AI Growth/GTM guy who understands how to turn AI into revenue for companies, creators, and investors. VP Growth @activeloop @forgenewco prev @airbnb @tubi @lyft
Creators reach out here →
Joined September 2009
- Tweets4.5K
- Following259
- Followers940
- Likes2.1K
Michael Sutyak retweeted
From an empty rooftop to fireworks over the skyline, directed while it played.
This is Director in OTOY Studio. The shot streams live and you steer it with cues:
→ Neon signs flicker on
→ Freeze time, orbit left
→ Rain gets heavier
→ A drone rises
→ Fireworks over the city
→ Push in on the brightest one
Six cues, one live session, nothing re-generated.
Try Director in OTOY Studio.
Michael Sutyak retweeted
Every model. One canvas. OTOY Studio is the VFX platform for filmmakers.
Blending the power of Hollywood rendering tools with the speed and ease of AI workflows.
Michael Sutyak retweeted
Hivemind is now on WhatsApp.
Ask it what your agents are doing:
“Can you check the status of my agents?”
“What did the team work on last week?”
• See agent activity and outputs
• Switch between organizations
• Get team recaps
All from one WhatsApp chat.
Michael Sutyak retweeted
@activeloop is heading to the @agisummitai this weekend at the Palace of Fine Arts!
Stop by our booth to meet the team, learn what we’re building, and talk about the future of AI infrastructure and continual learning.
And don’t miss Activeloop Founder & CEO @DBuniatyan's keynote on Sunday:
🎤 Closing the Continual Learning Loop
📍 AGI Stage
🗓️ Sunday, 11:25 to 11:45 AM
#AGISummit #ArtificialIntelligence #AGI #ContinualLearning #Activeloop
Michael Sutyak retweeted
We have 3 new updates to Hivemind that are in service of our mission: making organizational agents more cost efficient and effective.
1. Proactive search experience for Claude Code + Cursor. After each user prompt, Hivemind automatically searches for relevant stored traces from the past and passes them to the agent as additional context. This improves reasoning and output.
2. Ability to share skills across teams and devices. Skill sharing is a simpler feature: it syncs the skills a user already has in their system through Deeplake, so everyone on the team can access them. Previously, we were more focused on generating and improving skills. This is about sharing existing skills across the team.
3. Ability to create and assign goals. Any teammate can create and assign goals to another that can persist across sessions until marked completed. Agents can automatically reason when the goal has been reached, and will notify the creator of the goal automatically.
All of these new features work to level up your team, reduce errors and redundant work and lower AI token spend.
Get set up with one command line install.
Michael Sutyak retweeted
Really excited to see that the project I’m leading, which we officially released four days ago, just got its first 1K GitHub stars, with 300+ gained in the last 24 hours.
It also reached #2 Repository of the Day on GitHub.
Really proud to be building this with @KamoAghbalyan, @khustup, @DBuniatyan, and the rest of the team.
Agree with the diagnosis. From the operator side, here's what happens when teams skip the work this post describes.
Monte Carlo surveyed 260 enterprise teams running agents in production (April 28, 2026):
64% deployed before they were ready.
75% of the engineers running them say the same.
63% found an agent touching data it shouldn't.
36% can't roll back a failing agent in minutes.
70% already know they're going to rebuild.
Gartner placed agentic AI at the Peak of Inflated Expectations and forecasts 40%+ of agentic projects get canceled by 2027.
None of these failures are the model. They're the runtime. If every step of a 12-step agent is 95% accurate, only 54% of full runs finish clean. At 99% per step, only 89%. Demos work because they're short. Production doesn't because it isn't.
What the post calls "writing down what good means" is what an enterprise engineering team calls a per-firm eval harness. Same thing. Trace-level observability, persistent memory across runs, automated evals on every model swap.
70% will rebuild. The only real question is whether they rebuild toward the runtime that compounds (their evals, their traces, their definition of resolved) or toward another wrapper that gets eaten when the next model drops.
Hivemind, the plugin that turns you agent traces into re-usable skills, just got smarter.
Now with SkillOpt: your skills don't just accumulate, they get sharpened over time.
Coding agents that actually get better the more your team uses them.
Introducing Hivemind: continual learning for AI coding agents.
Hivemind turns the traces from every agent your team runs (Claude Code, Codex, Cursor, Hermes, OpenClaw, Pi) into reusable skills, then pushes those skills across all of them. All on your cloud storage.
Now with SkillOpt built in, your skills get trained: +19.1 points of accuracy in Claude Code, +24.8 in Codex, best or tied on all 52 setups tested.
Open source, one line install.
Michael Sutyak retweeted
Coding agents that actually get better the more your team uses them.
Introducing Hivemind: continual learning for AI coding agents.
