Ty Dunn @ty.energy · 09/09/2026And join us on Friday for the Superdark Factory lecture at @grayarea.org: luma.com/fawtwq0bluma.comAgentworld x Superdark Factory Antikythera x Disintegrator · LumaA double-lecture by Benjamin Bratton (Antikythera) and Marek Poliks (Disintegrator) on how the coming world of trillions of agents and fully automated agentic… 001
Ty Dunn @ty.energy · 09/09/2026Read the entire paper in the Antikythera Journal: superdark.antikythera.orgsuperdark.antikythera.orgThe Superdark Factory | Antikythera 110
Ty Dunn @ty.energy · 09/09/2026How we build software is about to be more like architecting a self-guided spacecraft Once we turn on a Superdark Factory, we let go of old ways of control and inherit new ways of steering We become strategic players who co-govern continuously evolving software with agent swarms 110
Ty Dunn @ty.energy · 09/09/2026Everyone has been asking me what I’m up to after the @continue.dev acquisition The answer: I wrote a paper with Disintegrator titled *The Superdark Factory: Toward the Full Automation of Software* that was published in Antikythera (MIT Press) today superdark.aisuperdark.aiThe Superdark FactoryToward the Full Automation of Software 340
Ty Dunn @ty.energy · 26/02/202630 tickets of product debt. 8 waves of parallel Claude Code sessions. Between waves, a trip to the post office. By the end: +283 net lines and a map of every moment that actually required a human. 100
Ty Dunn @ty.energy · 26/02/2026Yesterday I shipped 29 PRs, renewed my passport, and had two conversations that landed on opposite ends of the same question. blog.continue.dev/283-net-line...blog.continue.dev283 Net Lines and a PassportI shipped 29 PRs, renewed my passport, and had two conversations that landed on opposite ends of the same question. 120
Ty Dunn @ty.energy · 18/02/2026PR merged: github.com/continuedev/... ✅ appreciate the kind words!github.comIncrease font size on mobile by TyDunn · Pull Request #99 · continuedev/amplified.devSummary Bumps base html font-size from 16px to 17px at the 768px tablet breakpoint Increases body text (.section p, .closing p) from 1.05rem to 1.1rem on tablet, and from 1rem to 1.05rem on mobile... 010
Ty Dunn @ty.energy · 17/02/202611 checks for one HTML file. Each protects a decision easy to violate when agents open PRs. Writing them forced us to say what "good" means. Tests for your taste. See the checks: github.com/continuedev/amplified.dev/tree/main/.continue/checks Try it: continue.dev/walkthrough 010
Ty Dunn @ty.energy · 17/02/2026"Resist Complexity": "The site is HTML, CSS, and images. That's the point. Don't add what you don't need." No npm. No Tailwind. No React. CSS in a style tag. Not a problem to solve — a decision. The check enforces it every PR, especially from agents that love adding deps. 110
Ty Dunn @ty.energy · 17/02/2026The site has two hidden features. Type "amplify" and the text lifts, lavender glows: "↑ the work moves up." There's also j/k navigation — zero UI. If you know, you know. Each has its own check. Both share one rule: zero visual footprint. 110
Ty Dunn @ty.energy · 17/02/2026"Accessibility as Care" — "The site should work for everyone — not just developers on MacBook Pros, but the people you love who don't have a framework for what's happening." Reduced motion. Descriptive alt text. WCAG contrast. Each rule is a small act of noticing someone exists. 110
Ty Dunn @ty.energy · 17/02/2026"No Product Creep" watches for the site drifting from essay toward marketing. Analytics scripts, newsletter signups, CTAs to commercial pages. The site's power comes from what it is not. The moment it starts selling, it stops being the thing people share. 110
Ty Dunn @ty.energy · 17/02/2026The essay has a six-section narrative arc. A check called "Editorial Voice & Narrative Arc" protects it. Flags corporate language. Enforces second-person voice — peer to peer, not product to customer. Pull quotes must be turning points, not summaries. No linter can do this. 110
