geek, scribe, coffee snob, and wanna-be cyclist. Contributor to Apache Spark and Delta Lake maintainer. Developer Relations at Databricks (opinions r my own)
Joined May 2008
- Tweets 18,744
- Following 1,908
- Followers 2,905
- Likes 26,168
334 Photos and videos
dennylee retweeted
This is super exciting - I’ve been using Omnigent at Databricks for a while, and today we open-sourced it.
Omnigent is a meta-agent for orchestrating a swarm of agents.
Why do we need this? The best results no longer come from a single model running in a single harness.
I used to run the same task with Codex and Claude Code, then pick the better one. But the obvious thing is to let them collaborate, debate, and converge on something better. Omnigent makes this smooth.
The other feature I love is real-time collaboration. You can invite people into an Omnigent session to watch, steer, and send commands.
Multi-agent, multi-human collaboration is the future.
Omnigent was built by @matei_zaharia and a very lean team in just 6 weeks, working every day out of a Databricks war room, truly amazing.
Databricks AI really feels like a startup.
Really excited to open source a new project: Omnigent, a meta-harness for AI agents.
It lets you build multi-agent coding and custom agents, sitting above Claude Code, Codex, Pi, and agent SDKs to let you compose them. It also adds live collaboration and rich control policies.
22
43
372
40,376
dennylee retweeted
There's a lot more coming, e.g. meta-optimization with @gepa_ai and programmatic tool calling as in RLM and MemEx. Read more in our blog: databricks.com/blog/introduc… or get started at omnigent.ai/
1
8
90
24,107
dennylee retweeted
Omnigent focuses on three problems above the level of a single harness: composition, collaboration and control. databricks.com/blog/introduc…
For composition, it lets you create multi-agent teams with different harnesses or swap harness and model mid-session and mid-loop.
2
6
39
4,489
dennylee retweeted
Omnigent is based on the trend we saw with AI usage at Databricks and Neon: engineers were combining multiple agents into loops and workflows, and this was difficult above the harness layer. We add a uniform API above any harness that enables rich features on top.
1
19
91
7,494
dennylee retweeted
19h
Check out Omnigent, an open source harness that lets you use all the existing code harnesses (Claude Code, Codex, OpenCode, pi), collaborate and share sessions in many modalities (e.g. Slack/Teams, cli, webui), while having a fine grained security model that really tightens the control on what agents can do/not do.
github.com/omnigent-ai/omnig…
6
39
238
11,339
dennylee retweeted
Really excited to open source a new project: Omnigent, a meta-harness for AI agents.
It lets you build multi-agent coding and custom agents, sitting above Claude Code, Codex, Pi, and agent SDKs to let you compose them. It also adds live collaboration and rich control policies.
62
153
805
110,159
dennylee retweeted
Introducing 𝗢𝗺𝗻𝗶𝗴𝗲𝗻𝘁, a meta-harness to combine, control, and share your agents.
The best teams already mix models and harnesses and design loops that drive teams of agents. No single harness can keep up with that alone. So we built the layer above — we call it a 𝗺𝗲𝘁𝗮-𝗵𝗮𝗿𝗻𝗲𝘀𝘀.
Omnigent sits above the tools you already use, Claude Code, Codex, Pi, and your own agents, and gives them one shared layer:
• 𝗖𝗼𝗺𝗽𝗼𝘀𝗶𝘁𝗶𝗼𝗻: combine models, harnesses, and techniques without rewriting code, and switch between them with one-line changes
• 𝗖𝗼𝗻𝘁𝗿𝗼𝗹: stateful, data-centric policies and cost budgets enforced at the meta-harness layer, not via prompts — let agents run without watching them
• 𝗖𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻: share a live agent session via URL with full history, so teammates can review, comment, and steer in real time
Every session is reachable from a terminal, the web, a desktop app, and your phone.
We built Omnigent for our own use at Databricks and are open sourcing it under Apache 2.0. databricks.com/blog/introduc…
10
56
263
17,242
dennylee retweeted
When the Seahawks came back from 12 down with just over 2 mins left to beat the Packers in the NFC Championship #Seahawks
16
61
1,674
40,202
dennylee retweeted
Stay tuned!
Jun 6
I want some kind of LLM workflow tool.
• Ability to manage a set of input files (Markdown or similar), plus other general-purpose context.
• With real-time collaboration. (And maybe some concept of snapshots or VCS integration.)
• And the ability to create/manage a inference workflows and a stored set of prompts.
• Access to general-purpose coding agents (and not just chat models).
• Some concept of compiled outputs/inference results (which ideally can be shared externally).
Many projects have this feeling: "there is all this stuff, which I want to process/compute over in this iterated way, with some build artifacts being important/worth saving." GNU Autotools x Notion or something. Is anyone building this?
