BrianClone/linkedin_posts

1urn,text,type,firstName,lastName,numImpressions,numViews,numReactions,numComments,numShares,numVotes,numEngagementRate,hashtags,createdAt (TZ=America/Los_Angeles),link2urn:li:activity:7437904645592862720,"""𝘐 𝘥𝘰𝘯'𝘵 𝘬𝘯𝘰𝘸 𝘩𝘰𝘸 𝘢𝘯𝘺 𝘰𝘧 𝘵𝘩𝘪𝘴 𝘸𝘰𝘳𝘬𝘴""3 4Our lead data scientist had created a service to support multi-modal embedding and retrieval from a vector database and a host of other functionality we needed for classifying and analyzing videos.5 6This was a greenfield project with an aggressive timeline.7 8He was the first person to introduce me to Opus. He's what I consider an AI expert with deep mathematical expertise and a PhD.9 10Still, there was no way a human could possibly understand the amount of code that Claude produced.11 12Not because it was bad.13 14Just because there was so much of it.15 16At one point he mentioned 𝗶𝘁 𝘄𝗼𝘂𝗹𝗱 𝗽𝗿𝗼𝗯𝗮𝗯𝗹𝘆 𝘁𝗮𝗸𝗲 𝘄𝗲𝗲𝗸𝘀 𝘁𝗼 𝗳𝘂𝗹𝗹𝘆 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘁𝗵𝗲 𝗰𝗼𝗱𝗲 𝗮𝗻𝗱 𝗮𝗹𝗹 𝘁𝗵𝗲 𝗲𝗱𝗴𝗲 𝗰𝗮𝘀𝗲𝘀 𝗮𝗻𝗱 𝗱𝗲𝘃𝗲𝗹𝗼𝗽 𝗮 𝘀𝘁𝗿𝗼𝗻𝗴 𝘁𝗲𝘀𝘁 𝘀𝘂𝗶𝘁𝗲.17 18And that got me thinking.19 20𝙄𝙨 𝙝𝙞𝙜𝙝-𝙡𝙚𝙫𝙚𝙡 𝙪𝙣𝙙𝙚𝙧𝙨𝙩𝙖𝙣𝙙𝙞𝙣𝙜 𝙚𝙣𝙤𝙪𝙜𝙝?21 22In some ways, yes. Developers work on systems we didn’t build MOST of the time. Understanding the architecture and intent is usually enough to move things forward.23 24But eventually someone needs to know where the bodies are buried.25 26Code breaks. Edge cases appear. Systems evolve.27 28And when that happens, someone has to understand the service deeply enough to debug it.29 30Which makes me wonder:31 32𝘈𝘳𝘦 𝘸𝘦 𝘢𝘤𝘵𝘶𝘢𝘭𝘭𝘺 𝘣𝘦𝘤𝘰𝘮𝘪𝘯𝘨 𝘮𝘰𝘳𝘦 𝘱𝘳𝘰𝘥𝘶𝘤𝘵𝘪𝘷𝘦…33 34𝘖𝘳 𝘢𝘳𝘦 𝘸𝘦 𝘫𝘶𝘴𝘵 𝘬𝘪𝘤𝘬𝘪𝘯𝘨 𝘵𝘩𝘦 𝘤𝘢𝘯 𝘥𝘰𝘸𝘯 𝘵𝘩𝘦 𝘳𝘰𝘢𝘥?35 36Shipping faster now, while pushing the real understanding cost into the future.",TEXT,Brian,Jenney,641,0,9,2,1,0,0.0187207488299532,,2026-03-12 09:57:37,https://www.linkedin.com/feed/update/urn:li:activity:743790464559286272037urn:li:activity:7437535914496765952,"You don’t need to learn how to code anymore.38 39All you need to learn is:40 41• How to use Git 42• How deployment works 43• Lambdas vs EC2 44• AWS vs GCP vs Vercel 45• SQL vs NoSQL databases46• Foreign key relationships 47• Indexes and query performance 48• LEFT JOIN vs INNER JOIN 49• Caching50• Environment variables 51• Authentication flows 52• Debugging production issues 53• CI/CD pipelines54 55…and then you just tell the AI what code to write.56 57It’s just that simple.",TEXT,Brian,Jenney,12346,0,94,18,2,0,0.009233759922242022,,2026-03-11 09:32:24,https://www.linkedin.com/feed/update/urn:li:activity:743753591449676595258urn:li:activity:7437531279597735936,"The story is deeper than the headline.59 60At least one Stanford economist calculated a productivity increase of 2.7% compared to this time last year.61 62Others see modest gains.63 64The 6,000 CEOs who were surveyed have seen basically little to no impact in their own orgs.65 66Either way - this isn't the dystopian reality we were promised.67 68Don't worry - it's just 12 months away 😉",IMAGE,Brian,Jenney,1568,0,7,5,2,0,0.008928571428571428,,2026-03-11 09:13:59,https://www.linkedin.com/feed/update/urn:li:activity:743753127959773593669urn:li:activity:7437183039337803776,"Outside of being a software developer, I also run a small business.70 71Here's where I use AI and where I absolutely do NOT.72 73✅ 𝗨𝗽𝗱𝗮𝘁𝗶𝗻𝗴 𝗰𝘂𝗿𝗿𝗶𝗰𝘂𝗹𝘂𝗺 - tedious updates are now trivial. This has saved me hundreds of hours74✅ 𝗟𝗮𝗻𝗱𝗶𝗻𝗴 𝗽𝗮𝗴𝗲𝘀 - I do not have a design eye. A landing page is a great target for Claude to create on its own (like this one: https://parsity.io/ai-dev)75✅ 𝗖𝗼𝗻𝘁𝗲𝗻𝘁 - I built a clone of myself, with access to thousands of examples of my writing (including this post) that helps me brainstorm and create articles/posts/transcripts76 77H͟e͟r͟e͟'͟s͟ ͟w͟h͟e͟r͟e͟ ͟I͟ ͟a͟v͟o͟i͟d͟ ͟A͟I͟:͟78 79❌ 𝗦𝘁𝘂𝗱𝗲𝗻𝘁 𝗳𝗲𝗲𝗱𝗯𝗮𝗰𝗸 - yeah, I could create a tool to do this but it seems cold and the error rate is too high. Human mentorship is our secret sauce. I cannot imagine automating this.80❌ 𝗣𝗿𝗼𝘀𝗽𝗲𝗰𝘁𝗶𝗻𝗴 - first off, LinkedIn will ban your ass if they catch you automating your messages. I also straight up don't trust an agent to hold a reasonable convo with a potential customer. We've seen massive companies fail spectacularly with this. I'm not going to beat them.81❌ 𝗘𝗺𝗮𝗶𝗹𝘀 - I played with the idea of having an agent read through my inbox and reply to emails. Again - the potential for error is just too high. We aren't drowning in emails either. The juice just isn't worth the squeeze.82 83I'll continue to look for areas we can squeeze out productivity without sacrificing quality or our human touch.84 85If you're a business owner, I'm curious where you're finding the most value when it comes to AI.",TEXT,Brian,Jenney,4876,0,21,3,0,0,0.004922067268252666,,2026-03-10 10:10:12,https://www.linkedin.com/feed/update/urn:li:activity:743718303933780377686urn:li:activity:7436803318062149632,"250 applications and barely an interview.87 88Bryan Peterson experienced what I hear from a lot of developers on the job hunt nowadays.89 90He updated his strategy and has had 16 interviews from 58 applications.91 92🤯93 94I met Bryan in Reno, NV last year and I saw the proof.95 96We sat down to chat about technical interviews, surprising ways he's failed and succeeded in the job search and a few mental shifts that can save your sanity when it feels like you're sending applications out into the void.97 98https://lnkd.in/gmwc7ikU",ARTICLE,Brian,Jenney,2818,0,19,2,0,0,0.007452093683463449,,2026-03-09 09:01:20,https://www.linkedin.com/feed/update/urn:li:activity:743680331806214963299urn:li:activity:7436171946171498496,"Last night I was riding through San Francisco in a Waymo with Alex Lau (author of Keep Calm, Code On), passing giant AI billboards promising to automate us out of jobs.100 101Naturally the conversation turned to the future of software engineers.102 103Neither of us claims to know the answer. But one thing is clear: the job is changing faster than most people realize.104 105I’ve started noticing a few predictable levels of AI leverage developers move through. Most people are stuck at level one.106 107The interesting opportunities are in the levels beyond that.108 109Here’s what those look like.",ARTICLE,Brian,Jenney,2533,0,15,2,0,0,0.006711409395973154,,2026-03-07 14:12:29,https://www.linkedin.com/feed/update/urn:li:activity:7436171946171498496110urn:li:activity:7435392216933482496,"I was demoing an AI agent (𝘳𝘦𝘢𝘭𝘭𝘺 𝘢 𝘸𝘰𝘳𝘬𝘧𝘭𝘰𝘸 𝘣𝘶𝘵, 𝘵𝘰𝘮𝘢𝘵𝘰 𝘵𝘰𝘮𝘢𝘵𝘰) — that broke at the worst possible time.111 112Our CEO asked it to find influencers in our database that charged less than $1000 per post.113 114No results.115 116I had tried this scenario so many times before demo day.117 118What happened?!119 120When I inspected the logs, I saw our agent was sometimes parsing 1K as $1 or $10 instead of $1000 😅121 122So the agent searched the database and confidently concluded:123 124🤖 “𝘕𝘰 𝘪𝘯𝘧𝘭𝘶𝘦𝘯𝘤𝘦𝘳𝘴 𝘧𝘰𝘶𝘯𝘥.”125 126A tiny interpretation mistake that completely broke the outcome.127 128This is something I’ve noticed a lot building AI agents:129 130They fail in subtle, unexpected ways like unit conversions, formatting quirks and edge cases you don't think about.131 132In our case, the fix wasn’t fancy. 133 134We added a regex parser to extract the price and compare it against what the AI produced. LLM observability tools like Langsmith allowed us to track these differences and determine if tweaks to our system prompt improved accuracy.135 136Not sexy.137 138But it’s a good reminder that when building agents, a lot of the work is anticipating the weird ways they might be 𝘢𝘭𝘮𝘰𝘴𝘵 right.",TEXT,Brian,Jenney,937,0,14,9,1,0,0.025613660618996798,,2026-03-05 10:34:07,https://www.linkedin.com/feed/update/urn:li:activity:7435392216933482496139urn:li:activity:7435022923012128768,"𝗬𝗼𝘂 𝗱𝗼𝗻’𝘁 𝗻𝗲𝗲𝗱 𝗮𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁.140 141You need a workflow.142 143Most teams asking for an “agent” really want something like this:144 145𝚄𝚜𝚎𝚛 𝚛𝚎𝚚𝚞𝚎𝚜𝚝 → 𝚌𝚕𝚊𝚜𝚜𝚒𝚏𝚢 𝚒𝚗𝚝𝚎𝚗𝚝 → 𝚛𝚘𝚞𝚝𝚎 𝚝𝚘 𝚝𝚑𝚎 𝚛𝚒𝚐𝚑𝚝 𝚝𝚘𝚘𝚕 → 𝚛𝚎𝚝𝚞𝚛𝚗 𝚝𝚑𝚎 𝚛𝚎𝚜𝚞𝚕𝚝.146 147That’s a 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝗱 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄.148 149Anthropic knows this.150 151LangGraph knows this too.152 153“Agents” get all the publicity but workflows get shipped.154 155One of the simplest ways to build this is with the 𝗿𝗼𝘂𝘁𝗲𝗿 𝗽𝗮𝘁𝘁𝗲𝗿𝗻 — where an LLM looks at a request and decides which path your system should take (search, API call, database query, etc).156 157Simple.158 159Predictable.160 161𝙔𝙤𝙪 𝙘𝙖𝙣 𝙜𝙧𝙖𝙗 𝙖 𝙨𝙢𝙖𝙡𝙡 𝙥𝙧𝙤𝙟𝙚𝙘𝙩 𝙩𝙤 𝙛𝙤𝙧𝙘𝙚 𝙮𝙤𝙪𝙧 𝙪𝙣𝙙𝙚𝙧𝙨𝙩𝙖𝙣𝙙𝙞𝙣𝙜: https://lnkd.in/gz2Q9hSf",TEXT,Brian,Jenney,798,0,14,3,0,0,0.021303258145363407,,2026-03-04 10:06:40,https://www.linkedin.com/feed/update/urn:li:activity:7435022923012128768162urn:li:activity:7435012931534077952,"Everybody loves to hate LinkedIn.163 164“𝘐𝘵’𝘴 𝘤𝘳𝘪𝘯𝘨𝘦.”165 166“𝘐𝘵’𝘴 𝘢𝘭𝘭 𝘈𝘐 𝘴𝘭𝘰𝘱.”167 168“𝘐’𝘮 𝘯𝘰𝘵 𝘱𝘰𝘴𝘵𝘪𝘯𝘨.”169 170Cool. But if your goal is a software engineering job, you don’t get to opt out... yet.171 172LinkedIn isn’t for your friends or your family. It’s for recruiters. And recruiters don’t