unfair.so/x/@johncrickett
John Crickett
John Crickett15.9K
John Crickett
This creator added themselves. Not verified by unfair.so.

Helping software engineers become better software engineers by building projects. With or without AI.

@johncrickettAI CodingSoftware EngineeringEngineering Leadership
15.9KFollowers
2.6KMedian impr. / post
1.4%Engagement
22+17
Paid partnerships
In their audience
10+ affiliated with @microsoft follow John Crickett
@brunoborges@LyalinDotCom@stroker@Richie4love@joeksluis@tweetmonster999+48+ from @ycombinator7+ affiliated with @google6+ affiliated with @jetbrains5+ affiliated with @amazon4+ affiliated with @meta
1,500+ verified followers analyzed
@brunoborges@LyalinDotCom@stroker@Richie4love@joeksluis@tweetmonster999+1494156+ engineers111+ AI engineers96+ devtools engineers71+ AI founders69+ founders38+ devtools founders
25–34 core age group
37%
25.5% of audience in United States
🇺🇸
Last 7 days185 posts analyzed · replies not counted · updates weekly
ImpressionsEngagements
332K166K0
5.1K2.5K0
Nov 13Nov 14Nov 15Nov 16Nov 17Nov 18Nov 19
Past collabs
John Crickett@johncrickett · Nov 2In August 2025, a developer called Will Burgess shut down his marketing agency to go all-in on building apps with AI. Less than a year later, he reported that he's making more money than he ever did running his agency. So what changed? He went all in on Base44. Base44 is an AI-powered app builder that lets anyone turn a simple description into a working application, complete with a frontend, backend, database, and hosting. While it’s powerful, it’s even more powerful in the hands of an experienced developer. Will realised that as more people started building apps on Base44, more people would also need help finishing them. So he started showing up in the community, answering questions, fixing problems, and helping people turn their ideas into working products. Over time, that turned into paid projects, consulting, debugging, repeat clients, and a full-time business. And Base44 makes it easy for other developers to follow a similar path. They have a Partner Program where software engineers can list their services and get discovered by businesses already building on the platform. Instead of spending months building and marketing your own SaaS, you can help businesses build theirs. You could specialise in building internal tools, integrating APIs, rescuing unfinished apps, or providing ongoing maintenance and support. Of course, finding clients and delivering reliable software still takes real work, but the partner program makes it easy to find motivated buyers, looking for the specific services you offer. With the tech job market as unpredictable as it is right now, this presents an interesting opportunity for developers to earn money on the side or build a full-time business could provide some much-needed financial security. If you’re a developer looking to do either, check out the @base44 partner program: https://app.base44.com/partners?program=trueBase44Paid partnership2.4K views
John Crickett@johncrickett · Oct 28Most AI conferences sell a vision of the future. I prefer conferences that focus on what actually works and how to build it. That’s the idea behind Forge ‘26, Fireworks’ flagship conference for specialized intelligence. On November 3rd, they’re bringing around 900 application and model builders, systems engineers, and technical leaders to Pier 27 in San Francisco. They'll be talking about what’s working, what’s broken, and why. Who’s on stage: A fireside chat between Jensen Huang (NVIDIA) and Lin Qiao (Fireworks). A fireside chat with David Heinemeier Hansson (DHH). Josh Weisberg (Zillow) on training AI that’s uniquely Zillow’s own. Sarah Sachs (Notion) on “Riding the Pareto Frontier”. And many more. You can expect candid conversations about the technical realities of production AI: Architecture decisions that worked, those that didn't, and the lessons learned. How teams are actually running evals for production systems. How to manage the trade-offs that impact quality, latency, and cost. The program spans frontier research, applied AI, hands-on workshops, a hackathon, and an invite-only executive track. It’s free to attend, but you need to apply for a place and attendance is intentionally selective. View the full speaker lineup and apply here: https://fandf.co/4yIg8tu Not in SF? The keynote will be livestreamed, and session recordings will follow. This post is sponsored by Fireworks. @FireworksAI_HQForge ‘26Paid partnership767 views
