GenAI PMAli Ghodsi
Databricks CEO mentioned as announcing ai_decide(). He connects the release to faster decision-making on data.
Key Highlights
- Ali Ghodsi consistently frames AI product strategy around cost, speed, governance, and running intelligence close to enterprise data.
- He spotlighted Smart Routing and inference-harness benchmarking as practical ways to cut AI costs without sacrificing quality.
- His comments on Genie emphasize that ontology, organizational context, and feedback loops are central to better enterprise AI answers.
- He has tied Databricks’ acquisitions and launches—including ElectricSQL, Row Zero, AI Extract, and ai_decide()—to a broader AI-native data platform strategy.
- His operational advice around database branching highlights the importance of safety controls for agentic products.
At a glance
- Role
- CEO
- Company
- Databricks
- X / Twitter
- @ali_ghodsi
Overview
Ali Ghodsi is the CEO of Databricks and a recurring signal source for where enterprise AI platforms are heading: lower-cost model orchestration, AI-native data workflows, and agent systems grounded in business data. In the newsletter coverage, he appears less as a generic executive voice and more as a product narrator for Databricks’ AI stack—especially around Genie, Unity AI Gateway, Lakebase, AI Extract, and ai_decide().For AI Product Managers, Ghodsi matters because his public comments consistently connect model capabilities to production constraints: cost, latency, governance, data safety, evaluation quality, and business context. His updates suggest a product strategy centered on running AI closer to enterprise data, routing across models intelligently, and favoring fast, typed, operational outputs over slower free-form generation when decisions need to scale.
Key Developments
- 2026-07-09: Ali Ghodsi shared results from an in-house evaluation across Databricks’ 3,000-engineer, multi-cloud codebase, arguing that swapping inference harnesses can cut AI costs roughly in half while preserving quality. He highlighted GLM 5.2 as a strong performer.
- 2026-07-18: He was mentioned in the context of Databricks raising funding at a $188 billion valuation to expand its AI stack, with emphasis on Unity AI Gateway for multi-model cost governance, Genie for business-data-aware AI coworkers, and Lakebase for serverless Postgres infrastructure aimed at AI agents.
- 2026-07-26: Ghodsi argued that agents that “cook” longer often do worse, using Genie as an example of faster result generation outperforming slower agentic workflows. He emphasized the importance of a strong ontology to give agents the context needed for accurate answers.
- 2026-08-12: He announced that Databricks acquired ElectricSQL, the team behind PGlite. The rationale was strategic: bring browser-based, sync-capable Postgres technology into the Lakebase Postgres roadmap.
- 2026-08-15: Ghodsi described how Smart Routing works in Databricks’ AI Gateway, calling it a simple approach that can lower AI costs by about 30% without reducing quality. He also explained that Genie uses signals such as org chart structure and popularity to choose the best answer, expose uncertainty, and learn from user feedback over time.
- 2026-08-21: He announced the release of AI Extract, a PDF field-extraction capability available directly from SQL. He claimed roughly 95% accuracy versus 87% for alternatives, at very low cost.
- 2026-08-28: Ghodsi recommended using branching databases on Neon Lakebase to protect against AI agents accidentally deleting production data, even sharing a concrete branch-creation command as an operational safeguard.
- 2026-09-22: He amplified Superhuman’s post on scaling GEC inference, calling its AI infrastructure work “super impressive,” signaling continued interest in practical inference systems at scale.
- 2026-09-25: Ghodsi announced that Databricks agreed to acquire Row Zero, noting internal usage where the finance team combined Genie with Row Zero’s Live Cloud Spreadsheet. He also teased a deeper combined Genie + Row Zero experience.
- 2026-10-02: He announced that ai_decide() had just been released, describing it as a way to run decision models natively on data in Databricks. He framed it as enabling fast, typed, “system one” decisions on big data, contrasting it with slower text generation.
Relevance to AI PMs
1. Model economics and routing are product features, not just infra details. Ghodsi’s comments on inference harnesses and Smart Routing show that PMs should benchmark multiple models and routing layers, not lock into a single default model. Cost reductions of 30%+ can materially change product margins and adoption.2. Grounding AI in enterprise context is a competitive advantage. His repeated emphasis on Genie, ontology design, org-chart signals, and data-native workflows suggests that better context architecture often matters more than simply using a bigger model. PMs should invest in semantic layers, metadata, and feedback loops.
