Sabitlenmiş Tweet
Oliver Bowman
653 posts
Oliver Bowman
@oliverbowman_
Building HerLyfe: Empowering women to run health experiments & discover what actually works! Previously @prolific @bookingcom
Manchester, England Katılım Şubat 2022
342 Takip Edilen156 Takipçiler
@sama Good sir, pray tell shall the next GPT model arrive with front end craft so SOTA that even the divs align themselves?
English
@daniel_mac8 I been experimenting with Symphony this week for a swift ios project and its been interesting, basically been lighting money on fire and the agent feedback loop is a bit rough but it works.
I am trying /goal right now so time will tell
English
@oliverbowman_ Symphony is great for more serious work and probably for teams (though I've only used it as an indie)
/goal is awesome when you want to do a quick experiment
English
GPT-5.5 represents a new class of model.
Can't wait for the METR time horizon results.
Esp in Codex CLI with /goal.
/goal + GPT-5.5 is insane.
English
I’ve officially cancelled both Claude Max plans and have 2 x Codex Max plans.
Codex - particular GPT5.5 is the best coding model.
English
@Dimillian What skills other than the build ios plugin are you using?
English
I like how clean a Swift codebase can be with the right Codex plugins and skills!
English
@Dimillian Very true, so many opportunities, if only I had the time.
English
Thanks to ChatGPT Image 2.0 I can finally "dream" almost any app I want. And to me, this is almost like an ideal state for modern retro gaming. A modern shell rendering an older game.
English
@vida0118 Soon. We think it’s maybe more of a deep not smart model in Amp mode terms. It’s good. We like it so far.
English
I’m not surprised that GitHub Copilot raised the effective prices of Opus and stopped allowing signups: github.blog/news-insights/…. We didn’t support Copilot subs in Amp because we knew this would happen and be a bad experience.
Quinn Slack@sqs
Codex sub maybe since they make the model. But the GitHub Copilot one: GitHub doesn’t lose money on things like this. They run a tight ship. If it ever gets too widely adopted to be meaningfully valuable, the terms will change and Amp users will be rugpulled. I would need to be convinced otherwise first.
English
@mattpocockuk I thought it was going in favour of domain-model
English
English
@oliverbowman_ @warpdotdev What kind of Swift support are you looking for? LSP?
English
You can now run any CLI agent with first-class support in Warp, including Claude Code, Codex, OpenCode and Gemini CLI.
• Vertical tabs
• Notifications when they need you
• Integrated code review
• Remote control from mobile
• Rich input editor
Download Warp for free today.
English
GPT-5.5 / 'Spud 🥔' will be released today.
But that's not all...
My sources say it's *more* than one model.
Excited for today.
Chubby♨️@kimmonismus
Quick update to GPT-5.5 / Spud: AI labs are shifting back toward making models smarter during pretraining rather than leaning on reasoning (test-time compute) to boost performance. OpenAI's Spud and Anthropic's Mythos both appear to reflect this trend, getting better answers with fewer tokens and less reliance on chain-of-thought reasoning. Spud is smarter "out of the box" from pretraining, it shouldn't need to burn through long chains of reasoning tokens to reach good answers. Fewer tokens means faster responses and lower cost per query. Even more excited for spud today.
English
Press this button on any Amp thread page on the web to see a new look for threads. Nothing huge, just a new look. Feedback appreciated.
English
@MilksandMatcha @cerebras I am building an app to help the underfunded area of women’s health!
English
Giving away 5 Windsurf Max ($200/month) plans
Each person will get 3 months of free Windsurf Max (highest tier). Try out SWE 1.6, Cognition's latest, fastest, and most intelligent model, powered by @cerebras.
Winners will be selected from comments in 48 hours, comment below why you want it.
Cognition@cognition
We’re releasing SWE-1.6, our best model in both intelligence & model UX. SWE-1.6 matches our Preview model on SWE-Bench Pro while dramatically improving on various behavioral axes. It’s available today in Windsurf in two modes: free tier (200 tok/s) and fast tier (950 tok/s).
English
Makes complete sense, for me I use a design-docs folder that captures my decisions which are basically my ADR's the skill for sure will be a big step up for most people that are not using such a structure. I think the skill is generic enough that people that are following a similar structure they shouldn't need to modify the skill to use it.
Another minor comment would be the ADR structure, could be slightly easier to parse for agent harnesses things like scan ability, progressive disclosure and enforcement (this could be a codex thing)
English
@oliverbowman_ Appreciate it! Yeah I wanted it to do lazy (not eager) DDD, where it only adds ADR's and CONTEXT.md docs when needed. Essentially it's just /grill-me with a thin layer on top.
English
My new skill lineup:
/domain-model - replaces /grill-me, integrates some DDD concepts and adds docs & ADR's during discussions
/to-prd - create a PRD
/to-issues - create issues with blocking
/github-triage - triage issues with a state machine-based labelling system
/tdd - do TDD where appropriate
Still more to flesh out, but this is feeling AWESOME
English
I am mid giving it a go, I used a modified grill me skill previously where i weaved in open ai's harness engineering practices with karpathy doc structure. I think the honing on on language in this skill will mean I have to do a lot less garbage collection as while drift before wasn't too much of an issue it still was a larger one than I would like. I will report back more when I use it more.
English
@oliverbowman_ Have you given it a go? Would love to know how you got on
English