want to take a chance on you. 173 174So your job is simple: 𝗱𝗲-𝗿𝗶𝘀𝗸 𝘆𝗼𝘂𝗿𝘀𝗲𝗹𝗳 and leave evidence you can actually do the work.175 176A few of the lowest-hanging fixes I see constantly (especially with junior devs):177 178• 𝗦𝘁𝗼𝗽 𝗰𝗮𝗹𝗹𝗶𝗻𝗴 𝘆𝗼𝘂𝗿𝘀𝗲𝗹𝗳 “𝗷𝘂𝗻𝗶𝗼𝗿” or “aspiring.” You’re a developer.179• 𝗨𝘀𝗲 𝘀𝘁𝗿𝗼𝗻𝗴 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲: 𝘣𝘶𝘪𝘭𝘵, 𝘴𝘩𝘪𝘱𝘱𝘦𝘥, 𝘭𝘦𝘥, 𝘪𝘮𝘱𝘳𝘰𝘷𝘦𝘥 𝘟 𝘣𝘺 𝘠 — not vague “worked on a team.”180• 𝗟𝗲𝗮𝗱 𝘄𝗶𝘁𝗵 𝗿𝗲𝗹𝗲𝘃𝗮𝗻𝘁 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 (projects, freelance, volunteer) so you don’t look like a career pivot with zero proof.181 182I broke all of this down, plus a simple strategy for getting more inbound opportunities without living on LinkedIn in this latest episode of Develop Yourself:183   184https://lnkd.in/gvrnv9HC185 186𝘗𝘚 - 𝘺𝘰𝘶 𝘭𝘪𝘬𝘦 𝘵𝘩𝘦 𝘯𝘦𝘸 𝘷𝘪𝘥𝘦𝘰 𝘧𝘰𝘳𝘮𝘢𝘵?",ARTICLE,Brian,Jenney,4469,0,46,18,1,0,0.014544640859252628,,2026-03-04 09:26:58,https://www.linkedin.com/feed/update/urn:li:activity:7435012931534077952187urn:li:activity:7434651315332444160,"It's honestly kinda wild how much coding you still need to know if you want to Vibe Code an app and have no technical background.188 189I wrote this article for non-devs who are interested in launching an app but don't want to learn to code.",ARTICLE,Brian,Jenney,1734,0,16,1,1,0,0.010380622837370242,,2026-03-03 09:30:02,https://www.linkedin.com/feed/update/urn:li:activity:7434651315332444160190urn:li:activity:7434648156744740864,"I'm a software developer and I own a small business. I'm not replacing a single person with AI.191 192Parsity has had a thin staff of mentors and an assistant for years. Most of our processes were automated since before I took over and well before anyone knew what Claude was.193 194I don't see a world where I have an AI assistant reaching out to prospective students, writing our weekly newsletter, grading work or ""mentoring"" people.195 196I'm not trying to take some moral high ground here either.197 198Parsity depends on humans to instruct, mentor and create curriculum. Believe it or not, but humans typically want to interact with... other humans!199 200Shaving away people to save money while delivering an inferior experience makes 0 sense to me.",TEXT,Brian,Jenney,5025,0,88,18,1,0,0.021293532338308458,,2026-03-03 09:17:29,https://www.linkedin.com/feed/update/urn:li:activity:7434648156744740864201urn:li:activity:7434284544440147968,"While non-coders are busy arguing about whether or not AI replaces coders, consider learning these skills:  202 2031. RAG 2042. AI agent patterns 2053. LLM ops 2064. Testing and eval strategies for models  207 208There is a major opportunity a few layers beneath learning how to use Claude better.209 210https://lnkd.in/gsNyXZkN",ARTICLE,Brian,Jenney,1740,0,25,0,0,0,0.014367816091954023,,2026-03-02 09:12:37,https://www.linkedin.com/feed/update/urn:li:activity:7434284544440147968211urn:li:activity:7433233332756193280,"I was writing an article to help Vibe Coders ship something. 212 213It was immediately clear how much technical knowledge you still need to have outside coding to actually launch:214 215• IDE set up 216• Cloud deployment 217• e2e tests (if you want to avoid regressions or bugs) 218• Github 219• Debugging220 221Anything I missed?",TEXT,Brian,Jenney,1240,0,19,6,0,0,0.020161290322580645,,2026-02-27 11:35:29,https://www.linkedin.com/feed/update/urn:li:activity:7433233332756193280222urn:li:activity:7433198293301723136,"This might lose me some business, but it needs to be said:223 224You can't stack AI skills on top of shaky software skills.225 226Before you join a program like https://parsity.io/ai-dev and dive into agents, RAG, or fine-tuning, you need real coding basics.227 228Do this first:229 230• 𝗚𝗲𝘁 𝗰𝗼𝗺𝗳𝗼𝗿𝘁𝗮𝗯𝗹𝗲 𝘄𝗶𝘁𝗵 𝗳𝘂𝗻𝗱𝗮𝗺𝗲𝗻𝘁𝗮𝗹𝘀: APIs, common design patterns and building and deploying full stack apps231• 𝗨𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘁𝗵𝗲 𝗲𝘀𝘀𝗲𝗻𝘁𝗶𝗮𝗹 𝗔𝗜 𝗽𝗿𝗶𝗺𝗶𝘁𝗶𝘃𝗲𝘀: Tokens, embeddings, context windows. If those concepts feel fuzzy, everything advanced will feel like magic.232 233Here’s the truth:234 235AI engineering isn’t a shortcut. It’s software development with more moving parts and a rule book that is still being written.236 237If you don’t have a solid foundation you'll be building on a house of cards.",TEXT,Brian,Jenney,2172,0,37,3,1,0,0.018876611418047883,,2026-02-27 09:16:15,https://www.linkedin.com/feed/update/urn:li:activity:7433198293301723136238urn:li:activity:7432813999882293248,"Before I was a software developer, I was a [redacted for legal reasons].239 240But before that, I honestly wanted to be a writer.241 242Kinda still do.243 244It's one of the many reasons I write so much.245 246If you've been connected to me for a while, you also know I don't do sponsorships. Not because I have some moral stance on sponsorships, it's just that I have my own business that I shamelessly promote.247 248Educative is a product I actually like, have paid for and used heavily when I first started studying for interviews. 249 250The thing that stands out about their platform is that it's text-first.251 252I prefer text to video 9 times out of ten. Easier to digest == quicker to understand.253 254They just published a short article I wrote about interviews in the AI era. I predict that your next interview is more likely to look like one I did recently than whatever you've seen in the past...",SHARE,Brian,Jenney,3356,0,21,4,0,0,0.007449344457687724,,2026-02-26 07:49:12,https://www.linkedin.com/feed/update/urn:li:activity:7432813999882293248255urn:li:activity:7432571818332368896,"There’s a massive digital divide that is only getting wider - especially in cities like Oakland.256 257When I bumped into Adamaka Ajaelo a few weeks ago in Oakland, CA I learned more about how she’s helping bridge that gap with Self-eSTEM.258 259Adamaka has a highly technical background and teaches practical skills to youth and adults.260 261Give her a follow if you fancy.",IMAGE,Brian,Jenney,1166,0,24,2,0,0,0.022298456260720412,,2026-02-25 15:46:51,https://www.linkedin.com/feed/update/urn:li:activity:7432571818332368896262urn:li:activity:7432271273872904192,"""𝘏𝘦𝘭𝘱! 𝘞𝘦 𝘨𝘢𝘷𝘦 𝘢𝘯 𝘈𝘐 𝘢𝘨𝘦𝘯𝘵 𝘢𝘤𝘤𝘦𝘴𝘴 𝘵𝘰 𝘰𝘶𝘳 𝘦𝘯𝘵𝘪𝘳𝘦 𝘥𝘢𝘵𝘢𝘣𝘢𝘴𝘦 𝘢𝘯𝘥 𝘴𝘰𝘮𝘦𝘰𝘯𝘦 𝘰𝘯 𝘰𝘶𝘳 𝘵𝘦𝘢𝘮 𝘫𝘶𝘴𝘵 𝘥𝘦𝘭𝘦𝘵𝘦𝘥 𝘢𝘭𝘭 𝘰𝘶𝘳 𝘶𝘴𝘦𝘳𝘴 𝘣𝘺 𝘢𝘤𝘤𝘪𝘥𝘦𝘯𝘵!""263 264We’re going to save this fictional (or not-so-fictional) company in a live event and learn how using human-in-the-loop could have prevented this mess.265 266If you’re curious about how to build AI agents in a practical way, you should join.267 268If you don’t see a problem with giving AI tools total access to your data, then maybe this isn’t the event for you 😅",EVENT,Brian,Jenney,902,0,17,3,5,0,0.02771618625277162,,2026-02-24 19:52:36,https://www.linkedin.com/feed/update/urn:li:activity:7432271273872904192269urn:li:activity:7432116119768834048,"Did you know you can run Python in your NextJS app?270 271I was so jealous when Ali Cagatay showed me his capstone project from Parsity's AI Accelerator program which included downloading the transcripts from popular YouTuber's so a user could ""talk"" with them through a chat interface.272 273The library he was using is only available in Python.274 275😢276 277Turns out, NextJS now supports Python runtime. I have API routes in both Typescript and Python. I'm sure this irks many of you...278 279A cron job in Python scrapes my most recent videos, chunks up the transcript and stores it in a vector database for easy searching and retrieval to help generate more content that sounds like me.280 281I can't wait to see what people build in the next cohort that I can steal 😅",IMAGE,Brian,Jenney,926,0,13,2,0,0,0.016198704103671708,,2026-02-24 09:36:04,https://www.linkedin.com/feed/update/urn:li:activity:7432116119768834048282urn:li:activity:7431750092530860032,"Cheating in interviews is rampant.283 284It's gotten so bad that there's a cottage industry dedicated to preventing cheaters and detecting AI usage.285 286Fardeen Khimani 😇 has a different approach: 287 288- no Leetcode289- unlimited AI usage290- no marathon interview rounds291- faster hiring based on challenges that reflect the actual work292 293He has some pretty spicy takes on cheating with AI, the death of the whiteboard interview and why most companies are behind the times with their process.294 295https://lnkd.in/gmZXcGdy",ARTICLE,Brian,Jenney,8083,0,48,7,5,0,0.007422986514907831,,2026-02-23 09:21:37,https://www.linkedin.com/feed/update/urn:li:activity:7431750092530860032296urn:li:activity:7431744752867254273,"If you’re a web developer, you’re not far from AI engineering.297 298But the shift isn’t (just) about using Claude better.299 300Here’s a practical roadmap.301 302           303𝟭. 𝗔𝗣𝗜-𝗙𝗶𝗿𝘀𝘁 𝗔𝗜 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁304 305Start calling models via API.306 307Learn how to structure prompts programmatically, manage context windows, constrain outputs, and control behavior through parameters.308 309The real skill isn’t “prompting.”310 311It’s designing the inputs so the model has the right information at the right time.312 313𝗛𝗼𝘄 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁:314 315Call LLMs directly via API. Experiment with system messages, structured prompts, output schemas, temperature, and response formats inside real applications.316 317 318𝟮. 