3. Operational safety matters for agent products. His database-branching advice for agent workflows is a practical reminder that AI PMs need rollback, sandboxing, and environment isolation built into product design—especially when agents can write, modify, or delete data.
Related
- Databricks: The company Ghodsi leads and the main platform behind the product launches and acquisitions mentioned here.
- Genie / Genie Code: Databricks AI assistants positioned around business-data fluency and fast, context-aware answers.
- Unity AI Gateway / AI Gateway / Smart Routing: Infrastructure for multi-model governance, routing, and cost optimization.
- Lakebase / Lakebase Postgres / Neon Lakebase: Databricks’ Postgres and agent-data infrastructure direction, including safety patterns like branching.
- AI Extract: A Databricks feature for structured extraction from PDFs via SQL.
- ai_decide: A Databricks capability for running decision models directly on data, emphasizing typed decisions over text output.
- ElectricSQL / PGlite: Acquired technology relevant to browser-based Postgres and synchronization for Lakebase.
- Row Zero: Spreadsheet-style analytics layer Databricks agreed to acquire, with integration potential alongside Genie.
- GLM / GLM 5.2: Models Ghodsi referenced in evaluation context as strong cost-quality performers.
- Superhuman / GEC inference: External example he highlighted to underscore interest in production-scale AI inference systems.
Frequently asked questions
What is Ali Ghodsi?
Databricks CEO mentioned as announcing ai_decide(). He connects the release to faster decision-making on data.
What should AI product managers know about Ali Ghodsi?
Ali Ghodsi consistently frames AI product strategy around cost, speed, governance, and running intelligence close to enterprise data. He spotlighted Smart Routing and inference-harness benchmarking as practical ways to cut AI costs without sacrificing quality. His comments on Genie emphasize that ontology, organizational context, and feedback loops are central to better enterprise AI answers.
Explore this topic
Newsletter Mentions (14)
“Ali Ghodsi said ai_decide() had just been released, running decision models natively on data in Databricks.”
#6 𝕏 Ali Ghodsi said ai_decide() had just been released, running decision models natively on data in Databricks. He said it enables system one decisions on big data, framing fast typed decisions as a game changer compared with slow text generation.
“Ali Ghodsi announced that Databricks agreed to acquire Row Zero after its finance team combined Genie with Row Zero’s Live Cloud Spreadsheet.”
#4 Ali Ghodsi announced that Databricks agreed to acquire Row Zero after its finance team combined Genie with Row Zero’s Live Cloud Spreadsheet. He also teased a Genie + Row Zero experience.
“Ali Ghodsi shared Superhuman’s blog post about scaling GEC inference, calling the company’s AI infrastructure scaling “super impressive.””
#10 𝕏 Ali Ghodsi shared Superhuman’s blog post about scaling GEC inference, calling the company’s AI infrastructure scaling “super impressive.”
“Ali Ghodsi recommended branching databases on Neon Lakebase to prevent agents from permanently wiping data, sharing the command `neonctl branches create --name newbranch`.”
Ali Ghodsi recommended branching databases on Neon Lakebase to prevent agents from permanently wiping data, sharing the command `neonctl branches create --name newbranch`.
“Ali Ghodsi says AI Extract was released to extract fields from PDFs, achieving 95% accuracy versus 87% for others at extremely low cost.”
#1 𝕏 Ali Ghodsi says AI Extract was released to extract fields from PDFs, achieving 95% accuracy versus 87% for others at extremely low cost. The function can be called directly from SQL and used throughout the platform.
“Ali Ghodsi shared how Smart Routing works on the author’s AI Gateway, describing it as a “super simple idea” that lowers costs by about 30% without sacrificing quality.”
#10 𝕏 Ali Ghodsi shared how Smart Routing works on the author’s AI Gateway, describing it as a “super simple idea” that lowers costs by about 30% without sacrificing quality. #12 𝕏 Ali Ghodsi commented that Genie uses org-chart and popularity signals to select the “right” answer, surfaces uncertainty, and remembers user answers across future sessions to improve over time.