Ty Dunn @ty.energy · 17/02/2026amplified.dev is a single-page essay arguing developers are amplified, not automated. One HTML file, inline CSS, vanilla JS. No frameworks. We wrote 11 checks for it. Checks don't have to protect code — they can protect taste. Here's what ours look like.amplified.devDevelopers amplified, not automatedEvery time a layer of the work became infrastructure, the role didn't shrink. AI is the next layer in it. 282
Ty Dunn @ty.energy · 08/10/2025Read more on the Continue Blog: blog.continue.dev/from-vibe-en...blog.continue.devFrom Vibe Engineering to Continuous AIVibe coding emerged earlier this year to describe fast, loose experimentation with AI. Now, Simon Willison has proposed "vibe engineering": Vibe coding is irresponsibly building software through dic... 000
Ty Dunn @ty.energy · 08/10/2025Continuous AI isn't about removing developers—it's about working at the right abstraction level. Sometimes that's hands-on vibe coding Sometimes careful vibe engineering Increasingly, it's orchestrating remote agents while staying accountable for what ships 110
Ty Dunn @ty.energy · 08/10/2025The shift happens when you trust an agent to run 5+ minutes unsupervised. You get bored waiting, so you spin up a second agent. Then three. Four. Five. Suddenly your whole team is doing this and you need: secure execution, event-driven triggers, overnight runs, monitoring, etc. 100
Ty Dunn @ty.energy · 08/10/2025We are evolving from vibe coding → vibe engineering → Continuous AI - Vibe coding = fast AI experimentation - Vibe engineering = sophisticated AI-assisted production development as individuals - Continuous AI = orchestrating agents across your entire dev lifecycle as a team 120
Reposted by Ty DunnAlan Nichol @alanmnichol.bsky.social · 04/09/2025So 🔥 I love when startups do research and just ship in product. Skip the benchmarking theatre. 052
Reposted by Ty DunnNate Sesti @natesesti.bsky.social · 04/09/2025Introducing Instinct 🐺, a new state-of-the-art open Next Edit model trained by @continue.dev Runnable with @ollamabot.bsky.social and available today in VS Code 1164
Ty Dunn @ty.energy · 03/07/2025blog.continue.dev/context-engi...blog.continue.devContext Engineering: The Bottleneck to Better Continuous AIBack in 2019, I gave a presentation about building contextual AI assistants at the O'Reilly AI Conference. While AI has progressed dramatically since then, one challenge has remained constant: the art... 010
Ty Dunn @ty.energy · 03/07/2025There is so much room for context engineering improvements at large organizations 140
Ty Dunn @ty.energy · 14/05/2025@continue.dev is hosting the AI Native Dev - SF Meetup at GitHub HQ (thanks to Kevin Crosby!) with @jedi.be and friends ️🔥 020
Ty Dunn @ty.energy · 29/04/2025It's hard to tell from your screenshot, but it looks like it didn't use your MCP server. You need to be in Agent mode, and the MCP needs to be connected, which you can verify in the notch. Then you should see the tool from the MCP being used 👍 000
Ty Dunn @ty.energy · 28/03/2025We launched on Hacker News yesterday! There was a great discussion on custom AI code assistants. In case you, missed it: 151
Ty Dunn @ty.energy · 26/02/2025It’s critical that an open architecture underlies the future of AI-enhanced software development. This is why we are launching Continue 1.0 with hub.continue.dev today: techcrunch.com/2025/02/26/c...techcrunch.comContinue wants to help developers create and share custom AI coding assistants | TechCrunchContinue helps developers create customized, contextual coding assistants that can connect with any model and development environment. 042
Ty Dunn @ty.energy · 30/11/2024Lineages of information structuring matter have led me to this Recursive Worlds lecture by Sara Imari Walker (and Bluesky, hello world!) antikythera.org/films/recurs...antikythera.orgRecursive Worlds | AntikytheraRecursive Worlds is a lecture by Sara Imari Walker recorded in July 2024 in London as part of Antikythera’s Cognitive Infrastructures Studio lecture series. 030
