Heavybit

Your agent should've been at DevGuild. Now it can catch up: Crucible, Heavybit's new MCP server, ships Agentic Engineering and Agent Infrastructure skills built from The Long Horizon Stack. Every talk is live for you too. 🛠️ crucible.heavybit.com 🎥 heavybit.com/devguild/long-h…

On episode 15 of Third Loop, the Progressive Delivery team unpacks what happens when generating software gets easier than knowing whether it’s any good. They explore AI evaluations, the difference between delegation and abdication, and why a cheaper model can create more work. Along the way, they ask who will develop tomorrow’s expertise if junior workers never get room to fail. Tune in! hubs.ly/Q04Dgkq80

On episode 46 of @opensourceready, Brian Douglas and John McBride sit down with JT Archie. When an agent skips a command or grabs the wrong branch, a better prompt may only get you so far. JT explains how Steps uses structured pipelines to give agents clearer inputs and boundaries, then joins the hosts to explore SQLite, tool permissions, and the limits of fully automated development. hubs.ly/Q04F3MHy0

On episode 61 of Generationship, @rachelchalmers speaks with Gabriela Moreira (@bugarela), co-founder and CEO of @quint_lang, about building trust in AI-generated code. Gabriela explains how executable specifications help engineers explore software behavior and why passing tests can offer false reassurance. They discuss making formal methods more accessible while preserving the judgment and curiosity that good engineering requires. hubs.ly/Q04F3C2D0

On episode 14 of Third Loop, the Progressive Delivery team sits down with Dr. Cat Hicks (@grimalkina) to ask what happens when teams generate code faster than they can understand it. From review bottlenecks to disappearing onboarding tasks, they examine the hidden costs of acceleration. Cat shares how deliberate learning, comprehension reviews, and psychological safety can help developers build expertise alongside their AI agents. hubs.ly/Q04C5XBh0

Never give an agent a 404. In this talk, @timothyjordan, longtime DevRel leader and creator of a14y, makes the case for returning a 200 OK and turning the URL into a search term, one of the product and docs changes behind what he calls agent-led growth. heavybit.com/library/video/a…

On episode 14 of Data Renegades, CL Kao (@clkao) speaks with Hannes Mühleisen, co-creator of @duckdb, about the shift from user experience to agent experience. They discuss why SQL remains useful for AI, how actionable error messages help agents recover, and what database interfaces should provide when machines choose the tools. Tune in! hubs.ly/Q04CfGWb0

The worst memory an agent can write is a perfect record of its last mistake. In this talk, @sarahwooders of @letta_ai shows why agents that log dated corrections end up with chaotic context, and why they should rewrite their own guidance instead. heavybit.com/library/video/p…

On episode 14 of Data Renegades, CL Kao (@clkao) joins Hannes Mühleisen, co-creator of @duckdb, to trace the project’s journey from an unconventional research gamble to widespread adoption. Hannes reflects on the emotional demands of founding a company, the separation of open source stewardship from commercial partnerships, and the relationship that preceded AWS’s acquisition of DuckLabs. Throughout the conversation, he returns to a guiding principle: solve users’ problems, even when it means changing your mind. hubs.ly/Q04CfGT_0

Writing code got easy. That doesn't make frameworks obsolete: if you don't use one, your AI has to write one. We spoke to RubyLLM creator @paolino about his project RubyLLM and why Ruby is actually the best language for building AI web apps. heavybit.com/library/article…

Attending #KubeCon North America? We've got your first stop for the night. Just a 5 minute walk from the Salt Palace, stop by for cold pints on us, pool tables, and a little friendly competition with Heavybit GP @josephruscio.  luma.com/rack-em-up

On episode 14 of Third Loop, the Progressive Delivery team speaks with Dr. Cat Hicks (@grimalkina) about why developers are more than brains in jars. They explore how AI amplifies coordination problems, why faster code generation can outpace understanding, and what teams lose when learning becomes an afterthought. Cat makes the case for collaboration and shared expertise as foundations for successful AI adoption. hubs.ly/Q04C5XDJ0

On episode 15 of Third Loop, the Progressive Delivery team discusses why AI makes human judgment more important than ever. Celebrating their book’s birthday, they revisit alignment, autonomy, abundance, and automation through the lens of AI agents. From evaluating models to recognizing good design, they explore what faster software creation still demands of people. hubs.ly/Q04DghD60

In this DevGuild panel, the founders of @exedev, @hexanalytics, @overmindlab, and Induction AI discuss what gets written to memory, what gets forgotten, and how completed runs and human corrections become new knowledge. 🎥 Watch the panel: heavybit.com/library/video/c… 🛠️ Want your agent to learn from it too? Lessons from DevGuild now power new Agent Infrastructure skills in Crucible, Heavybit's MCP server: crucible.heavybit.com/mcp

On episode 45 of @opensourceready, Brian Douglas (@bdougieyo) and John McBride (@johncodes) join @dotta to unpack Paperclip’s evolution from a provocative vision of zero-human companies to the practical challenges of managing agents at work. The conversation covers human oversight, model diversity, and the infrastructure needed to delegate safely. Dotta also shares the realities of maintaining a popular project amid thousands of community pull requests. hubs.ly/Q04BXHVG0

Agents can do anything you want but can't know what you value. Dotta B of Paperclip Labs argues that's why humans still set the goals. heavybit.com/library/video/g…

On episode 6 of Lab Notes, Amir Zohrenejad (@amirzo) speaks with James Evans (@profjamesevans) about a paradox of AI-powered science: faster answers may come with a narrower range of questions. They explore how diverse perspectives, productive disagreement, and new experimental capabilities could help AI expand scientific discovery. Evans also considers how research incentives shape what humans and machines choose to investigate. hubs.ly/Q04C1cPl0

There are five systems that still need to mature before most engineering teams can deploy long horizon agents. @amirzo opened DevGuild with a talk on memory, goal orchestration, durable execution, learning, and tools. heavybit.com/library/video/t…

Judgment and taste matter more as agents take on more of the building. In this DevGuild talk, @wesmckinn revisits The Mythical Man-Month for the age of agent factories. heavybit.com/library/video/t…

On episode 14 of The Third Loop, the Progressive Delivery team sits down with Dr. Cat Hicks (@grimalkina), psychologist and author of The Psychology of Software Teams. Cat unpacks the “Brains in Jars” mental model, the belief that engineers are interchangeable pure-cognition units, and explains why organizations that hold it struggle to use AI well. The conversation covers learning debt, overproduction pressure, and what teams should do with the time agents save. hubs.ly/Q04C5XMG0