𝗟𝗟𝗠 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀 (𝗗𝗲𝘃𝗲𝗹𝗼𝗽 𝗜𝗻𝘁𝘂𝗶𝘁𝗶𝗼𝗻)319 320You don’t need deep math, but you do need a mental model.321 322Understand, at a high level, how transformers, embeddings, attention, and feed-forward layers work. This sharpens your intuition about what models can and can’t do and helps you avoid hype-cycle nonsense.323 324𝗛𝗼𝘄 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁:325 326Build a tiny language model in Python with PyTorch. Or watch 3Blue1Brown’s transformer series and the first few episodes of his linear algebra series.327 328 329𝟯. 𝗔𝗴𝗲𝗻𝘁𝘀, 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 & 𝗧𝗼𝗼𝗹 𝗖𝗮𝗹𝗹𝗶𝗻𝗴330 331Learn the difference between deterministic workflows and autonomous agents. Study patterns like ReAct, human-in-the-loop systems, orchestrator patterns, and tool calling — and when each makes sense.332 333More autonomy isn’t always better. Reliability is a design decision.334 335https://lnkd.in/gz2Q9hSf336 337𝗛𝗼𝘄 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁:338 339Build a small agent that uses tools. Then refactor it into a structured workflow and compare reliability and control.340 341  342𝟰. 𝗥𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹-𝗔𝘂𝗴𝗺𝗲𝗻𝘁𝗲𝗱 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 (𝗥𝗔𝗚)343 344Most real AI products leverage retrieval systems.345 346Learn how to chunk documents, generate embeddings, and design retrieval pipelines for internal knowledge, semantic search, and context-aware applications.347 348𝗛𝗼𝘄 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁:349 350Build a simple RAG pipeline on your own documentation or notes.351 352https://lnkd.in/d6Pzfcg6353 354 355𝟱. 𝗟𝗟𝗠 𝗢𝗽𝘀 (𝗘𝘃𝗮𝗹𝘀 & 𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆)356 357AI systems don’t fail loudly — they drift. (𝘙𝘦𝘮𝘦𝘮𝘣𝘦𝘳 𝘵𝘩𝘦 𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘤𝘦 𝘣𝘦𝘵𝘸𝘦𝘦𝘯 𝘎𝘗𝘛-4 𝘢𝘯𝘥 5?)358 359You need testing strategies, evaluation datasets, and observability to understand performance over time.360 361𝗛𝗼𝘄 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁:362 363Create a test suite for the agents you built earlier. Add tools like Helicone or LangSmith to introduce observability and structured evaluation.364 365 366𝗧𝗵𝗶𝘀 𝗶𝘀 𝘁𝗵𝗲 𝘀𝗵𝗶𝗳𝘁 𝗳𝗿𝗼𝗺 𝘂𝘀𝗲𝗿 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱𝗲𝗿.",TEXT,Brian,Jenney,1558,0,27,2,2,0,0.01989730423620026,,2026-02-23 09:00:24,https://www.linkedin.com/feed/update/urn:li:activity:7431744752867254273367urn:li:activity:7431022202671677440,"What software developers can learn from an insecure fictional character from the 19th century.368 369And why it’s especially relevant in the AI hype cycle.",ARTICLE,Brian,Jenney,1761,0,15,2,0,0,0.009653605905735378,,2026-02-21 09:09:14,https://www.linkedin.com/feed/update/urn:li:activity:7431022202671677440370urn:li:activity:7430302390148485120,"In 2024, I remember sitting in the office, nervous what the CEO might think of me using ChatGPT to write some code.371 372Maybe he'd re-consider if I was the right fit? That I wasn't a ""real"" programmer.373 374In less than 2 years the tables have completely flipped.375 376If you're NOT using AI coding tools nowadays, it's a bright red flag for most(?) companies.377 378Your next interview is likely to reflect this new reality.379 380I break down a recent coding interview that allowed AI tools... and how that didn't make it any easier: https://lnkd.in/gnDt3i7q",ARTICLE,Brian,Jenney,5636,0,21,1,1,0,0.0040809084457061745,,2026-02-19 09:28:58,https://www.linkedin.com/feed/update/urn:li:activity:7430302390148485120381urn:li:activity:7429932716353208320,"When I built the curriculum and homework for Parsity's AI Cohort, I didn't do it on vibes.382 383I took a mix of what I learned working at 2 different AI startups as a full stack developer and the skills I keep seeing in the market.384 385It's the reason I decided against MCP, heavy math and a diving deeper into fine-tuning.386 387I stand by my decision to keep the codebase in Typescript, for now. The concepts translate easily across stacks and the amount of material out there for TS devs when it comes to AI is shockingly low.388 389𝗧𝗵𝗲 𝗴𝗼𝗮𝗹 𝗶𝘀 𝗽𝗿𝗲𝘁𝘁𝘆 𝘀𝗶𝗺𝗽𝗹𝗲: make you the goto person in your organization when it comes to AI integration OR help you transition into an full stack AI engineer role.390 391You can apply here: https://parsity.io/ai-dev",IMAGE,Brian,Jenney,1123,0,16,0,0,0,0.014247551202137132,,2026-02-18 09:00:01,https://www.linkedin.com/feed/update/urn:li:activity:7429932716353208320392urn:li:activity:7429574081773166592,"It’s not just you.393 394Expectations around software delivery have gotten unhinged.395 396For years we said, “𝘈𝘯𝘺𝘰𝘯𝘦 𝘤𝘢𝘯 𝘭𝘦𝘢𝘳𝘯 𝘵𝘰 𝘤𝘰𝘥𝘦.”397 398Now anyone can generate code.399 400For better or for worse.401 402𝗕𝗨𝗧, generating code isn’t the same as building software.403 404A feature that works on localhost isn’t production-ready.405 406When non-technical leaders can ship something that ""works,"" it creates the illusion that software development was always just typing.407           408You might be dealing with:409 410• Eroded trust (""𝘉𝘶𝘵 𝘪𝘵 𝘸𝘰𝘳𝘬𝘦𝘥 𝘸𝘩𝘦𝘯 𝘐 𝘳𝘢𝘯 𝘪𝘵 𝘭𝘰𝘤𝘢𝘭𝘭𝘺."")411• Inflated expectations (""𝘐 𝘣𝘶𝘪𝘭𝘵 𝘵𝘩𝘪𝘴 𝘪𝘯 𝘢 𝘸𝘦𝘦𝘬𝘦𝘯𝘥 𝘧𝘳𝘰𝘮 𝘴𝘤𝘳𝘢𝘵𝘤𝘩!"")412• The belief that typing speed was the bottleneck all along.413 414I think there's also a lot of opportunity in this odd stage we're in.415 416If you take the time to articulate why that ""simple service"" requires tests, additional infra and a rollback strategy - then you can position yourself as a leader on the team.417 418Obviously, your mileage will vary.419 420 421𝙏𝙝𝙞𝙨 𝙬𝙚𝙚𝙠𝙚𝙣𝙙, 𝙄'𝙡𝙡 𝙗𝙚 𝙬𝙧𝙞𝙩𝙞𝙣𝙜 𝙖𝙗𝙤𝙪𝙩 𝙝𝙤𝙬 𝙮𝙤𝙪 𝙘𝙖𝙣 𝙜𝙚𝙩 𝙮𝙤𝙪𝙧 𝘾𝙮𝙧𝙖𝙣𝙤 𝘿𝙚 𝘽𝙚𝙧𝙜𝙚𝙧𝙖𝙘 𝙢𝙤𝙢𝙚𝙣𝙩 𝙖𝙨 𝙖 𝙙𝙚𝙫𝙚𝙡𝙤𝙥𝙚𝙧 𝙞𝙣 𝙩𝙝𝙚 𝘼𝙄 𝙚𝙧𝙖 𝙖𝙣𝙙 𝙬𝙝𝙮 𝙩𝙝𝙞𝙨 𝙤𝙥𝙥𝙤𝙧𝙩𝙪𝙣𝙞𝙩𝙮 𝙬𝙤𝙣'𝙩 𝙡𝙖𝙨𝙩. https://lnkd.in/gmEZY7K8",TEXT,Brian,Jenney,1763,0,21,5,0,0,0.01474758933635848,,2026-02-17 09:14:55,https://www.linkedin.com/feed/update/urn:li:activity:7429574081773166592422urn:li:activity:7429199815932071937,"Hiring is obviously broken.  423 424To make it worse, you're using the same tactics that haven't worked since 2019.  425 426𝗦𝗼𝗺𝗲 𝗵𝗮𝗿𝘀𝗵 𝘁𝗿𝘂𝘁𝗵𝘀:  427- Your resume is increasingly less important 428- Luck is a factor no one wants to admit 429- Mass applying is like playing the lotto 430- LinkedIn is a social media site pretending to be a job board  431 432𝗦𝗼𝗺𝗲 𝘀𝘁𝘂𝗳𝗳 𝘆𝗼𝘂 𝘀𝗵𝗼𝘂𝗹𝗱 𝘀𝘁𝗮𝗿𝘁 𝗱𝗼𝗶𝗻𝗴:  433- Get 500 connections on LinkedIn to be more discoverable 434- Remove any mention of junior/aspiring/student from your profile 435- Don't apply for only junior roles - let the market decide 436- Do a BFS of your network to find hidden jobs437- Build in public - this is uncomfortable and effective (𝘺𝘰𝘶'𝘭𝘭 𝘢𝘭𝘴𝘰 𝘣𝘦 𝘮𝘰𝘳𝘦 𝘷𝘪𝘴𝘪𝘣𝘭𝘦 𝘪𝘧 𝘳𝘦𝘤𝘳𝘶𝘪𝘵𝘦𝘳𝘴 𝘢𝘳𝘦 𝘶𝘴𝘪𝘯𝘨 𝘧𝘪𝘭𝘵𝘦𝘳𝘴 𝘧𝘰𝘳 𝘢𝘤𝘵𝘪𝘷𝘪𝘵𝘺 😉)438 439I'm not saying this is easy or fair.440 441If you've already gone to a coding bootcamp and are looking to land your first role - I'd personally work with David Roberts 442 443If you're at the beginning of your journey and realize it's going to take more than coding skills to break in, join Parsity https://parsity.io",TEXT,Brian,Jenney,11385,0,59,14,1,0,0.006499780412823891,,2026-02-16 08:27:43,https://www.linkedin.com/feed/update/urn:li:activity:7429199815932071937444urn:li:activity:7428119230572728320,"Apocalypse marketing is the most effective AI growth strategy right now.445 446Step 1: CEO predicts white-collar job collapse in X months.447Step 2: Internet panics.448Step 3: Engagement explodes.449 450Yale Budget Lab’s analysis of the labor market so far shows no clear sign that AI exposure is causing large employment shifts… yet.  451 452So either:453- executives know something the data doesn’t…454- or fear is great marketing.455 456Sometimes it feels like we’re using two different AIs.",TEXT,Brian,Jenney,931,0,12,8,0,0,0.021482277121374866,,2026-02-13 08:53:52,https://www.linkedin.com/feed/update/urn:li:activity:7428119230572728320457urn:li:activity:7427791445362733057,"Imagine being a junior developer right now.458 459You have CEOs of major tech companies telling you that no one will write code in the next few months.460 461Every other tech influencer is showing off their bespoke workflow with multiple agents.462 463Is it any wonder that more early career developers are over-relying on AI to do their work?464 465Turns out, there is a very real cost to doing this, based on a recent study from Anthropic.466 467If you don't want to turn into a prompt-squirrel, I outline some things you might want to consider...",ARTICLE,Brian,Jenney,11084,0,49,10,1,0,0.0054132082280765065,,2026-02-12 11:11:22,https://www.linkedin.com/feed/update/urn:li:activity:7427791445362733057468urn:li:activity:7427768581465186305,"𝘛𝘩𝘦 𝘤𝘰𝘥𝘪𝘯𝘨 𝘭𝘢𝘯𝘨𝘶𝘢𝘨𝘦 𝘰𝘧 𝘵𝘩𝘦 𝘧𝘶𝘵𝘶𝘳𝘦 𝘸𝘪𝘭𝘭 𝘣𝘦 𝘌𝘯𝘨𝘭𝘪𝘴𝘩!469𝘈𝘐 