2
7
99
23,674
dennylee retweeted
This is a devastating interview.
Scott Pelley tells the NYT that Bari Weiss directly put a “thumb on the scale” for Trump over the killing of Renee Good.
Here’s his explanation of exactly what happened.
1,279
15,963
53,455
3,104,666
dennylee retweeted
Jun 6
There is something happening in Texas.
We are uniting Texans on one team to take back our state and take back our country.
We will end 30 years of one party rule in this state and elect a Senator who is going to serve the people, not billionaire mega-donors.
2,583
3,646
16,966
283,919
dennylee retweeted
Jun 6
95 days until Seahawks opening night....
Let's watch Rashid Shaheed take it 95 yards to the house to start the playoff game against the 49ers 😏
8
50
684
15,002
dennylee retweeted
There’s a ton of interest in custom model tuning as agents reach production and scale up. Here is how we made Databricks Knowledge Assistant 3x faster using our new Instructed Retriever model trained end-to-end to do parallel test-time compute. It’s rolling out to customers now!
Most agentic search systems get better by thinking longer: more tool calls, more reason-act loops, each step waiting on the last. Quality goes up, but so does latency.
Instructed-Retriever-1 takes a different route. Instead of scaling test-time compute sequentially, it scales it in parallel. One retrieval-specialized model fans the work out: it generates multiple query and filter formulations to widen recall, then reranks the merged evidence with a multi-pivot reranker to sharpen precision. Both stages run at once, so searching more broadly no longer means searching more slowly.
The result inside Knowledge Assistant: search time drops more than 3x and answer time 2x, with time to first token around two seconds, and no drop in quality (it matches Claude Sonnet 4.5 retrieval quality on KARLBench). For the people using it, that means far less waiting between question and answer, the freedom to ask more follow-ups, and more of the knowledge base actually surfaced. Rolling out to all customers now, with no reconfiguration.
Read how we did it: databricks.com/blog/3x-faste…
5
10
115
17,356
dennylee retweeted
Jun 5
🛠️ Building apps should be fast. Managing infrastructure shouldn’t slow developers down.
Watch @andrelandgraf from Databricks and Kevin Niparko from @cursor_ai walk through how to build and deploy a production-ready app on Databricks using built-in governance and serverless Postgres.
You’ll see how teams can:
✓ Prototype and iterate in real time
✓ Branch, scale, and manage transactional data
✓ Govern data securely with SSO and Auth
✓ Deploy apps where their data already lives
Watch the on-demand webinar ↓ databricks.com/resources/web…
7
36
3,637
dennylee retweeted
Jun 4
"We don't really need AI to get smarter. What we need is for it to have context."
Databricks co-founder and CEO @alighodsi joined @CarolineHydeTV and @EdLudlow at the Bloomberg Tech event in San Francisco to make case that the real breakthrough will come not from smarter models but from giving AI the data context it needs to perform. That is the problem Databricks Genie is built to solve.
Watch the full interview: youtube.com/watch?v=CipJDvsH…
1
9
49
3,690
dennylee retweeted
Jun 3
Introducing the lab sponsors for the Databricks Grounded Reasoning Cup at #DataAISummit 2026: @AnthropicAI, @OpenAI, and @GoogleDeepMind.
Each lab is partnering with leading academic teams to build agents that tackle grounded reasoning over complex government data using the latest models and tooling.
Meet the teams, see the agents they’re bringing to the competition, and join us as they push the boundaries of enterprise AI reasoning live on stage! 🏆
10
46
9,514
dennylee retweeted
Jun 2
We just launched Sites into Codex!
Software creation was always about more than writing code. Sites in Codex fundamentally gives the power of end-to-end software creation to every user, no matter their technical fluency.
These Sites are fully deployed to a URL, private to workspaces, come with authentication, can have static files, and can store dynamic data in databases.
It is in preview for business and enterprise teams and will be rolling out to all workspaces over the next day. Give it a try by typing @ Sites into Codex and ask it to build anything!
This project took a massive amount of effort across hundreds of people at OpenAI - proud that we were able to get this out and excited to see what you all build with it!
240
265
3,840
1,021,248
dennylee retweeted
May 27
❤️ @Lovable now integrates with Databricks, providing a natural language interface that allows anyone, regardless of technical skills, to build live applications that can read and write data stored in Databricks.
See how teams can build dashboards, operational tools, internal chatbots, and other custom apps directly on governed Databricks data without ETL, replication, or sync jobs.
2
9
43
3,718
dennylee retweeted
When rookie Devon Witherspoon announced himself to the league on Monday Night Football vs the Giants. 🍿
24
208
2,611
71,465