DC

Co-Founder/CTO @mindcloudhq ($3.5M ARR). Working on Stealth.

Context is everything.

[DC @devdcdev]

Integration platforms are essentially normalization layers for the private internet.

I predict most future agentic services will depend on them as their router.

[DC @devdcdev]

@alvishbaldha @mindcloudhq Super happy with this 🤌

[DC @devdcdev]

@sama We automated what would have been about 50 man-years of work, in 8 weeks with an 11 man team.

[DC @devdcdev]

@sama I did something really cool in my company would love to share

[DC @devdcdev]

@jjferman Sage

[DC @devdcdev]

@mvrkke @jamesm @mindcloudhq @alvishbaldha Oh we know it!

[DC @devdcdev]

@alvishbaldha

[DC @devdcdev]

@alvishbaldha Nice desk!

[DC @devdcdev]

@theirongolddev @mattpocockuk Yeah. Most people call these prds but I had more success with canon.

[DC @devdcdev]

@joshpuckett We should train all future models and this and mootools.

[DC @devdcdev]

Oh so busy.

[MindCloud @mindcloudhq]

We've been busy — this week we built 368 new apps, exposing 8,280 actions agents can use.

These apps are accessible to any agent platform, letting their users connect to the tools they already use. They're also for Cirra — coming soon.

[Alvish 🧙‍♂️ @alvishbaldha]

✨ Charting my way through it.

[DC @devdcdev]

@jamesm Require user login + twitter and tie public names to each drawing :). Guests should sign their name after all

[DC @devdcdev]

@coreyching @OpenAIDevs @SlackHQ @figma @NotionHQ @gmail Hey do you mind PMing me on this?

[DC @devdcdev]

@TTrimoreau @garrytan Someone that is effective.

[DC @devdcdev]

@jamesm @wireframe 👏

[DC @devdcdev]

I’m not so sure. Your one example of reasoning is the one thing baked into models now.

I’ve been building harnesses for a month. Client/server side is determined by access.

Web search: server
Universal MCP: server
Knowledge: server
Compaction: server

Local file system: client by necessity
Local exec command: client
Exec tools (python, git): client or server based on containerization

None of those can or should be in models except for maybe compaction.

[DC @devdcdev]

Agents running on a schedule seem to perform worse if they have the knowledge they were run on a schedule. There seems to be more eagerness to complete, and therefore do a less thorough job.

[DC @devdcdev]

A process is ready to automate when it becomes boring.

By then, you understand it well enough to build the right abstraction.

[DC @devdcdev]

@theo I yo-yo between Codex and Opus like two forbidden lovers.