𝘸𝘪𝘭𝘭 𝘳𝘦𝘱𝘭𝘢𝘤𝘦 92.8% 𝘰𝘧 𝘥𝘦𝘷𝘦𝘭𝘰𝘱𝘦𝘳𝘴 𝘣𝘺 [𝘯𝘦𝘹𝘵 𝘦𝘢𝘳𝘯𝘪𝘯𝘨𝘴 𝘤𝘢𝘭𝘭]!470𝘠𝘰𝘶 𝘸𝘰𝘯'𝘵 𝘣𝘦 𝘳𝘦𝘱𝘭𝘢𝘤𝘦𝘥 𝘣𝘺 𝘈𝘐, 𝘺𝘰𝘶'𝘭𝘭 𝘣𝘦 𝘳𝘦𝘱𝘭𝘢𝘤𝘦𝘥 𝘣𝘺 𝘴𝘰𝘮𝘦𝘰𝘯𝘦 𝘶𝘴𝘪𝘯𝘨 𝘈𝘐!471 472Sufficiently rage-baited?473 474I don't have a crystal ball and I don't know what the future of AI means for code authors.475 476I do know this:477 478In the last 12 months I went from knowing nothing about AI/LLMs to using vector databases to create RAG applications and building data pipelines to ingest tons of data to train AI models.479 480This has been a ton of fun and seems to have made me a hotter commodity on the job market than I have been in quite some time.481 482Because I like you - I've created a small project to teach you the basics of RAG, a popular and useful way to retrieve relevant info for LLMs.483 484Grab it here: https://lnkd.in/d6Pzfcg6",TEXT,Brian,Jenney,3380,0,37,2,0,0,0.011538461538461539,,2026-02-12 09:40:31,https://www.linkedin.com/feed/update/urn:li:activity:7427768581465186305485urn:li:activity:7427044451094585345,"This guy AIs. 486 487I've met plenty of ""AI experts"" who know a lot about theory but don't know how to build or think through practical problems that map to business use cases.488 489Tad didn't ask me to write this but with so much misinformation and snake oil being peddled, we need more Tad's out there.",SHARE,Brian,Jenney,602,0,10,2,1,0,0.02159468438538206,,2026-02-10 09:43:04,https://www.linkedin.com/feed/update/urn:li:activity:7427044451094585345490urn:li:activity:7426669081438720000,"Most AI projects start off with:491 - “Which model should we use?”492- “LangGraph vs CrewAI vs Vercel AI SDK?”493- “Python or Typescript?”494 495When they should start off with:496- “Where’s the data?”",ARTICLE,Brian,Jenney,1280,0,11,4,0,0,0.01171875,,2026-02-09 08:51:29,https://www.linkedin.com/feed/update/urn:li:activity:7426669081438720000497urn:li:activity:7426351032420573184,"❌ 10 prompts to 10x your workflow!498❌ Launch your first agent with ZERO code (or planning)!499❌ How use OpenClaw to drain your bank account!500 501Sorry AI bros - this post isn't meant for you.502 503Just wrapped up 4 weeks of building a complex RAG application with a small group of developers.504 505The results were well beyond what I expected.506 507They made projects and products which have some serious value.508 509Before we launch the next cohort, I want to do a live session for builders.510 511For Typescript devs who want to party with AI - I think you'll find this valuable.",POLL,Brian,Jenney,2083,0,10,1,0,30,0.019683149303888623,,2026-02-08 11:47:41,https://www.linkedin.com/feed/update/urn:li:activity:7426351032420573184512urn:li:activity:7425565696803287040,"We wrapped our first AI Cohort with a special guest, John Crickett , who showed us how to build:513 514• an agent that writes code515• and another one that impersonates a certain over-posting, mildly cringey LinkedIn influencer (me)516 517In the middle of the workshop, John said something that stuck with us:518 519“𝘼𝙄 𝙙𝙤𝙚𝙨𝙣’𝙩 𝙧𝙚𝙢𝙤𝙫𝙚 𝙩𝙝𝙚 𝙣𝙚𝙚𝙙 𝙛𝙤𝙧 𝙙𝙚𝙫𝙚𝙡𝙤𝙥𝙚𝙧𝙨 — 𝙞𝙩 𝙨𝙝𝙞𝙛𝙩𝙨 𝙩𝙝𝙚 𝙟𝙤𝙗 𝙩𝙤𝙬𝙖𝙧𝙙 𝙩𝙝𝙞𝙣𝙠𝙞𝙣𝙜, 𝙨𝙘𝙤𝙥𝙞𝙣𝙜, 𝙖𝙣𝙙 𝙙𝙚𝙘𝙞𝙨𝙞𝙤𝙣-𝙢𝙖𝙠𝙞𝙣𝙜.”520 521Then he followed it with this:522 523“𝘼 𝙡𝙤𝙩 𝙤𝙛 𝙥𝙚𝙤𝙥𝙡𝙚 𝙖𝙧𝙚 𝙣𝙤𝙩 𝙩𝙝𝙞𝙣𝙠𝙞𝙣𝙜 — 𝙩𝙝𝙚𝙮 𝙟𝙪𝙨𝙩 𝙖𝙘𝙘𝙚𝙥𝙩 𝙚𝙫𝙚𝙧𝙮𝙩𝙝𝙞𝙣𝙜 𝙩𝙝𝙚 𝘼𝙄 𝙨𝙪𝙜𝙜𝙚𝙨𝙩𝙨. 𝙏𝙝𝙚 𝙥𝙧𝙤𝙟𝙚𝙘𝙩 𝙨𝙘𝙤𝙥𝙚 𝙗𝙖𝙡𝙡𝙤𝙤𝙣𝙨. 𝙏𝙝𝙖𝙩’𝙨 𝙣𝙤𝙩 𝙨𝙤𝙛𝙩𝙬𝙖𝙧𝙚 𝙚𝙣𝙜𝙞𝙣𝙚𝙚𝙧𝙞𝙣𝙜.”524 525TBH, the hard part about building AI agents isn’t prompts.526 527In this cohort, most developers spent far more time:528 529– gathering and cleaning data530– deciding what the system should actually do531– cutting scope to deliver a useful product in a short time frame532 533The real skill wasn’t “using AI.”534 535It was engineering judgment.536 537And in just a few weeks, this group built some seriously impressive stuff.538 539We’re running it again.540 541 542 543𝗡𝗲𝘅𝘁 𝗔𝗜 𝗗𝗲𝘃 𝗖𝗼𝗵𝗼𝗿𝘁 𝘀𝘁𝗮𝗿𝘁𝘀 𝗶𝗻 𝗔𝗽𝗿𝗶𝗹.544https://parsity.io/ai-dev",IMAGE,Brian,Jenney,2260,0,39,15,0,0,0.023893805309734513,,2026-02-06 07:47:02,https://www.linkedin.com/feed/update/urn:li:activity:7425565696803287040545urn:li:activity:7424865775259389952,"Last year I took a week off work and traveled to beautiful (kinda) Reno, NV to create a new AI curriculum for Parsity.546 547That week I got a chance to speak to Alex Hormozi on his show, met a connection from LinkedIn in real life and gambled away my life savings! 548 549Fun times.550 551At the time, I was working at my 2nd AI start-up and learning things like RAG, agents, workflows, tool-calling and LLM-ops.552 553There's surprisingly little information out there on these topics. Even less for Typescript developers!554 555I built a program and material that I wish I had when I started out: light on theory, heavy on practical application.556 557Last week, our first cohort met for the last official session and nearly a dozen developers are now building some truly amazing products.558 559This has been some of the most fun I've had teaching in a long time. Looking forward to cohort #2!",SHARE,Brian,Jenney,1309,0,16,1,0,0,0.012987012987012988,#2,2026-02-04 09:25:48,https://www.linkedin.com/feed/update/urn:li:activity:7424865775259389952560urn:li:activity:7424503662942793728,"Side Effects of Vibe Coding:561 562• Sensitive API keys were embedded directly in client-side JavaScript bundles, making them publicly accessible.563• Row Level Security (RLS) was disabled, allowing unauthenticated users full administrative read and write access to the database.564• Approximately 1.5 million agent authentication tokens, 35,000 user email addresses, and over 4,𝟬𝟬𝟬 𝗽𝗿𝗶𝘃𝗮𝘁𝗲 𝗺𝗲𝘀𝘀𝗮𝗴𝗲𝘀 𝘄𝗲𝗿𝗲 𝗲𝘅𝗽𝗼𝘀𝗲𝗱.565• Reported participation metrics were misleading — 𝟭.𝟱 𝗺𝗶𝗹𝗹𝗶𝗼𝗻 “𝗮𝗴𝗲𝗻𝘁𝘀” 𝘄𝗲𝗿𝗲 𝗼𝗽𝗲𝗿𝗮𝘁𝗲𝗱 𝗯𝘆 𝗷𝘂𝘀𝘁 𝟭𝟳,𝟬𝟬𝟬 𝗵𝘂𝗺𝗮𝗻 𝘂𝘀𝗲𝗿𝘀 (an 88:1 ratio).566• Unauthenticated users could modify any live post on the platform, enabling widespread defacement and manipulation of content.567• 𝗣𝗿𝗶𝘃𝗮𝘁𝗲 𝗱𝗶𝗿𝗲𝗰𝘁 𝗺𝗲𝘀𝘀𝗮𝗴𝗲𝘀 𝗰𝗼𝗻𝘁𝗮𝗶𝗻𝗲𝗱 𝘁𝗵𝗶𝗿𝗱-𝗽𝗮𝗿𝘁𝘆 𝗢𝗽𝗲𝗻𝗔𝗜 𝗔𝗣𝗜 𝗸𝗲𝘆𝘀 stored in plaintext, putting users’ external accounts at risk.568 569Maybe Moltbook is the warning we need.570 571""𝘚𝘩𝘪𝘱 𝘧𝘢𝘴𝘵 𝘢𝘧 𝘶𝘴𝘪𝘯𝘨 𝘈𝘐 𝘵𝘰𝘰𝘭𝘴!""572 573What could go wrong? 574 575A lot apparently.",IMAGE,Brian,Jenney,3183,0,27,7,1,0,0.010995915802701853,,2026-02-03 09:26:53,https://www.linkedin.com/feed/update/urn:li:activity:7424503662942793728576urn:li:activity:7424134545815560192,"At Parsity, we’ve been against AI tool use while students are learning to code.577 578This was not an easy decision and I'd be lying if I said I was 100% sure that this was the right move.579 580Emerging research (and what we see in practice) supports our position:581 582• 𝗖𝗼𝗻𝗰𝗲𝗽𝘁𝘂𝗮𝗹 (𝗺𝗶𝘀)𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴: Students finish tasks without building real mental models583• 𝗗𝗲𝗯𝘂𝗴𝗴𝗶𝗻𝗴 𝗶𝗻𝗮𝗯𝗶𝗹𝗶𝘁𝘆: Fewer mistakes == fewer chances to learn how systems actually break584• 𝗟𝗮𝗰𝗸 𝗼𝗳 𝗰𝗼𝗱𝗲 𝗰𝗼𝗺𝗽𝗿𝗲𝗵𝗲𝗻𝘀𝗶𝗼𝗻: Reading AI output != knowing how to write or fix it yourself585• 𝗦𝗸𝗶𝗹𝗹 𝗿𝗲𝘀𝗶𝘀𝘁𝗮𝗻𝗰𝗲: Performance drops when the AI “exoskeleton” is removed586 587Yes, AI can boost short-term productivity.588But learning to code was never about speed. It’s about building the brain.589 590That’s why our students learn fundamentals first.591 592AI comes later. Once they can evaluate, debug, and supervise it.593 594We’re not anti-AI.595We’re pro-skill formation.",IMAGE,Brian,Jenney,1330,0,20,7,0,0,0.02030075187969925,,2026-02-02 09:00:09,https://www.linkedin.com/feed/update/urn:li:activity:7424134545815560192596urn:li:activity:7423838044660678656,"While my YouTube channel has gained a lot of traction, my podcast is steadily declining in listeners.597 598On the podcast, I feel like I've formed a deeper connection with listeners and can be a bit more nuanced in my advice.599 600I'm making some changes to hopefully increase listeners on Spotify by going all in on video episodes.601 602Tomorrow, the first video podcast drops and I'd love to hear from you if you're a listener of the show.603 604What are topics you really want to know about? What are you struggling with?605 606DM me or leave a comment and let me know if you want a shout out if I use your topic/question. 607 608THANK YOU FOR YOUR ATTENTION TO THIS MATTER 🫡",IMAGE,Brian,Jenney,1780,0,16,6,0,0,0.012359550561797753,,2026-02-01 13:21:58,https://www.linkedin.com/feed/update/urn:li:activity:7423838044660678656609urn:li:activity:7423413415185969152,"The same company telling you ""𝘞𝘦 𝘮𝘪𝘨𝘩𝘵 𝘣𝘦 6–12 𝘮𝘰𝘯𝘵𝘩𝘴 𝘢𝘸𝘢𝘺 𝘧𝘳𝘰𝘮 𝘮𝘰𝘥𝘦𝘭𝘴 𝘥𝘰𝘪𝘯𝘨 𝘮𝘰𝘴𝘵, 𝘮𝘢𝘺𝘣𝘦 𝘢𝘭𝘭, 𝘰𝘧 𝘸𝘩𝘢𝘵 𝘴𝘰𝘧𝘵𝘸𝘢𝘳𝘦 𝘦𝘯𝘨𝘪𝘯𝘦𝘦𝘳𝘴 𝘥𝘰 𝘦𝘯𝘥-𝘵𝘰-𝘦𝘯𝘥"" is also rapidly expanding.610 611I wonder what roles they will be hiring for? 🤔",IMAGE,Brian,Jenney,34873,0,79,18,2,0,0.0028388724801422303,,2026-01-31 09:14:38,https://www.linkedin.com/feed/update/urn:li:activity:7423413415185969152612urn:li:activity:7423056734421798913,"I helped design the interview process for an AI start up where I used to work.613 614The job listing was for an ""AI Engineer"".615 616What that meant, practically, was:617 6181. 𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 𝘄𝗶𝘁𝗵 𝗥𝗔𝗚: chunking && re-ranking and opinions on retrieval strategies6192. 