John Crickett@johncrickett · Oct 27Did you know that on teams using AI coding agents, code now waits 114% longer for review? And 76% more code is merged without human review, and for teams using agentic review 41.4% doesn’t even get that. If you’ve been working with agents a lot, that might not surprise you. Intuitively we’ve all seen that the code gets written faster now, but we’re not shipping to production as fast as we’re opening PRs. AI has accelerated the work at the start of the development process, we can now go from idea to code far quicker. In fact, the average AI-enabled development team is completing 69% more tasks. But at the same time only merging 16.6% more PRs, and deploying just 13.8% more frequently. So why aren’t many teams getting the promised productivity gains? Well, our intuition is reflected in the data from Faros’ new AI Speed Trap quarterly report. They analyzed data from 22,000 developers across 4,000 teams and found that it’s not just code review that’s a bottleneck. QA time has increased by 300% and despite that, monthly incidents are up 125%. So whilst AI has huge potential to accelerate software engineering, many teams are not seeing the full potential, and worse, many are seeing significantly more incidents. The data from Faros’s report shows that we can code quicker with AI, but we still need the proven software engineering processes to ensure that we improve end-to-end engineering productivity without compromising quality. If you’re running a software engineering team or organization, check out Faros’ report: https://fandf.co/4zdtmzt This post was sponsored by @Faros_AI.FandfPaid partnership4.3K views
John Crickett@johncrickett · Oct 12I've a lot of old coding projects I'd love to pick up again, but I haven't touched them in years, and I can't remember where I left off. So, rather than spending hours digging through the code to get back up to speed, I've been using AI to help. And since JetBrains Air in IDEs launched, I’ve been using it to manage the AI that is helping me figure out what I was working on, what works, and what needs fixing or finishing. Let me show you how. For this example, I'm revisiting my Go implementation of Lox from the excellent Crafting Interpreters book. I haven't touched it in a couple of years, and I'm not entirely sure what state I left it in. Step 1: Get the context back. I give an agent the entire project and ask it to explain what I was working on and where I left off. Step 2: Find out what works. I start other agents to explore the existing test scripts and run the tests to see what's working and what's broken. Step 3: Identify and fix problems. As the tests reveal issues I can use the IDE to address them or start another agent to do so. Step 4: Keep track of progress. I use Air to track which agents need my attention. What I like about this workflow isn't just that I can run multiple agents in parallel. It's that I can manage them from one place while staying hands-on with my code. I can inspect changes, edit code myself, run tests, debug issues, and review commits, all within a familiar IDE workflow. Even better, the agents can use the debugger themselves. And I'm not locked into one agent. I can choose between Codex, Claude, and Junie depending on what I need. Now I've got a much clearer picture of what's working, what's broken, and what I need to tackle next. If you've got an old project you'd like to revisit, you can try JetBrains Air yourself: https://www.jetbrains.com/air/ides/ @jetbrains @getsomeairJetBrains AirPaid partnership3.1K views
John Crickett@johncrickett · Sep 29Did you know you can route different AI requests to different models automatically? There’s a Kestra blueprint that does exactly that. It first classifies each request as coding, reasoning, search or general. Kestra then routes it to a different model depending on the category. So a coding question might go to Anthropic, a search query to Perplexity, a reasoning task to Gemini, and a general question to OpenAI. The idea is that you don’t need to use the same model for every workload. You can choose models based on things like capability, cost or latency, while keeping the routing logic in one workflow. This is just one example of what you can build with a Kestra blueprint. Blueprints are ready-to-use workflow templates with the tasks, triggers and integrations already configured. You can import one, adapt it to your environment, and use it as a starting point instead of building everything from scratch. And if you’re taking part in Kestra’s Hacktober, one of the tracks is to build and contribute a new blueprint. If there’s a platform engineering workflow you’ve automated internally, or one you wish somebody had already built, turn it into something other engineers can reuse. You could win: 🥇 13” MacBook Neo 🥈 Apple iPad 🥉 $150 Amazon gift card Submissions close 31 October. Hackathon details and registration: https://fandf.co/4yVsuiq Browse the existing blueprints for inspiration: https://kestra.io/blueprints @kestra_ioKestraPaid partnership746 views
