Patrity

Why am I building cognova.dev? Because it’s a tool that I actually needed. There’s a gap that needs filling. Agentic orchestration, custom tools, task management, and documentation management. Designed for devs, build by a dev. #BuildInPublic #AITools #DevTools

✅Telegram integration ✅Discord integration Cognova allows you to interact with your agents through the web or your choice of integration! Early access soon👉cognova.dev #BuildInPublic #AITools #DevTools #Agents #AIAgents

@johncooper19863 Yes sir! I work on the project-side managing earned value mostly, but also tracking spent hours, productivity, schedule data, etc. I'm essentially the technical bridge between IT and AWP. I completely agree it's underserved, happy to connect! My DMs are always open!

@Patrity wait you're in construction too?? epc data management is such a pain point. we're solving a similar problem but focused on boq extraction from tender pdfs for indian rcc projects. would love to exchange notes — the construction + ai intersection is so underserved. let's connect!

Real-time collaborative editing coming to Cognova. Two users. Same document. Live sync in both editor and preview. Markdown, code files, everything your team touches. Your AI agents and your team, sharing one workspace. Early access👉 cognova.dev #BuildInPublic #DevTools #AI

@TheJokerHut The chat is for LLMs! I haven't posted a demo yet but you can talk to a generalized agent or create specific expert agents that have context of docs from the knowledge base you build. We also provide MCP tooling that allows your code agents to call your custom agents/knowledge!

@Patrity Not that i can think of off the top of my head🤔 is the chat for LLMs or for users to communicate between eachother?

@TheJokerHut Thanks for your reply about the presentation on my app, wish I could take the credit, but I use @ScreenCharm It handles all the zooms/pans automatically and has a built in editor as well. It'd fit really nice on this demo😃

Working on guided programming projects for Ludocode. The idea is you get instructions → write real code → run it directly in the web editor. Will be shipping them in less than a week, stay tuned! #BuildInPublic #WebDev #WebDesign #EdTech

@TheJokerHut Thank you! Anything you'd specifically like to see in a tool like this? Trying to tie up everything pretty soon and release early access🙏

@Patrity Really nicely presented!

@johncooper19863 Yep! the context switching between users and agents kills productivity! I'm building this app mostly for me and I work in the construction industry managing data for EPC projects on enterprise/domain-specific systems. I'm glad that you see the vision!!

@Patrity ai agents + humans sharing the same workspace is the future. right now most industries still pass files back and forth like it's 2005. live sync changes everything — especially for teams working on complex docs like construction specs or legal contracts.

@SynabunAI Yep! The MCP/Skill API Supports memory and full document endpoints are being worked right now! cognova.dev/docs/rest-api Instruct your claude.md file to store memories and reference them, or better yet - use hooks to force it! Our vectorization and semantic search handles the rest!

task lists nail the "what to build next" problem. the gap that usually stays open is the "why" layer - decisions made, constraints hit, patterns that worked. when that carries over between sessions the agent doesn't need re-explaining on session 10 what was settled on session 2. both layers together are powerful

I gave my AI agent a PRD about Horse Tinder It read the doc, created a project, broke it into tasks, and completed one — across separate conversations. This is Cognova: project memory for your AI coding workflow. Plan here. Build in Claude Code, Codex, Cursor — whatever via MCP/Skills Early access soon👇 cognova.dev #BuildInPublic #AI #DevTools #AIAgents

Your AI coding agent can check its own task list before writing a single line of code. Claude Code → Cognova MCP → knows exactly what to build next. Just update your Claude.MD/Agents.MD and you're ready to roll! This is the workflow💪

Also shoutout to @sergeynazarovx for building @ScreenCharm Absolutely amazing application. Best lifetime license I've ever purchased😅

@Adidotdev yooo, did you. steal my code?? why does it look just like the one claude code built for me??

claude code is fucking insane i know literally NOTHING about coding. ZERO. and i just built a fully functioning web app in minutes. http://localhost:3000/ check it out

Call me a conspiracy theorist, but prior to the government and Department of War beef with Anthropic, I NEVER saw this much downtime on @claudeai Claude Code's API's.... 🤔👀

@finalbossblues Would love to hear your ideas and possibly collaborate on something. I’ve built several web applications over the years and love the game development space.

I just got a kick in the nutz from a writing software that I really liked. When you download my art assets (free or paid), you get straight-up PNG files you can use forever. Imagine if all of the assets I'd released over the past 20 years were tied to a cloud account that you had to login to use, and I could randomly one day decide to shut it all down. Software as a service is a scam. REALLY I should have known better, gotta find real alternatives from real people on itch.io.

@AutismCapital @AnthropicAI Best comment ever 🤣

@AnthropicAI New Anthropic

We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.

@benjamincanac Massive improvement!

Thanks for the inspiration! 🔥 I added this to Nuxt UI as well when using unique ids: #deduplicated-toasts" target="_blank" rel="nofollow noopener">ui.nuxt.com/docs/component…

@superactro @Patarino @shawmakesmagic I’m running OpenClaw. Using opus as a man model but I utilize qwen-3 for most of my agentic tasks!

If you're into self-hosted AI, check out OpenClaw — it's an open-source AI agent that runs 24/7 on your own hardware. Works with cloud models (Claude, GPT) or local ones. We actually sell a pre-configured box running it on a Jetson Orin Nano for people who want plug-and-play. With your 3090 rig you could run it with local LLMs too.

How the fuck are you guys surviving on 2 max plans

@KyleHessling1 @TheAhmadOsman years ago? claude code is not even 1 yr old....

I started using Claude code years ago, and it’s made me a lot of money that I don’t otherwise deserve, and hasn’t really ever let me down, also max is 100/month vs 200 for codex and I never rate limit on Claude, also idk I’m scared of trusting Chinese models I can’t run locally for client codebases maybe. And the front end designs always seem to be better for my use cases with Apple swift programming. Thoughts?

what do people use Opus for nowadays? Kimi, GLM, and MiniMax are overall a better, cheaper, and faster models Codex is more intelligent as well why would anyone pay Anthropic for a Claude subscription that gets nerfed?

@Patarino @shawmakesmagic Awesome, well I'd love to throw some of my local compute at it! I have a 3x 3090 rig (hopefully 4 soon) that I'd be happy to test on.

@Patrity @shawmakesmagic We are training our own language model specifically for it!