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝗻𝗱 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀: tool-calling vs orchestration, evals and testing6203. 𝗔𝗴𝗲𝗻𝘁 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸𝘀: LangGraph, Vercel's AI SDK, etc. and when it makes sense to use them... or not6214. 𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆: how do you control token costs and see where agents fail or drift from their expected behavior?622 623These skills really aren't too difficult for most experienced devs to learn.624 625Yet, it's incredibly difficult to find people who already have practical experience, made the costly mistakes and know what to avoid.626 627If you're a developer looking to build these skills for fun or to launch a product or maybe you're an engineering manager who is looking for a more technical understanding of RAG and agents, 𝗜 𝗰𝗿𝗲𝗮𝘁𝗲𝗱 𝗮 𝗵𝗮𝗻𝗱𝘀-𝗼𝗻 𝗽𝗿𝗼𝗷𝗲𝗰𝘁 𝗳𝗼𝗿 𝘆𝗼𝘂 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗯𝘆 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴: https://lnkd.in/d6Pzfcg6",TEXT,Brian,Jenney,2674,0,19,0,1,0,0.0074794315632011965,,2026-01-30 09:37:19,https://www.linkedin.com/feed/update/urn:li:activity:7423056734421798913628urn:li:activity:7423054982716891136,"Code is just 1 tool in a developer’s box.629 630There are some equally important ones you’ll want to pick up along the way:631 632- project management633- public speaking 634- system design635 636Not only will these nice-to-haves become must-haves at some point in your career but they are more difficult to offload to an AI assistant.",TEXT,Brian,Jenney,509,0,5,0,1,0,0.011787819253438114,,2026-01-30 09:30:21,https://www.linkedin.com/feed/update/urn:li:activity:7423054982716891136637urn:li:activity:7422688743532433409,"Tech Layoffs Translated:638 639“𝘞𝘦’𝘳𝘦 𝘧𝘭𝘢𝘵𝘵𝘦𝘯𝘪𝘯𝘨 𝘵𝘩𝘦 𝘰𝘳𝘨.” 640𝗧𝗿𝗮𝗻𝘀𝗹𝗮𝘁𝗶𝗼𝗻: We overhired, and now we’re cutting back.  641 642“𝘞𝘦’𝘳𝘦 𝘣𝘦𝘤𝘰𝘮𝘪𝘯𝘨 𝘮𝘰𝘳𝘦 𝘦𝘧𝘧𝘪𝘤𝘪𝘦𝘯𝘵 𝘸𝘪𝘵𝘩 𝘈𝘐.” 643𝗧𝗿𝗮𝗻𝘀𝗹𝗮𝘁𝗶𝗼𝗻: We told investors our massive bet on AI would pay off. It didn’t. Now we need to cut you.  644 645“𝘋𝘰𝘪𝘯𝘨 𝘮𝘰𝘳𝘦 𝘸𝘪𝘵𝘩 𝘭𝘦𝘴𝘴.” 646𝗧𝗿𝗮𝗻𝘀𝗹𝗮𝘁𝗶𝗼𝗻: Tariffs, the markets, or simply bad bets have eaten our lunch. AI 𝘮𝘪𝘨𝘩𝘵 save us, but we don’t actually know how.  647 648We’ve heard some version of this nonsense since the pandemic. It likely won’t stop any time soon.   649 650Tech hiring continues to rise overall, but companies still over-hired, over-invested, and made short-sighted moves that are now being paid for by workers getting laid off.  651 652I was laid off a couple years ago.   653It was brutal.   654 655I had just bought a business, I have three kids and a mortgage. My biggest regret is not taking more time to reflect before jumping back into the job market. 656 657But I get it: you need a paycheck and the guise of stability. I did too.  658 659I ended up becoming part of the great reshuffle: moving from management and web dev into working with LLMs, AI and product engineering.  660 661If you’re impacted by layoffs, I’m sorry and I can relate.   662 663But if there’s any silver lining, it’s this: the next era of tech is being built 𝘳𝘪𝘨𝘩𝘵 𝘯𝘰𝘸. And the skills you build next could shape your entire career.  664 665Good luck out there.666 667 668 669I made a project for full stack developers who want to learn AI practically: https://lnkd.in/d6Pzfcg6",ARTICLE,Brian,Jenney,7080,0,36,3,1,0,0.005649717514124294,,2026-01-29 09:15:03,https://www.linkedin.com/feed/update/urn:li:activity:7422688743532433409670urn:li:activity:7422676002063110144,"Junior developers are inherently risky (in every industry, not just tech btw).671 672𝘉𝘶𝘵... 𝘐 𝘢𝘮 𝘢 𝘫𝘶𝘯𝘪𝘰𝘳, you say.673 674Yeah, I get it, I get it, just bare with me here.675 676The word junior is subjective. 𝗪𝗵𝗮𝘁 𝗶𝘁 𝗯𝗮𝘀𝗶𝗰𝗮𝗹𝗹𝘆 𝗯𝗼𝗶𝗹𝘀 𝗱𝗼𝘄𝗻 𝘁𝗼 𝗶𝘀 𝘀𝗼𝗺𝗲𝗼𝗻𝗲 𝘄𝗵𝗼 𝗵𝗮𝘀 𝘀𝗼 𝗹𝗶𝘁𝘁𝗹𝗲 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 𝘁𝗵𝗮𝘁 𝘁𝗵𝗲𝘆 𝗰𝗮𝗻𝗻𝗼𝘁 𝗯𝗲 𝘁𝗿𝘂𝘀𝘁𝗲𝗱 𝘁𝗼 𝗱𝗼 𝗺𝘂𝗰𝗵 𝗼𝗻 𝘁𝗵𝗲𝗶𝗿 𝗼𝘄𝗻.677 678• They need significant hand-holding for months on the job.679• They will take on the easy stuff and maybe break some things.680• They won't contribute much.681 682𝗛𝗲𝗿𝗲'𝘀 𝘁𝗵𝗲 𝘁𝗵𝗶𝗻𝗴: nearly every developer will be junior when they first join a company unless they are very senior.683 684You don't actually need to a be a mid level developer to be a safe hire. You just need to avoid the tell-tale signs that you are, in fact, a n00b.685 686Quite simply, stop down-playing yourself.687 688• 𝗥𝗲𝗺𝗼𝘃𝗲 𝗷𝘂𝗻𝗶𝗼𝗿, 𝗮𝘀𝗽𝗶𝗿𝗶𝗻𝗴 𝗮𝗻𝗱 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 from your resume and LinkedIn.689• DON'T talk about the project from your school or bootcamp690• 𝗗𝗢 𝗧𝗮𝗹𝗸 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗮𝗽𝗽 𝗼𝗿 𝘄𝗲𝗯𝘀𝗶𝘁𝗲 𝘆𝗼𝘂'𝗿𝗲 𝗹𝗮𝘂𝗻𝗰𝗵𝗶𝗻𝗴 - yes, you need to have a project to talk about.691• 𝗨𝘀𝗲 𝘀𝘁𝗿𝗼𝗻𝗴 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝘁𝗼 𝘁𝗮𝗹𝗸 𝗮𝗯𝗼𝘂𝘁 𝘆𝗼𝘂𝗿 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 and speak to the benefits of features you created. 692• Instead of created a full stack app with a laundry list of tech try migrated a JS app to Typescript to improve developer experience and velocity.693 694I break down why you need to be building a product instead of a portfolio in 2026 with some resources to help you get started: https://lnkd.in/gNuYpYfR",ARTICLE,Brian,Jenney,2063,0,20,3,0,0,0.011148812409112942,,2026-01-29 08:24:25,https://www.linkedin.com/feed/update/urn:li:activity:7422676002063110144695urn:li:activity:7421979306819772416,"Apparently, if you’re not using:696 697• Opus 4.5 (but only for certain tasks)698• A different model for planning vs coding vs refactoring699• Custom rule files700• Per-project AI configs701• MD spec / rules docs702• Spec-driven planning phases703• Git subtrees704• “Ralph loops”705• Spec kit706• Carefully segmented prompting workflows707 708…then you’re ""not using AI correctly.""709 710To be fair, some of this is spot on. 711 712𝗕𝘂𝘁 𝗮𝘁 𝘀𝗼𝗺𝗲 𝗽𝗼𝗶𝗻𝘁 𝘄𝗲 𝘀𝗵𝗼𝘂𝗹𝗱 𝗽𝗮𝘂𝘀𝗲 𝗮𝗻𝗱 𝗮𝘀𝗸:713 714If it takes this much infrastructure to get AI to help us write decent code…715 716𝗜𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝗿𝗲𝗮𝗹𝗹𝘆 𝘁𝗵𝗲 𝘂𝘀𝗲𝗿?717 718Or is the tooling still not as intuitive as we pretend it is?719 720The biggest performance gains I’ve seen don’t come from elaborate toolchains. They come from:721 722• Understanding the problem723• Being able to describe the problem clearly724• Giving the model the right context (look at 𝚊𝚐𝚎𝚗𝚝𝚜.𝚝𝚜 as an example to scaffold the next agent)725• Reviewing and iterating726 727Yes, rules files help. 728Yes, better workflows help. 729Yes, model choice matters.730 731But if someone needs a 14-step AI ritual just to build a CRUD app, we may have crossed from ""power tools"" into magic spells.732 733𝗔𝗿𝗲 𝘄𝗲 𝗮𝘂𝗴𝗺𝗲𝗻𝘁𝗶𝗻𝗴 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀 𝗼𝗿 𝗷𝘂𝘀𝘁 𝗮𝗱𝗱𝗶𝗻𝗴 𝗮 𝘀𝗲𝗰𝗼𝗻𝗱 𝗷𝗼𝗯 𝗮𝘀 𝗮𝗻 𝗔𝗜 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗼𝗽𝗲𝗿𝗮𝘁𝗼𝗿?734 735I'm not saying either is wrong... I'm just saying.",TEXT,Brian,Jenney,17788,0,88,42,5,0,0.00758938610299078,,2026-01-27 10:16:00,https://www.linkedin.com/feed/update/urn:li:activity:7421979306819772416736urn:li:activity:7421597826318618625,"""𝘞𝘢𝘳 𝘪𝘴 𝘱𝘦𝘢𝘤𝘦. 𝘍𝘳𝘦𝘦𝘥𝘰𝘮 𝘪𝘴 𝘴𝘭𝘢𝘷𝘦𝘳𝘺. 𝘐𝘨𝘯𝘰𝘳𝘢𝘯𝘤𝘦 𝘪𝘴 𝘴𝘵𝘳𝘦𝘯𝘨𝘵𝘩.""737 738I can only dissociate through code and AI hot takes for so long. It feels like a dark time is upon us in the US. But I'm always optimistic.739 740You may not know me personally, but I hope you're doing alright and keeping sane in an increasingly chaotic world.741 742RIP Alex Pretti",TEXT,Brian,Jenney,1237,0,29,0,1,0,0.024252223120452707,,2026-01-26 09:00:08,https://www.linkedin.com/feed/update/urn:li:activity:7421597826318618625743urn:li:activity:7420542285995712512,This guy said it better than I can.,SHARE,Brian,Jenney,1604,0,15,4,0,0,0.011845386533665835,,2026-01-23 11:05:47,https://www.linkedin.com/feed/update/urn:li:activity:7420542285995712512744urn:li:activity:7420152017597366274,"If you’re using AI to write your code, here’s something you should know.745 746If you are:747 748• using AI to generate production code749• using AI to generate the tests750• using AI to validate its own output751• and then shipping it752 753Then you are MORE likely than a human to ship bugs and vulnerabilities.754 755Specifically, you are more likely than a human to:756 757• Introduce security vulnerabilities (~20-25% higher likelihood)758• Ship critical or major bugs (≈1.5–1.7× more per PR)759• Miss edge cases and failure paths in tests760• Over-test happy paths and under-test bad inputs761• Have “clean” builds that hide production risks762 763AI-written tests are less likely to test edge cases and over-use mocks.764 765LGTM! 