Let's Build with AI

[Let's Build with AI @AvinashAeon]

@mattpocockuk I have Pockok Skills Github page as the permanently open tab. It's always a good place to dive in for quick reads.

Leitwörter - is my current Leitwört
Leading words have tremendous influence on both human neural psychology and agentic neural prediction.

Thanks Matt.

[Let's Build with AI @AvinashAeon]

@PrajwalTomar_ Mobile
I'm particularly interested in how you handle the edge cases.

[Let's Build with AI @AvinashAeon]

@NousResearch For my Hermes running on VM, and accessing behind Tailscale via Hermes dashboard on my local mac is great.
But my poor old Intel Mac doesn't have a release for the new Hermes Desktop app. That's a bummer.
If someone has a fix for Intel Macs, please do help me out

[Let's Build with AI @AvinashAeon]

@Teknium @NousResearch Can't hermin with the desktop app on poor old Intel Macs 😢

[Let's Build with AI @AvinashAeon]

@ParagArora Been waiting to give Pi and CommandCode a try. Have heard a lot about their token efficiency without a drop in performance.
I had deferred Pi, as it didn't have a built-in MCP bridge, but iterating a custom one is a good learning curve.
Will do it soon.

[Let's Build with AI @AvinashAeon]

@karpathy Congrats Andrej,
Even, your rough thoughts have been shaping the AI engineering landscape for quite a while from the backseat. Now you will lead from the front. Excited for how you will enlighten us along the way.
You're a solid role model in this hype world.
My role model 🫡

[Let's Build with AI @AvinashAeon]

@trq212 This will signal the model to take a bit more leeway to depart from the <SPEC>. That can be good where the plan had uncovered bases, while in tightly defined scopes, it might produce trackable yet unintended approaches.
I'll experiment with this for passing tasks between agents.

[Let's Build with AI @AvinashAeon]

@gregisenberg Great ideas to daydream the wide world of agentic possibilities. I need to do this more often to get out of my agentic infra related projects' tunnel vision. Thanks Greg. I'll keep coming back for more ideas.

[Let's Build with AI @AvinashAeon]

@budapp My OpenClaw is locked in a VM manging root & coding agents without outside access. An able outside world facing collaborator would be great. Was planning for Hermes to do that, but probably Bud can do a fine job here, with security already taken care of. Credits to experiment 🙏

[Let's Build with AI @AvinashAeon]

@rileybrown I'm using both. OpenClaw as the Operator of the internal shell sandbox and Hermes as the outbound Scout, communicating via contractual tasks and only sanitized structured data flows back in.
Both have a different feel, even when running on the same model.

[Let's Build with AI @AvinashAeon]

@shipper_now SHIPPER

[Shipper @shipper_now]

🚨 JUST IN: @shipper_now just ended vibe coding forever. From now on, Claude Opus 4.6 in Shipper will build and run a full business all by itself.

We just launched Shipper a package for Claude to:

→ Build web/mobile apps and Chrome extensions
→ Code, design, monetize, launch
→ Do email marketing for you
→ Translate the entire app instantly
→ Self-maintain in the long run

Claude Code Opus 4.6 can do all the above from a 7-word prompt for as low as $0.28/app... And it takes minutes!

Simply go to Shipper, then ask Claude to "create a talent-hiring platform" or "build an analytics SaaS that charges $29/mo"!

To celebrate the launch, we're giving away free credits randomly. Repost rand comment "SHIPPER" and we'll pick the winners.

[Let's Build with AI @AvinashAeon]

I did try to make something similar with Perplexity and Comet, which could only be semi automated via multiple tabs. But then I made the 'LLM Council' on AntiGravity as a runnable script with output schema and proof of work from models. Now it's co-opted by Perplexity.

[Perplexity @perplexity_ai]

Introducing Model Council in Perplexity.

Run three frontier models at once, compare outputs, and get a more accurate, higher‑confidence answer.

Available now on web only for Perplexity Max subscribers.

[Let's Build with AI @AvinashAeon]

@perplexity_ai A while back I did try to make something similar with Perplexity and Comet, which could only be semi automated via multiple tabs. But then I made the LLM Council on AntiGravity as a runnable script with output schema and proof of work from models. Now it's co-opted by Perplexity.

[Let's Build with AI @AvinashAeon]

@theo The AI brand wars are getting damn serious. It's great for all of us builders, but hope they don't burn down all of the sense of security, that we somewhat know what we are doing.

[Let's Build with AI @AvinashAeon]

No way! OpenAI and Anthropic are releasing ground shattering LLMs in the space of minutes. This is unbelievable 😳

[Sam Altman @sama]

I love building with this model; it feels like more of a step forward than the benchmarks suggest.

Also you can choose "pragmatic" or "friendly" for its personality; people have strong preferences one way or the other!

[Let's Build with AI @AvinashAeon]

@theo The AI brand wars are getting damn serious. It's great for all of us builders, but hope they don't burn down all of the sense of security, that we somewhat know what we are doing.

[Let's Build with AI @AvinashAeon]

Anthropic is back on top of the AI news cycle to rain all over OpenAI's 'Codex app launch' parade. Let's see what's now possible that wasn't before.

[Claude @claudeai]

Opus 4.6 is state-of-the-art on several evaluations including agentic coding, multi-discipline reasoning, knowledge work, and agentic search.

We're also shipping new features across Claude in Excel, Claude in PowerPoint, Claude Code, and our API to let Opus 4.6 do even more.

[Let's Build with AI @AvinashAeon]

@v_computer Anthropic's Claude Opus 4.6 is sure to lift up vibe (intent) coding as the only viable way to make apps. User’s intent is now sacred and is to be treated as the only meaningful state around which disposable apps are to be built and used on demand. Very eager to try it out 😁

[Let's Build with AI @AvinashAeon]

@matt_elevenlabs 11B