Suggestions
Davit Buniatyan
Building Deep Lake, the Database for AI
Davit Buniatyan is the founding CEO of Activeloopai, an expert in AI/ML, Data, and Infrastructure with a strong background in academia and industry. He holds a Ph.D. in Computer Science from Princeton University and a Bachelor's degree in Computer Science from UCL. Davit was a Y Combinator S18 participant and is currently focused on revolutionizing data as part of his work on Data 2.0.
Having founded Activeloop, Davit Buniatyan has a deep understanding of cutting-edge technologies and a passion for innovation. He brings a unique blend of academic research and entrepreneurial spirit to his endeavors, making significant contributions to the fields of AI, ML, and data management. With a strong educational foundation and practical experience, Davit continues to drive advancements in technology and shape the future of data processing and analysis.
Highlights
still wild how many teams keep production data on a filesystems, loose it and then blame the vendor.
Databases exist for a reason consistency, safety, versioning. You can even make agent-driven ops reliable if you use the right primitives.
This isn’t about Cursor, Railway, or whatever the vibe coding trend is. It’s just fundamentals.
When you bypass those fundamentals to get to the business fast, sometimes you pay the price for those shortcuts.
Jack's right: "Companies move fast or slow based on information flow." But framing it as a worker hierarchy problem is losing the plot.
Look at where the actual work is moving: coding agents.
Quick question for any dev reading this:
Do you have a coding agent? Probably yes. Do you store its traces? Almost certainly no.
That's the whole problem. Every run your agent does (the reasoning, the tool calls, the dead ends, the fix that finally worked) evaporates the second the session closes. Tomorrow it starts from zero. Your teammate's agent starts from zero. Every agent on your team is a goldfish with a GitHub token.
This is not a RAG-over-Slack problem. Indexing your docs and Notion and email doesn't fix it, because the thing that actually needs to be shared isn't sitting in those tools. It's sitting in the trace that just got thrown away.
Deeplake Hivemind stores agent traces and makes them shared memory. One install and every coding agent on your team, Claude Code, OpenClaw, Codex, whatever, can read what every other agent has already tried, decided, and shipped. Across sessions. Across teammates. Across tools.
Your agent hits a bug your coworker's agent solved last Tuesday? It already knows the fix. Your agent is about to refactor a module someone else's agent just rewrote? It already knows. No standup, no Slack thread, no "hey did you push that."
This is not another RAG integration. It's the missing substrate underneath all of them: the place agents actually remember.
Slack was for humans. Hivemind is for the things actually writing the code now.
Comment HIVEMIND and we'll DM you $100 in free credits. Plug it into your coding agent and run the experiment with your team.