John Crickett@johncrickett · Sep 24If you’re planning to contribute to open source this Hacktoberfest, Kestra has two ways to make your contribution count, and both ways allow you to win prizes and grab some free merch! The first way: Fix an issue in their open-source codebase. They’ve identified and scoped a number of issues across their core and plugin repositories. They’re tagged good-first-issue, so you can find one that interests you. The second way: Build a blueprint, a ready-to-use flow template that other developers can pick up and run. A good place to start is with something you’d find useful yourself. Scratch your own itch. And if you’re new to Kestra, you can use the next week to work through the Kestra Fundamentals course in the Kestra Academy before Hacktoberfest starts and earn a certificate: https://academy.kestra.io/course/kestra-fundamentals It’s a self-led course covering workflow orchestration, Kestra’s core concepts, plugins and blueprints, with hands-on examples throughout. Pass the quiz and you’ll earn a certificate you can add to your LinkedIn profile. And once you’ve finished Fundamentals, there are three newly launched courses you can explore depending on your interests: Data Pipelines with Kestra AI Workflows with Kestra Infrastructure Automation with Kestra Then, from October 1st, you can start putting what you learn into practice. And you don’t have to win one of the top prizes to get something for your contribution. Everyone who lands a merge will get Kestra merch and goodies. There are also prizes for each track: 🥇 1st place: 13” MacBook Neo 🥈 2nd place: Apple iPad 🥉 3rd place: $150 Amazon gift card Submissions close on 31 October and the winners will be announced in November. You can find all the details and register to take part here: https://kestra.io/hacktoberfest @kestra_ioKestraPaid partnership630 views
John Crickett@johncrickett · Sep 21This Hacktoberfest, you can fix a real open-source issue or build something other developers will actually use. And win a prize for doing so. I’m excited to be part of the Kestra Hacktoberfest Hackathon this year. There are two tracks you can enter. Track 1: Fix something real in the codebase. Kestra has identified and scoped a number of issues across its core and plugin repositories. They’re tagged good-first-issue, so you can find one that interests you. Once you find one, comment on it to let others know you’re working on it for Hacktoberfest, and start contributing. Track 2: Build a flow other people will reuse. Create a blueprint, a ready-to-use flow template that someone else can pick up and run. The best place to start is with something you’d find useful yourself - scratch your own itch! If you need inspiration, browse Kestra’s existing blueprints or check out the Code Review Agent from Coding Challenges: https://codingchallenges.substack.com/p/coding-challenge-131-automated-code There are prizes for each track: 🥇 1st place: 13” MacBook Neo 🥈 2nd place: Apple iPad 🥉 3rd place: $150 Amazon gift card Submissions close on 31 October and the winners will be announced in November. You can find all the details and register to take part here: https://fandf.co/4y7VQtG We’re also kicking things off with a launch livestream at the start of October - more details to follow. If you’re thinking about taking part, go register now. If you have any questions, hit reply. @kestra_ioKestraPaid partnership671 views
John Crickett@johncrickett · Sep 15In the age of AI, nobody knows what anything costs anymore. Per-token pricing is the reason, but perhaps there’s a better way? If you’re curious to find out, then Corey Quinn’s upcoming keynote “Nobody Knows What Anything Costs Anymore,” at the Honeycomb O11yDay London is worth checking out. According to the programme, Corey Quinn will be looking at AI costs as runtime behaviour and arguing that traces are increasingly important for understanding cost per outcome. The event looks like a great couple of days. Day 1 is focused on hands-on workshops. One that caught my attention is Martin Thwaites’s “AI Battleships: can you beat my battleships just by vibecoding?” It uses a battleships challenge to explore vibecoding, telemetry, model choice and cost management. Day 2 is the main conference, with several talks I’d be keen to hear. Sam Newman’s opening keynote asks whether more code is actually better, and whether humans still need to read code. Yana Chen is speaking on “You can’t vibe code uptime… or can you?” Given her experience supporting LLM infrastructure at Baseten, her perspective on reliability should be interesting. Rebecca Murphey’s “Your AI Problem is your 2016 Problems Dressed in a Trenchcoat” also stands out. A brilliant title and a key message engineering leaders need to hear: making code faster to produce doesn’t automatically make software faster to deliver. Constraints around review, environments, risk and deployment still need to be tackled. Interested? Save your spot here: https://fandf.co/3SivGVK @honeycombioHoneycomb O11yDay LondonPaid partnership633 views
Planning a launch?John Crickett is one of 200 curated B2B creators. Tell us what you are launching and we will put the shortlist together.
Built with unfair.so
1 picked$500
John Crickett (@johncrickett) — sponsored posts & rates · unfair.so