Hivemind turns the traces from every agent your team runs (Claude Code, Codex, Cursor, Hermes, OpenClaw, Pi) into reusable skills, then pushes those skills across all of them. All on your cloud storage.
Now with SkillOpt built in, your skills get trained: +19.1 points of accuracy in Claude Code, +24.8 in Codex, best or tied on all 52 setups tested.
Open source, one line install.
Michael Sutyak retweeted
Physical AI still requires manual annotation despite all advancements in visual understanding.
@activeloop collaborated with @AgentField_ai. outcome is Roboscribe-AF to automatically annotate multi sensory data with an agentic swarm.
how it works
> AgentField runs the reasoning graph to produce new annotation rows
> Deeplake stores the corpus and annotation versions
> raw and derived annotation fields share one schema
> disagreements between the visual and action reasoners become queryable dataset fields
> all open source
Read guest blogpost below.
Agentic LLM systems this year have mostly gone to coding, browsing, research.
With the @activeloop (Deeplake) team, we pointed agents at a job they don't usually do: curating a robotics dataset.
Roboscribe-AF automates the work a human curator would do — segment episodes, cross-check video vs trajectory, flag the messy ones.
Disagreements between reasoners get written back as queryable Deeplake fields.
Open source: deeplake.ai/blog/agentfield
Michael Sutyak retweeted
Hivemind just crossed 250 stars on github
2K weekly downloads on NPM. 🚀
Connect coding agents to a shared brain
> Collect traces into deeplake
> Auto-optimize skills
> Share across agents, machines and teammates
Your agents continuously learn from each other's experience.
Get them to compound your intelligence.
Hivemind just crossed 100 stars on github 🌟
github.com/activeloopai/hive…
Beyond memory. Let's compound intelligence!
This really is a game changer for anyone who uses coding agents.
Learning from your agent's traces and codifying that knowledge into skills makes your agents smarter across your org.
We got tired of re-explaining context, duplicate/inconsistent work and miscommunication across our agents.
So we built Hivemind.
today, we're going beyond memory.
your org's agents shouldn't just remember what happened. they should learn from their experience.
Hivemind takes agent traces and codifies them into skills every agent on your team can use.
> no more explanations
> no more duplicate work
> no more repeat bugs
we make your intelligence compound.
here's how 👇
Michael Sutyak retweeted
today, we're going beyond memory.
your org's agents shouldn't just remember what happened. they should learn from their experience.
Hivemind takes agent traces and codifies them into skills every agent on your team can use.
> no more explanations
> no more duplicate work
> no more repeat bugs
we make your intelligence compound.
here's how 👇
Michael Sutyak retweeted
"What would the developer experience even be like for someone using a continual learning model? Any idea how you’d actually steer it?" @garrytan asked, @demishassabis answered.
"Not having continual learning right now is one of the things holding agents back from doing full end-to-end tasks.
They’re really useful for aspects of tasks today. You can compose them together and do some really cool things, but they don’t adapt well to the specific context you’re operating in.
That’s the missing piece for them to become truly fire-and-forget agents that can just figure it out themselves.
They need to be able to learn about the specific context you put them in. So I think we have to crack continual learning to reach full general intelligence."
Michael Sutyak retweeted
Bigger context helps, but too much old context can still hurt performance (“context rot”)
Claude Code works better when you actively manage context.
Simple rules.
Use the same session for closely related work.
Start a new session for new tasks.
Use /rewind instead of piling on corrections.
Use /compact when you want a summarized carryover.
Use /usage to reason about how you’re using Claude Code
Delegate to subagents when you only need the result, not all the intermediate work.
Ever get tired of tracking your team down on Slack to get updates?
Dread putting together updates for your execs?
Miscommunication rampant amongst your team and 20 red notifications terrifying you on Monday?
Try the Hivemind plugin for ClaudeCode, Codex or OpenClaw and let your agents do the heavy lifting.
Engineering managers shouldn’t have to play detective every morning.
But too often, that’s the job:
standups get skipped
tickets stay stale
and status updates become a scavenger hunt.
What if your tools just told you what changed?
An AI layer across Claude Code, Codex, and OpenClaw could surface progress, blockers, and momentum automatically.
No nagging required.
You get your mornings back.
Engineers feel less monitored.
Everyone stays in sync.
Michael Sutyak retweeted
Engineering managers shouldn’t have to play detective every morning.
But too often, that’s the job:
standups get skipped
tickets stay stale
and status updates become a scavenger hunt.
What if your tools just told you what changed?
An AI layer across Claude Code, Codex, and OpenClaw could surface progress, blockers, and momentum automatically.
No nagging required.
You get your mornings back.
Engineers feel less monitored.
Everyone stays in sync.