😅766 767This doesn’t mean “don’t use AI.”768 769It means 100% AI-generated systems should make you pause and consider the risks.770 771Output scales fast. Judgment and maintenance won't.",TEXT,Brian,Jenney,5108,0,28,10,2,0,0.007830853563038372,,2026-01-22 09:15:00,https://www.linkedin.com/feed/update/urn:li:activity:7420152017597366274772urn:li:activity:7419455185086914560,"If you're a Typescript developer and you want to get started with AI agents, just do this:773 7741. Put $5 into OpenAI's API platform (or use Gemini's free tier) https://lnkd.in/gPM8D2Pq7752. Install Vercel's AI SDK: https://lnkd.in/gh-sND6r7763. Read up on common patterns for agents (human in the loop, ReAct, orchestration): https://lnkd.in/g3ekb3xr7774. Create a simple agent using tool-calling to do an internet search and summarize the results. You can use this API: https://www.searchapi.io/7785. Re-build your agent with a popular framework like LangGraph: https://lnkd.in/g9yB-_ec7796. Add unit tests and/or evals for your agent7807. Add an observability layer with Helicone or Langsmith: https://www.helicone.ai/7818. Grab this project to build your first agent: https://lnkd.in/gMv3x3tS782 783𝗦𝘁𝗲𝗽 𝟴.𝟱 - spend less time online arguing about whether or not LLMs will replace developers with people who can't solve FizzBuzz.",TEXT,Brian,Jenney,2454,0,33,1,0,0,0.013854930725346373,,2026-01-20 11:06:02,https://www.linkedin.com/feed/update/urn:li:activity:7419455185086914560784urn:li:activity:7419046027644678144,"Nearly 100% of my code is AI-generated.785 786That doesn’t mean I’m not building software - it just means I’m not the one writing the code.787 788It's a weird time to be a coder but there's also massive opportunity hiding behind the hype and fear mongering...",ARTICLE,Brian,Jenney,4972,0,33,10,0,0,0.008648431214802896,,2026-01-19 08:00:12,https://www.linkedin.com/feed/update/urn:li:activity:7419046027644678144789urn:li:activity:7417986432369078272,Wonder if Claude can fix the terminal flicker next 😅,SHARE,Brian,Jenney,3854,0,8,0,0,0,0.0020757654385054488,,2026-01-16 09:49:44,https://www.linkedin.com/feed/update/urn:li:activity:7417986432369078272790urn:li:activity:7417349877958914048,"Do not attend if:791 7921. You're studying for Google, Meta, Netflix or massive tech orgs7932. You want some magical prompts to ""hack"" the interview7943. You'd just use a cache 😉795 796We're going to cover 3.5 patterns to help you nail your next technical interview outside big tech.",EVENT,Brian,Jenney,4318,0,34,3,0,0,0.00856878184344604,,2026-01-14 15:40:18,https://www.linkedin.com/feed/update/urn:li:activity:7417349877958914048797urn:li:activity:7417264206913200128,"We get it. AI wRiTeS thE COde!798 799While most devs are optimizing their prompts, I think the fun (and opportunity) is a couple layers below:800 8011. 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁𝘀 using frameworks like LangChain, Vercel's AI SDK or rolling your own8022. Learning how to to 𝗶𝗺𝗽𝗿𝗼𝘃𝗲 𝘀𝗲𝗮𝗿𝗰𝗵 𝗾𝘂𝗮𝗹𝗶𝘁𝘆 𝗶𝗻 𝘃𝗲𝗰𝘁𝗼𝗿 𝗱𝗮𝘁𝗮𝗯𝗮𝘀𝗲𝘀 like Qdrant and Pinecone8033. 𝗦𝗰𝗿𝗮𝗽𝗶𝗻𝗴 𝗶𝗻 𝘁𝗵𝗲 𝗔𝗜 𝗲𝗿𝗮 using tools like Firecrawl and 𝚌𝚛𝚊𝚠𝚕𝟺𝚊𝚒8044. 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 - basically figuring out how to organize what an LLM needs to know at what stage to give you a good response8055. 𝗟𝗟𝗠-𝗼𝗽𝘀 𝗮𝗻𝗱 𝗼𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆. Monitoring token costs and model drift to see what breaks and why.806 807While everyone is glazing Claude or arguing over replacement theories, you could build up these skills and be prepared for software 3.0",TEXT,Brian,Jenney,3103,0,27,5,0,0,0.0103126007089913,,2026-01-14 09:59:53,https://www.linkedin.com/feed/update/urn:li:activity:7417264206913200128808urn:li:activity:7416910212789542912,"Challenge accepted.809 810Last year I spent a lot of time rallying against AI hype.811 812I also said a lot of things that are now less true than they were 6 months ago.813 814The models have improved.815Our collective workflows have improved.816 817I'm honestly not sure where the limit is and I'm excited John Crickett is working to find it.",SHARE,Brian,Jenney,1940,0,21,1,0,0,0.01134020618556701,,2026-01-13 10:33:14,https://www.linkedin.com/feed/update/urn:li:activity:7416910212789542912818urn:li:activity:7416886820073869312,"2 things:819 8201. Are CEOs of large companies really begging their eng department for... anything? Let alone to build a prototype? Why?8212. If you ARE doing this... cool! Using app-gen tools like v0, Lovable or Replit are great ways to create a ""cheap"" proof of concept with basic functionality.822 823Oh and lastly, if you're one of those dudes who claims they built an app over the weekend that is now doing 10K MRR and you have no clue how it works... you can take a long walk off a short bridge.",IMAGE,Brian,Jenney,25578,0,48,16,2,0,0.0025803424818203143,,2026-01-13 09:00:16,https://www.linkedin.com/feed/update/urn:li:activity:7416886820073869312824urn:li:activity:7416520611809640448,"White board interviews are biased, archaic and don't reflect the kind of work we actually do as developers.825 826Ok.827 828Let's pretend you're correct.829 830What do you do now?831 832You can either:833 834a. Avoid DSA and all companies that ask these questions (𝘐 𝘵𝘳𝘪𝘦𝘥 𝘵𝘩𝘪𝘴 𝘧𝘰𝘳 𝘺𝘦𝘢𝘳𝘴 𝘸𝘪𝘵𝘩 𝘷𝘢𝘳𝘺𝘪𝘯𝘨 𝘳𝘦𝘴𝘶𝘭𝘵𝘴)835b: Learn some of the most common patterns and data structures that make up the majority of questions out there836 837You can still avoid whiteboard interviews, because you WANT to, not because you HAVE to.838 839 840PS 👇841 842I break down 3 patterns you WILL encounter on the interview circuit and a free guide for developers who want to learn through hands-on challenges: https://lnkd.in/gEYY9sF5",TEXT,Brian,Jenney,1024,0,8,1,0,0,0.0087890625,,2026-01-12 08:45:06,https://www.linkedin.com/feed/update/urn:li:activity:7416520611809640448843urn:li:activity:7415784721013506048,"During a live coding interview last year, I instinctively switched from Cursor to VS Code.844 845The CTO stopped me.846 847“No — leave Cursor on. We want to see how you solve this with AI.”848 849I thought the interview would be easier.850 851It wasn’t.",ARTICLE,Brian,Jenney,9743,0,46,5,1,0,0.005337165144206096,,2026-01-10 08:00:56,https://www.linkedin.com/feed/update/urn:li:activity:7415784721013506048852urn:li:activity:7415238834323636224,"So you can just write Python in your NextJS app now.853 854What will those maniacs at Vercel cook up next? 🙇‍♂️",IMAGE,Brian,Jenney,2598,0,31,10,0,0,0.01578137028483449,,2026-01-08 19:51:46,https://www.linkedin.com/feed/update/urn:li:activity:7415238834323636224855urn:li:activity:7414349222793117696,"I'm genuinely excited to walkthrough any of these topics. 856 857If there's something else you'd like to see, let us know in the comments or DM me.",SHARE,Brian,Jenney,1411,0,8,0,1,0,0.0063784549964564135,,2026-01-06 08:56:46,https://www.linkedin.com/feed/update/urn:li:activity:7414349222793117696858urn:li:activity:7413993429594103808,"My absolute worst interview experiences:859 8601. That time the guy at Google told me ""𝘵𝘩𝘪𝘴 𝘪𝘴 𝘸𝘩𝘺 𝘸𝘦 𝘥𝘰𝘯'𝘵 𝘩𝘪𝘳𝘦 𝘴𝘦𝘭𝘧-𝘵𝘢𝘶𝘨𝘩𝘵 𝘥𝘦𝘷𝘦𝘭𝘰𝘱𝘦𝘳𝘴"" when I struggled to implement a BST.8612. When I ended the Meta phone screen in the first 15 minutes after I was asked a problem that involved recursion.8623. The interview that turned into a lesson on objects in JS when I thought {} === {} 🤦 (𝘴𝘵𝘪𝘭𝘭 𝘦𝘮𝘣𝘢𝘳𝘳𝘢𝘴𝘴𝘦𝘥 𝘣𝘺 𝘵𝘩𝘪𝘴 𝘰𝘯𝘦)8634. Had a literal panic attack during the last stage for a role I wanted. It was completely non-technical too. I asked if we could start over... 😅864 865Listen, I've sucked at interviews and through lots of trial, error, practice and more practice, I forced myself to suck less.866 867I break down the exact things I did in this video and give away my repo full of coding challenges (𝘣𝘶𝘵 𝘰𝘯𝘭𝘺 𝘧𝘰𝘳 𝘢𝘣𝘰𝘷𝘦 𝘢𝘷𝘦𝘳𝘢𝘨𝘦 𝘭𝘰𝘰𝘬𝘪𝘯𝘨 𝘥𝘦𝘷𝘦𝘭𝘰𝘱𝘦𝘳𝘴): https://lnkd.in/gt4Uzrmm",TEXT,Brian,Jenney,135072,0,203,49,0,0,0.0018656716417910447,,2026-01-05 09:22:58,https://www.linkedin.com/feed/update/urn:li:activity:7413993429594103808868urn:li:activity:7413981477610303491,"My mom kicked off my coding career with an intervention.869 870Over a decade ago, my mom walked into my girlfriend's condo and handed me a note with an ultimatum:871 872Get sober or get out of your kids lives.873 874I was a mess.875 876I owed people money. I was getting threats on my phone. I almost lost my life in a robbery gone wrong. My friend had taken his life and his brother got sentenced to state time.877 878It was only a matter of time before I lost my life or my freedom.879 880Didn't matter. I wasn't ready to quit yet.881 882I told her I'd try just so she'd stop crying.883 884One day turned to a few days. Then a week went by. Then a month.885 886I either couldn't sleep or would sleep for 12 hours. I ate too much candy. I lost ""friends."" I got better.887 888Coding became my new addiction. It didn't make any sense - I had no technical background and didn't own a computer for most of my life but I loved solving problems with code.889 890This new addiction led to a new career and my habits snowballed.891 892I lost weight, stopped smoking and picked up reading. I became a better father.893 894𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘁𝗼 𝗰𝗼𝗱𝗲 𝗱𝗶𝗱 𝗡𝗢𝗧 𝗰𝗵𝗮𝗻𝗴𝗲 𝗺𝘆 𝗹𝗶𝗳𝗲. Getting sober did. Coding certainly changed my wallet however.895 896Sometimes I reveal this embarrassing aspect of my life because I know how it feels to feel like you're alone or like the odds are impossibly stacked against you.897 898If you're going through something similar I hope you know it's not impossible and the world can really open up once you get of your own way.899 900I break down some practical steps to building habits that stick here and hopefully you find it helpful: https://lnkd.in/gGzwfbSW",ARTICLE,Brian,Jenney,1792,0,27,4,1,0,0.017857142857142856,,2026-01-05 08:35:29,https://www.linkedin.com/feed/update/urn:li:activity:7413981477610303491901urn:li:activity:7412920350268846080,"If learning JavaScript feels like you’re starting over every few months,  just stop. 902 903The issue is the way you’re learning.904 905Here’s the 5-step system I’d use in 2026:906 9071. 