We had to choose: build @trysecretsauce with templates or go fully agentic. Templates would have shipped faster. Here's why we didn't. Templates put a ceiling on what AI can do for your brand. They're fast to produce and consistent in output - but consistency comes from constraint. You can only get what the template was designed to give you. And the template was designed for everyone. If your brand doesn't fit, you adapt the brand. That's backwards. An agentic system has no ceiling. You get the best practices baked in, but you're not borrowing someone else's aesthetic. The output is shaped by your brand - your colours, your tone, your vibe - not by what the template allows. If you can describe what you want, you can make it. The tradeoff is that it's harder to build, harder to control, and harder to make consistent at scale. Those are real problems. They're also the problems worth solving. We didn't want to build the ceiling.

You can't have a decent conversation with it AND it will 100% whoop your ass.

TLDR: Instinct persistent memory is markdown files, git-tracked and interlinked, with the LLM maintaining the knowledge + cron job to maintain ~once a day. KW matching/grep for retrieval, no vector search or BM25. Simple and elegant. Very cool! Thanks @DhravyaShah !

Some will claim this is AI.

As a creative, I am incredibly relieved to hear that Joel Kuwahara did not, in fact, lose all artistic judgment, moral integrity, hard-earned skills, standards and his will to live when he started using a new piece of technology. Incredible news for all us creatives using AI, who were in great danger of absolute moral bankruptcy.

Distribution is king. Ship long enough and you’ve been “there” too. It doesn’t matter how smart, fun, or exciting the work is. If no one sees it, it doesn’t exist. And that’s only getting truer as the noise gets louder.

"He told me AI has yet to prove its worth and it's all just hype. I didn't try convince him otherwise." Same. Two years ago, I was still engaging with skeptics to at least consider what is exciting about the tech. I completely stopped. Life's too short.

The number of engineers in the replies thinking this interaction is fake is mind boggling...

🐱

in the very fast process of getting my whole Engineering team Jev pilled

I really like the direction Hermes is taking. Keep the core clean and lean. Move everything else to plugins, let the original "solution providers" maintain their own integrations. Users pick and chose what they need. And let the Hermes team focus on high impact development.

Le shade.

Some dream of "Her". Me, I just dream about my Clippy. And he runs on Hermes by @NousResearch, so I gave him a voice using GPT-Live-1 full duplex, and it's been incredible. Enables high level, cross sessions, orchestration of my work or deeper research and conversation through incredibly voice based conversations. Mad stuff.

This is so incredibly good!

Astra is mind-blowingly powerful, yet I suspect it will change surprisingly little about how most people work with AI. At least for a while. There's a massive gap between model capability and our own ability to put it to work. If your AI workflow is still “ask a question, read the answer, copy something useful, do the work yourself,” a better model gives you a better answer. That's valuable. But you're still doing most of the same work. Using AI every day isn't the same as learning to delegate to it. That applies to developers too. The biggest change most have not really been able to implement is handing over complete objectives: giving an agent the goal, the context, access and boundaries it needs to work, then knowing how to judge what comes back. Understanding when to split the work between agents, how to connect them to your tools, and how to preserve context becomes part of doing the job. I don't think everyone should need to become an expert in that machinery. The tools need to make it easier. But right now, knowing how to work this way changes what you can get out of a model like Astra. The models are moving faster than we're learning to put them to work. Even those of us working at the bleeding edge are still figuring this out.