Ankit Mathur

Its the beginning of the end of subsidized AI subscriptions. GH Copilot is moving to usage-based billing, as has Claude (for business customers.) Fair to assume more will follow. I expect this change will also be a great boost for open models - cheaper, and pretty good already

GPT 5.5 and Codex are now available and manageable on Databricks! Both support Unity AI Gateway so you can manage access and costs, add guardrails, secure access to MCPs centrally, and audit usage. databricks.com/blog/openai-gp…

.@OpenAI GPT-5.5 is now available on Databricks, with Codex coding workflows and model inference fully governed through Unity AI Gateway. With GPT-5.5 on Databricks, you can: - Power coding workflows with Codex or other coding agents - Build custom agents grounded in enterprise data - Ask business questions in natural language from enterprise data with Genie - Automate document intelligence pipelines with Lakeflow Spark Declarative Pipelines Hear more from our co-founder, @pwendell, and OpenAI CRO @dhdresser about how Databricks and OpenAI are bringing frontier AI to the enterprise. Get started today: databricks.com/blog/openai-gp…

Databricks Unity AI Gateway now lets teams connect agents to external MCP servers like GitHub, Glean, and Atlassian through managed OAuth, with all governance running through Unity Catalog. Auth has been the consistent bottleneck for enterprise agent deployments, and Unity AI Gateway removes it entirely: • Register and control external MCP servers through Unity Catalog with fine-grained permissions and a full audit trail • Agents act on behalf of individual users, not overprivileged service accounts, so access stays scoped and traceable • Databricks handles the full OAuth lifecycle server-side, across AWS, Azure, and GCP Here’s how it works: databricks.com/blog/ai-gatewa…

Agentic and AI Coding Night returns on Thursday, May 7 in San Francisco. This one is engineered for backend and infra engineers, AI engineers, and founders moving beyond basic wrappers and into complex agentic workflows, repository-level reasoning, and high-scale execution. Speakers from Databricks, @AnthropicAI, @OpenAI, @datafoldcom, and more will cover: - The core tension and balancing productivity vs. accountability for coding agents - How to balance individual developer productivity gains vs. enterprise-wide operational risk - Fireside chat with @rohanvarma from OpenAI on coding and agents Doors open at 3:30pm. Come build with us! luma.com/agenticaiobsni…

@TheVixhal @alighodsi @pwendell @matei_zaharia Our moat is Data, AI, and most importantly, people.

Super excited to have @Yuchenj_UW on the team!

Life update: I joined Databricks this week! I thought I’d do another startup after Hyperbolic, but I was surprised by how startup-y Databricks AI is. @alighodsi, @pwendell, @matei_zaharia are in full founder mode. They’re the best founders I’ve met. I like working with people who aren’t “normal” and they definitely aren’t. For example, they invited me to an all-hands before I joined. I’m also impressed by how many former founders are here. @akhilgupta and @hanlintang are incredible leaders. A big bonus: I finally have unlimited Claude Code & Codex tokens! AI adoption on the Databricks AI team is insanely high. Every engineer I’ve met uses AI heavily and shares their own ways to drive agents. Many talented people here. I’m super pumped for this new journey!

Using Opus to reason over scanned insurance claims is the most expensive way to get a number wrong. Frontier agents can reason and code, but they're bottlenecked by the messy reality of enterprise data. At @databricks AI Research, we identify these silent failure points and build specialized systems to complement frontier models. We optimize relentlessly along the accuracy-cost frontier so agents perform at their best on real enterprise workflows. We build evals, train models, design agent harnesses, and serve the entire system at scale — millions of unstructured documents per tens of thousands of companies per year. Our ultimate goal: build the best agents that reason over data and drive decision-making in the enterprises at scale. We're hiring — reach out if this aligns with your passion!

AI Gateway in Databricks now extends Unity Catalog governance to agentic AI. As agents call LLMs, pull data through MCP servers, and invoke external APIs, every step touches sensitive data and audit requirements. With the latest updates to AI Gateway, you can now bring all of that into one governance model: • Set access policies once across every LLM and tool • Trace the full agent call chain end-to-end • Centralize logging for FinOps, engineering, and security databricks.com/blog/ai-gatewa…

Replit now deploys directly to Databricks. Your apps run inside your Databricks environment while inheriting its security, governance, and data access. Beta is live. Enterprises are already building with it and seeing massive acceleration in BI and internal tools.

One big problem with agentic coding today is that models are pretty “spiky.” For example, Claude Opus is better at frontend + agentic workflows, while GPT-5.4 is better at backend + distributed systems. But Claude Code and Codex are locked into their own models. You also often have to jump between them. I sometimes write code with Claude, then when it has a complex bug, I have to spin up a separate terminal to have Codex review it. Ideally, you’d want multiple models collaborating within the same context. Automatic model routing and cross-model or agent collaboration will be a huge unlock. There are a few technical challenges. Early model routers (like ChatGPT’s) were pretty rough. (Cursor and OpenCode seem to be in the best position to do this. Let me know if they already have strong model routers.)

At this rate, we'll all be spending every last dollar on Claude! Claude crushing it. The easy button if you want Claude forever. But many enterprise CTOs remind me single-vendor agent stacks are tomorrow's lock-in story. They want agents running across Claude, GPT, Gemini, and open source. Still an opportunity to build. Other winners will be the orchestration layers that treat models like interchangeable parts.

any bozo can do y-axis butchery, it needs real conviction to commit x-axis war crimes

Before Trump’s war the Strait of Hormuz was open, sanctions were in place, and gas was $2.39. Now Iran controls the Strait, sanctions will be lifted, gas is $4.99, 13 Americans are dead, over 300 injured, and the loss of hundreds of billions in aircraft and U.S. base repairs.

Iran Deal By Obama: •Strait of Hormuz open for free •Iran limits Uranium enrichment •Iran agrees to make no nuclear weapons •Iran allows Int'l inspectors to ensure compliance •Inspectors confirm Iran's full compliance Iran Ceasefire By Trump: •Strait of Hormuz closed, only open for $2M Per Ship •Iran makes no guarantee of limit on uranium enrichment •Iran makes no guarantee of no nuclear weapons •Iran makes no guarantee to allow Int'l inspectors MAGA: Trump Playing 5D Chess! Art Of The Deal!!

nah it's not that deep

the bigger story may be what comes next. Ghodi said it out loud: "With the sort of SaaS disruption that we’re seeing, Databricks will definitely partake in that disruption" He has the data, the AI... and now he's building the apps. Its the AWS playbook for the AI age.

Databricks enters cybersecurity... with Anthropic's models inside. The company that stores data now wants to secure it too. The incumbents prepared for AI entering the market.... less clear they're prepared for the data vendor thats already inside the building Sitting down w Databricks CEO @alighodsi today around 10:30a PT