𝗞𝗶𝗹𝗹 𝘁𝘂𝘁𝗼𝗿𝗶𝗮𝗹 𝗮𝗱𝗱𝗶𝗰𝘁𝗶𝗼𝗻: Watch just enough to start, then pause. Rewrite, extend, or break the example before you hit play again.908 9092. 𝗠𝗮𝘀𝘁𝗲𝗿 𝗳𝘂𝗻𝗱𝗮𝗺𝗲𝗻𝘁𝗮𝗹𝘀: Variables, functions, arrays/objects, this/binding, async/await. Litmus test: build a vanilla JS form with validation that posts to an API.910 9113. 𝗨𝘀𝗲 𝗯𝘂𝘁𝘁𝗼𝗻-𝗱𝗿𝗶𝘃𝗲𝗻 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁: Add a button, decide what should happen, then make it real. Ship tiny experiments in CodePen/StackBlitz.912 9134. 𝗚𝗲𝘁 𝗳𝗲𝗲𝗱𝗯𝗮𝗰𝗸: Mentor > meetups > paid review > post your code. Pair program with someone slightly ahead of you.914 9155. 𝗕𝘂𝗶𝗹𝗱 𝗮 𝘀𝘆𝘀𝘁𝗲𝗺: 10–30 minutes daily beats 8 hours on Saturday. Plan tomorrow’s task tonight. Keep the habit alive.916 917If you want structure, accountability, and feedback to actually stick the fundamentals, start with Dev30 (30 days, JS fundamentals with office hours): 918 919https://dev30.xyz920 921I break down how I'd be learning JS in 2026 if I was starting over here:  https://lnkd.in/gkPW5nUP",TEXT,Brian,Jenney,6625,0,40,4,1,0,0.006792452830188679,,2026-01-02 10:18:56,https://www.linkedin.com/feed/update/urn:li:activity:7412920350268846080922urn:li:activity:7412546240384794625,"The 2026 goal for Parsity is simple:923 924More student success.925 926This is a small subset of congratulatory posts I found in Slack. I've said ""𝙘𝙤𝙣𝙜𝙧𝙖𝙩𝙪𝙡𝙖𝙩𝙞𝙤𝙣𝙨!"" to more people than I can remember over the years.927 928It never gets old.929 930This market is not easy for career changers.931 932It's also not impossible.933 934There is no silver bullet either.935 936Some were hired before completing our program. 937Others took months to land a role. 938Some students did internships. 939Some built in public. 940Some mass applied.941 942You have enough doom and gloom in your feed. 943 944Whether or not you ever consider joining Parsity (https://parsity.io) - I hope this gives you a little boost heading into the new year.",IMAGE,Brian,Jenney,1824,0,56,7,0,0,0.03453947368421053,,2026-01-01 09:32:22,https://www.linkedin.com/feed/update/urn:li:activity:7412546240384794625945urn:li:activity:7411805138450935808,"Here are 4 mistakes I made when building agents:946   947❌ 𝗧𝗵𝗶𝗻𝗸𝗶𝗻𝗴 𝗜 𝗻𝗲𝗲𝗱𝗲𝗱 𝗮𝗻 𝗮𝗴𝗲𝗻𝘁 𝗶𝗻 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝗽𝗹𝗮𝗰𝗲948✅ If the steps are deterministic and predictable, 𝗮 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗶𝘀 𝘂𝘀𝘂𝗮𝗹𝗹𝘆 𝘁𝗵𝗲 𝗯𝗲𝘁𝘁𝗲𝗿 𝗼𝗽𝘁𝗶𝗼𝗻.949Agents make sense when planning, exploration, or autonomy are required.950 951❌ 𝗣𝗮𝘀𝘀𝗶𝗻𝗴 𝘁𝗼𝗼 𝗺𝘂𝗰𝗵 (𝗼𝗿 𝘁𝗵𝗲 𝘄𝗿𝗼𝗻𝗴) 𝗰𝗼𝗻𝘁𝗲𝘅𝘁952✅ Good agents get the right information at the right moment. Context should be intentional and managed, not a full memory dump.953 954❌ 𝗣𝗿𝗼𝗺𝗽𝘁 (𝗼𝘃𝗲𝗿)𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴955✅ Clear structure, consistent formatting, and a few concrete examples outperform clever wording.956 957❌ 𝗨𝘀𝗶𝗻𝗴 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗺𝗼𝗱𝗲𝗹 𝗳𝗼𝗿 𝗲𝘃𝗲𝗿𝘆 𝘁𝗮𝘀𝗸958✅ Different tasks benefit from different models.959Routing, extraction, and classification don’t need heavyweight reasoning.960 961I break down more common  pitfalls (and how I avoid them now) in this video 👇962 963https://lnkd.in/gv-Nqjuz",ARTICLE,Brian,Jenney,1768,0,8,1,0,0,0.005090497737556561,,2025-12-30 08:27:29,https://www.linkedin.com/feed/update/urn:li:activity:7411805138450935808964urn:li:activity:7411483215850508288,"Typical advice to junior developers on Reddit:965 966The market is cooked.967No one is hiring juniors.968AI has replaced you.969Just give up.970 971Truly, thought provoking stuff.972 973Let's go past the headlines and edge-lord takes and learn what you can practically do to create a larger surface area for luck going into 2026.",ARTICLE,Brian,Jenney,3748,0,30,16,2,0,0.012806830309498399,,2025-12-29 11:08:17,https://www.linkedin.com/feed/update/urn:li:activity:7411483215850508288974urn:li:activity:7411462800864272384,"2025 was the year of layoffs, hype cycles, and broken promises. 975 976Some trends clearly moved the industry forward and others did real damage. 977 978Here’s the best and absolute worst tech trends coming out of 2025 (IMO).979 980❌ 𝗪𝗼𝗿𝘀𝘁 𝗧𝗿𝗲𝗻𝗱𝘀981 982𝟭. 𝗕𝗼𝗼𝘁𝗰𝗮𝗺𝗽𝘀 𝗰𝗼𝗹𝗹𝗮𝗽𝘀𝗶𝗻𝗴983Sentiment about coding is at an all time low. So many bootcamps closed. 𝗧𝗵𝗲 𝗰𝘆𝗻𝗶𝗰𝘀 𝗮𝗿𝗲 𝗵𝗮𝗽𝗽𝘆 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗶𝘀. It does leave a massive gap for adults who want to change careers. This makes me sad.984 985𝟮. 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗵𝘆𝗽𝗲 𝗿𝗲𝗽𝗹𝗮𝗰𝗲𝗱 𝗺𝗲𝗮𝘀𝘂𝗿𝗲𝗺𝗲𝗻𝘁986We were told output would explode. Instead, we got more bugs, more rewrites, and more pressure to ship faster without better systems in place.987 988𝟯. 𝗝𝘂𝗻𝗶𝗼𝗿𝘀 𝗴𝗼𝘁 𝘀𝗾𝘂𝗲𝗲𝘇𝗲𝗱989“Entry-level” now means “ready on day one.”990 991✅ 𝗕𝗲𝘀𝘁 𝗧𝗿𝗲𝗻𝗱𝘀992 993𝟭. 𝗕𝗼𝗼𝘁𝗰𝗮𝗺𝗽𝘀 𝗰𝗼𝗹𝗹𝗮𝗽𝘀𝗶𝗻𝗴 (𝘆𝘂𝗽, 𝗼𝗻 𝗯𝗼𝘁𝗵 𝗹𝗶𝘀𝘁𝘀)9943 months to 100K is dead. This creates room for honest education: no timelines, no guarantees, no pretending this is easy. That’s a win long-term IMO.995 996𝟮. 𝗦𝗲𝗻𝗶𝗼𝗿 𝗮𝗻𝗱 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝘁 𝗱𝗲𝗺𝗮𝗻𝗱 𝗾𝘂𝗶𝗲𝘁𝗹𝘆 𝘀𝘂𝗿𝗴𝗲𝗱997Smaller, leaner teams meant more demand for people who are full stack and senior.998 999𝟯. 𝗔𝗜 𝘁𝗼𝗼𝗹𝘀 𝗴𝗼𝘁 𝗴𝗲𝗻𝘂𝗶𝗻𝗲𝗹𝘆 𝘂𝘀𝗲𝗳𝘂𝗹1000Claude and Cursor went from fun little tools to core infrastructure with crazy high retention among developers. Agents are moving past the hype cycle into boring, practical use cases.1001 1002𝟰. 𝗛𝗶𝗿𝗶𝗻𝗴 𝘀𝘁𝗮𝗯𝗶𝗹𝗶𝘇𝗲𝗱 𝗶𝗻𝘀𝘁𝗲𝗮𝗱 𝗼𝗳 “𝗯𝗼𝗼𝗺𝗶𝗻𝗴”10032025 will end with more open roles than it began with. The market is lop-sided but trending in a positive direction.1004 1005𝗛𝗼𝘄 𝘁𝗼 𝘁𝗮𝗸𝗲 𝗮𝗱𝘃𝗮𝗻𝘁𝗮𝗴𝗲 𝗼𝗳 𝘁𝗵𝗶𝘀 𝗴𝗼𝗶𝗻𝗴 𝗶𝗻𝘁𝗼 𝟮𝟬𝟮𝟲:10061. 	Stop chasing shortcuts and build projects over portfolios10072. 	Learn how systems work by reading up on system design10083. 	Treat education like a long-term investment, not a lottery ticket10094. If you're front-end-only, expand your skillset towards backend1010 1011Curious what trends you think I missed. Or got wrong.1012 1013https://lnkd.in/gc3S2Rhm",ARTICLE,Brian,Jenney,1350,0,18,1,0,0,0.014074074074074074,,2025-12-29 09:47:10,https://www.linkedin.com/feed/update/urn:li:activity:74114628008642723841014urn:li:activity:7410391300547739649,"""Maybe you should just quit.""1015 1016I honestly didn't expect Michael Greenspan to be so direct during a meeting we had a few months back.1017 1018I was unhappy, stressed and burnt out.1019 1020Michael saw that and led me to the same conclusion I was trying to avoid.1021 1022We talked through the worst case scenario which honestly wasn't so bad. My kids wouldn't go hungry or lose the roof over their head. This wasn't life or death. It just felt like it.1023 1024I made a plan and handed in my 2 weeks notice.1025 1026Now what?1027 1028• I spent the first 2 weeks on Parsity's new AI curriculum and launched our first cohort which is nearly sold out. This was so much fun to create.1029• Renewed focus on YouTube to attract students and pumped out more videos. Nothing, nothing, nothing then 1 video hit well over 100K views. So many trolls. Way more love.1030• Accepted an offer with a company and team I'm excited to work with. I swear the lord works in mysterious ways.1031• That massive tension headache I'd gotten used to disappeared.1032 1033Thanks Michael Greenspan for giving me room to vent and pushing me to do the uncomfortable.",TEXT,Brian,Jenney,6438,0,81,5,0,0,0.013358185771978875,,2025-12-26 10:49:24,https://www.linkedin.com/feed/update/urn:li:activity:74103913005477396491034urn:li:activity:7409675055577763842,"AI didn’t fail this year.1035 1036It failed expectations.1037 1038No mass job extinction. No solo unicorn founders. No “prompt your way to riches.”1039 1040Reality was less marketable:1041 1042• uneven usefulness1043• uneven adoption1044• a widening gap between people who know how to think and people hoping AI will think for them1045 1046Developers felt this first.1047 1048AI raised expectations. Output increased. So did bugs. We all felt a historic number of outages.1049 1050𝗠𝘆 𝟮𝟬𝟮𝟲 𝗽𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝗼𝗻𝘀1051 10521. 𝘈𝘐 𝘴𝘦𝘵𝘵𝘭𝘦𝘴 𝘪𝘯𝘵𝘰 𝘪𝘯𝘧𝘳𝘢𝘴𝘵𝘳𝘶𝘤𝘵𝘶𝘳𝘦.1053 1054Agents improve in reliability and scope. Teams stop chasing novelty and start standardizing workflows. Excitement fades. Usefulness increases.1055 1056Agent design patterns solidify the same way frameworks and cloud architectures eventually did.1057 10582. 𝘓𝘦𝘢𝘳𝘯𝘪𝘯𝘨 𝘤𝘩𝘢𝘯𝘨𝘦𝘴 𝘴𝘩𝘢𝘱𝘦1059 1060Learners reach for AI as a teacher first. Tutorials lose relevance because the fastest path to understanding becomes interactive and contextual.1061 1062I also don't believe this will work. Hallucinations and overly general learning paths frustrate users moving past the beginner stage.1063 1064People rediscover mentorship and opt for more human experiences.1065 10663. 𝘛𝘦𝘤𝘩 𝘪𝘯𝘵𝘦𝘳𝘷𝘪𝘦𝘸𝘴 𝘨𝘦𝘵... 𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘵1067 1068 Interviews evolve around AI fluency.1069 1070Candidates are evaluated on how they reason with tools, how they validate output, and how they recover when AI is confidently wrong. The skill becomes collaboration && delegation.1071 10724. 𝘈𝘐 𝘱𝘳𝘦𝘴𝘴𝘶𝘳𝘦 𝘴𝘱𝘳𝘦𝘢𝘥𝘴 𝘵𝘰 𝘰𝘧𝘧𝘪𝘤𝘦 𝘸𝘰𝘳𝘬𝘦𝘳𝘴1073 1074Office and knowledge workers begin to feel the same pressure developers felt earlier to ""move faster""1075 1076They’re expected to produce more with AI but given little guidance on how to structure work, evaluate output, or integrate tools into real processes.1077 1078That gap creates space for a new role: people who translate ambiguous business needs into structured workflows, systems, and AI-assisted processes. Not traditional developers but systems-oriented operators embedded in non-technical teams.1079 1080 1081 1082 1083How off base am I?",TEXT,Brian,Jenney,9953,0,56,21,1,0,0.007836833115643524,,2025-12-24 11:23:18,https://www.linkedin.com/feed/update/urn:li:activity:74096750555777638421084urn:li:activity:7409273301421744129,"LGTM 👌1085 1086This won't be a shocker to developers who use AI coding tools everyday but it makes you wonder:1087 1088𝘐𝘴 𝘵𝘩𝘦 𝘫𝘶𝘪𝘤𝘦 𝘸𝘰𝘳𝘵𝘩 𝘵𝘩𝘦 𝘴𝘲𝘶𝘦𝘦𝘻𝘦?1089 1090𝘐𝘴 𝘪𝘵 𝘱𝘶𝘳𝘦 𝘤𝘰𝘪𝘯𝘤𝘪𝘥𝘦𝘯𝘤𝘦 𝘵𝘩𝘢𝘵 𝘴𝘰 𝘮𝘢𝘯𝘺 𝘮𝘢𝘫𝘰𝘳 𝘰𝘶𝘵𝘢𝘨𝘦𝘴 𝘩𝘢𝘷𝘦 𝘰𝘤𝘤𝘶𝘳𝘳𝘦𝘥 𝘵𝘩𝘪𝘴 𝘺𝘦𝘢𝘳 𝘢𝘴 𝘸𝘦 𝘳𝘦𝘭𝘺 𝘰𝘯 𝘰𝘶𝘳 𝘈𝘐 ""𝘤𝘰-𝘸𝘰𝘳𝘬𝘦𝘳𝘴""?1091 1092The tools are here to stay. I'm not cancelling my Claude or Cursor subscriptions but reading CodeRabbit's report validates what many of us have experienced.1093 1094AI coding tools can be amazing BUT:1095 1096• They introduce 𝗺𝗼𝗿𝗲 𝗰𝗿𝗶𝘁𝗶𝗰𝗮𝗹 & 𝗺𝗮𝗷𝗼𝗿 𝗶𝘀𝘀𝘂𝗲𝘀1097• 𝗘𝗿𝗿𝗼𝗿 𝗵𝗮𝗻𝗱𝗹𝗶𝗻𝗴 𝗴𝗮𝗽𝘀 show up far more often1098• Performance bugs are rarer, but 𝗱𝗶𝘀𝗽𝗿𝗼𝗽𝗼𝗿𝘁𝗶𝗼𝗻𝗮𝘁𝗲𝗹𝘆 𝗔𝗜-𝗱𝗿𝗶𝘃𝗲𝗻1099• 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗺𝗶𝘀𝘁𝗮𝗸𝗲𝘀 scale faster than human ones1100• AI PRs often 𝗰𝗼𝘀𝘁 𝗺𝗼𝗿𝗲 𝗿𝗲𝘃𝗶𝗲𝘄 𝘁𝗶𝗺𝗲, not less1101 1102This isn't some ""dunk"" on AI-generated code. It's better the devil you know than the devil you don't.1103 1104LGTM!",IMAGE,Brian,Jenney,3093,0,27,5,2,0,0.010992563853863564,,2025-12-23 08:46:52,https://www.linkedin.com/feed/update/urn:li:activity:74092733014217441291105urn:li:activity:7408902577347653632,"5 things that nearly derailed my developer career:  1106 11071. 𝗡𝗼𝘁 𝗻𝗲𝗴𝗼𝘁𝗶𝗮𝘁𝗶𝗻𝗴. I would've job hopped a lot less if I had been up front with what I wanted. 11082. 𝗧𝗵𝗶𝗻𝗸𝗶𝗻𝗴 ""𝗽𝗹𝗮𝘆𝗶𝗻𝗴 𝗶𝘁 𝘀𝗮𝗳𝗲"" 𝘄𝗮𝘀 𝘀𝗮𝗳𝗲. No opinion, no questions... no visibility. 11093. 𝗔𝘃𝗼𝗶𝗱𝗶𝗻𝗴 𝗗𝗦𝗔. Seriously, it's not that hard to learn and gave me more tools to solve coding problems. 11104. 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸𝘀 𝗼𝘃𝗲𝗿 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲𝘀. It's really tough to debug or move between frameworks when you don't have a solid foundation. 11115. 𝗥𝘂𝗯𝗯𝗲𝗿-𝘀𝘁𝗮𝗺𝗽𝗶𝗻𝗴 𝗰𝗼𝗱𝗲 𝗿𝗲𝘃𝗶𝗲𝘄𝘀. This is embarrassing - I let too many issues slide by because I thought I had better things to do 😬  1112 1113If I was starting off today, I'm sure I'd be over-relying on AI tools instead of StackOverflow and following the advice of every bald headed influencer online who I think has it figured it out.  1114 1115Hindsight is 20/20 and it's one of the reasons I write online in the hopes maybe you'll avoid some of these pitfalls (and maybe even join me at Parsity: https://parsity.io).1116 1117If you've written code for a while, what are some mistakes you wish you avoided when you first started?",TEXT,Brian,Jenney,7907,0,40,10,0,0,0.006323510813203491,,2025-12-22 08:13:45,https://www.linkedin.com/feed/update/urn:li:activity:74089025773476536321118urn:li:activity:7407834608500789248,"""𝘞𝘦 𝘴𝘩𝘰𝘶𝘭𝘥 𝘣𝘦 𝘢𝘣𝘭𝘦 𝘵𝘰 𝘣𝘶𝘪𝘭𝘥 𝘵𝘸𝘪𝘤𝘦 𝘢𝘴 𝘧𝘢𝘴𝘵 𝘸𝘪𝘵𝘩 𝘩𝘢𝘭𝘧 𝘵𝘩𝘦 𝘱𝘦𝘰𝘱𝘭𝘦.""1119 1120That was the promise of AI.1121 1122It's failing a lot of developers.1123 1124Leaders believe the hype. Teams get thrown into AI workflows with no training. And we're all supposed to ""figure it out"" from viral posts.1125 1126𝗛𝗲𝗿𝗲'𝘀 𝘁𝗵𝗲 𝘀𝘆𝘀𝘁𝗲𝗺 𝘁𝗵𝗮𝘁 𝘄𝗼𝗿𝗸𝘀 𝗳𝗼𝗿 𝗺𝗲 (𝗳𝗼𝗿 𝗻𝗼𝘄 𝗮𝘁 𝗹𝗲𝗮𝘀𝘁):1127 11281. 𝗣𝗹𝗮𝗻 𝗳𝗶𝗿𝘀𝘁: I voice out the idea to ChatGPT to clarify direction and confirm my understanding.11292. 𝗙𝗶𝗻𝗱 𝗮 𝗽𝗮𝘁𝘁𝗲𝗿𝗻:  Either existing examples or one I write myself to feed my robot.11303. 𝗦𝘁𝗮𝗿𝘁 𝘁𝗶𝗻𝘆:  Never trust AI with big refactors upfront.11314. 𝗗𝗲𝗾𝘂𝗲𝘂𝗲 𝘁𝗮𝘀𝗸𝘀: Once a pattern works, I feed the tool more tasks.11325. 𝗥𝗲𝘃𝗶𝗲𝘄 𝗺𝗮𝗻𝘂𝗮𝗹𝗹𝘆: Draft PR, check everything, then commit.1133 1134𝘞𝘩𝘦𝘯 𝘐'𝘮 𝘦𝘹𝘱𝘦𝘳𝘪𝘮𝘦𝘯𝘵𝘪𝘯𝘨 𝘰𝘳 𝘴𝘵𝘳𝘦𝘴𝘴-𝘵𝘦𝘴𝘵𝘪𝘯𝘨 𝘢𝘯 𝘪𝘥𝘦𝘢?1135 1136I ignore the rules and let AI run wild just to see if the approach isn't a waste. Small experiments are cheaper to try with AI.1137 1138I break down some practical ways I do and absolutely do NOT use AI in this article: https://lnkd.in/gfHut-7q",ARTICLE,Brian,Jenney,1935,0,16,8,0,0,0.012403100775193798,,2025-12-19 09:30:01,https://www.linkedin.com/feed/update/urn:li:activity:74078346085007892481139urn:li:activity:7407468096439037952,"My most popular video on YT isn't controversial at all.1140 1141I just said what most developers already know:1142 1143AI tools are great.1144 1145They also:1146 1147• Often slow us down1148• Create unrealistic expectations around productivity1149• Allow non-coders to create a mess at breakneck speed1150• Aren't the great replacement that the media keeps shoving down our throat1151 1152Look - I pay for the tools. I use them literally every day.1153 1154I've spent a solid portion of this year sharing my experience with Claude, Cursor and Gippity and where they fall short.1155 1156Looks like I'm not alone.",ARTICLE,Brian,Jenney,4149,0,40,24,2,0,0.015907447577729574,,2025-12-18 09:13:38,https://www.linkedin.com/feed/update/urn:li:activity:74074680964390379521157urn:li:activity:7407102288575647744,"Web devs: here’s a practical path to learn AI without quitting your day job:1158 1159What you’ll be able to build in ~6–8 weeks (1–2 hrs/night): a docs Q&A bot for your repo, a customer-support search tool, or a summarizer that plugs into your app. 1160 1161Real projects, not demo toys.1162 1163𝗧𝗵𝗲 𝗽𝗮𝘁𝗵:1164• 𝗚𝗲𝘁 𝘁𝗵𝗲 𝗺𝗲𝗻𝘁𝗮𝗹 𝗺𝗼𝗱𝗲𝗹 𝗿𝗶𝗴𝗵𝘁: tokens, embeddings, vectors, context window, latency/cost. You don’t need PhD math - just enough to reason about tradeoffs.1165• 𝗦𝗵𝗶𝗽 𝗮 𝘁𝗶𝗻𝘆 𝘀𝘁𝗮𝗿𝘁𝗲𝗿: Next.js + Vercel AI SDK (or OpenAI/Anthropic SDK). One route, one prompt, stream the response. Push to prod day 1.1166• 𝗔𝗱𝗱 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 (𝗥𝗔𝗚): chunk your own docs/README/issues, embed them, store in pgvector/Supabase or Pinecone, then ""ground"" answers with your data.1167• 𝗠𝗼𝗻𝗶𝘁𝗼𝗿 𝘂𝘀𝗮𝗴𝗲/𝗰𝗼𝘀𝘁𝘀: log tokens and latency per request. Set monthly caps.1168• 𝗨𝗽𝗴𝗿𝗮𝗱𝗲 𝘁𝗵𝗲 𝗨𝗫: add tool/function calling for structured outputs, retries with fallbacks, and guardrails for edge cases.1169• 𝗦𝗵𝗶𝗽 𝗮 𝗿𝗲𝗮𝗹 𝘂𝘀𝗲 𝗰𝗮𝘀𝗲: deploy, add analytics, and get someone (even just you at work) using it for a week. Iterate on what breaks.1170 1171𝗪𝗵𝗮𝘁 𝘁𝗼 𝘀𝗸𝗶𝗽 (𝗳𝗼𝗿 𝗻𝗼𝘄):1172• Building a ChatGPT clone1173• Rewriting your stack around the “framework of the week”1174• Deep-diving transformer math before you’ve shipped anything1175• Chasing the newest model every 48 hours1176 1177𝗔 𝗿𝗲𝗮𝗹𝗶𝘀𝘁𝗶𝗰 𝘁𝗶𝗺𝗲𝗹𝗶𝗻𝗲:1178• Week 1: Starter app + streaming + basic prompt1179• Weeks 2–3: RAG with your data + evals1180• Weeks 4–5: Monitoring, cost controls, tool calls1181• Weeks 6–8: Ship a real workflow and iterate1182 1183𝗪𝗮𝗻𝘁 𝗮 𝗻𝗼‑𝗕𝗦 𝘀𝘁𝗮𝗿𝘁𝗶𝗻𝗴 𝗽𝗼𝗶𝗻𝘁?1184• AI with JavaScript (hands-on guide): https://lnkd.in/gPWqvEff1185• AI with RAG (project guide): https://lnkd.in/d6Pzfcg61186 1187If you can build a CRUD app, you can build with AI. The trick is focusing on data and shipping small (but useful) tools.",TEXT,Brian,Jenney,3166,0,23,11,0,0,0.010739102969046115,,2025-12-17 09:00:02,https://www.linkedin.com/feed/update/urn:li:activity:74071022885756477441188urn:li:activity:7406740305170948096,"AI wrote this.1189 1190I’m testing a new app I built that can draft posts in my voice, suggest relevant lead magnets, and pull in examples from past content. It uses tool-calling, RAG, and a vector index of 1,000+ of my posts to get close to how I actually write.1191 1192Why share this? Because “AI for content” doesn’t have to be fluff. Here’s a practical example and how I got it working.1193 1194𝗪𝗵𝗮𝘁 𝗶𝘁 𝗱𝗼𝗲𝘀:1195 1196• 𝗜𝗺𝗶𝘁𝗮𝘁𝗲𝘀 𝗺𝘆 𝘃𝗼𝗶𝗰𝗲: It retrieves similar posts from my archive and mirrors tone, length, and structure.1197• 𝗞𝗻𝗼𝘄𝘀 𝘄𝗵𝗮𝘁 𝘁𝗼 𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲: It can suggest relevant resources (like an AI lead magnet) and drop the right links.1198• 𝗦𝘁𝗮𝘆𝘀 𝗼𝗻-𝗺𝗲𝘀𝘀𝗮𝗴𝗲: It pulls my business context (programs, voice, audience) before it writes, so it’s not guessing.1199 1200𝗛𝗼𝘄 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀 (𝗵𝗶𝗴𝗵 𝗹𝗲𝘃𝗲𝗹):

Showing the first 1,200 of 16680 lines. Download the file for the rest.