“"#8 𝕏 Ali Ghodsi announced that “we” acquired ElectricSQL, the team behind PGlite."”
#8 𝕏 Ali Ghodsi announced that “we” acquired ElectricSQL, the team behind PGlite. Its browser-based WebAssembly implementation of Postgres can sync asynchronously with Postgres instances and will enhance the Lakebase Postgres offering.
“Ali Ghodsi notes that agents that “cook” longer often underperform, while Genie's rapid result generation proves more efficient.”
GenAI PM Daily July 26, 2026 GenAI PM Daily 🎧 Listen to this brief 3 min listen Today's top 18 insights for PM Builders, ranked by relevance from X, Blogs, and LinkedIn. Perplexity unveils CLI for live web data #1 𝕏 OpenAI calls the Hugging Face incident an unprecedented AI safety event and is reviewing it with external advisors and its Safety and Security Committee. It will publish a technical report of findings in the coming weeks. #2 𝕏 Demis Hassabis reports that Gemma 4 models have been downloaded over 300 million times, driving the total Gemma open model series downloads past 900 million. #3 𝕏 Sundar Pichai celebrates Google’s commitment to open source, highlighting that they’ve long contributed and released open-weight AI models via the Gemma platform from Google DeepMind and Demis Hassabis. #6 𝕏 Ali Ghodsi notes that agents that “cook” longer often underperform, while Genie's rapid result generation proves more efficient. He argues that a solid ontology is vital for giving these agents the context they need to deliver accurate answers quickly.
“Ali Ghodsi is raising funding at a $188 billion valuation to turbocharge Databricks’ AI stack with Unity AI Gateway (multi-AI cost governance), Genie (AI coworkers fluent in your business data), and Lakebase (serverless Postgres for AI agents).”
#22 𝕏 Ali Ghodsi is raising funding at a $188 billion valuation to turbocharge Databricks’ AI stack with Unity AI Gateway (multi-AI cost governance), Genie (AI coworkers fluent in your business data), and Lakebase (serverless Postgres for AI agents).
“Ali Ghodsi ran an in-house evaluation on his 3,000-engineer, multi-cloud codebase and found that simply swapping inference harnesses can halve AI costs while maintaining quality, with GLM 5.2 emerging as a top performer.”
Today's top 25 insights for PM Builders, ranked by relevance from X, Blogs, and YouTube. OpenAI launches GPT-Live full-duplex voice API #1 𝕏 Sam Altman announced that GPT-5.6 Sol launches Thursday, urging builders to start integrating and experimenting with the new model. #2 📝 OpenAI News Introducing GPT-Live - OpenAI is launching GPT‑Live, a full‑duplex voice model that can listen and speak simultaneously, use conversational cues like “mhmm,” and delegate deeper searches or reasoning to GPT‑5.5 in the background; two versions (GPT‑Live‑1 and GPT‑Live‑1 mini) are rolling out to ChatGPT users globally today with an API sign‑up available. OpenAI says over 150 million people use ChatGPT voice weekly, reports users strongly prefer GPT‑Live to Advanced Voice Mode (GPT‑Live‑1 preferred ~75.7%), and shows large evaluation gains — GPQA rising from 45.3% (AVM) to up to 84.2% and BrowseComp from 0.7% to up to 75.2%. Also covered by: @Sam Altman #3 𝕏 OpenAI rolled out GPT-Live voice models in ChatGPT on iOS, Android, and web starting today (full rollout over the next few days), with API access coming soon—just tap the Voice button to talk with ChatGPT. Also covered by: @Sam Altman #4 𝕏 Mistral AI launched Robostral Navigate, its first embodied navigation model with 8B parameters that guides robots to perform natural-language specified tasks using a single RGB camera. It achieves state-of-the-art results on the R2R-CE benchmark. #5 𝕏 Logan Kilpatrick rolled out “import from GitHub” in Google AI Studio Build, automagically converting your repo into a runtime-compatible format. Now you can seamlessly iterate on it in AI Studio, deploy it, and more. #6 📝 OpenAI News Separating signal from noise in coding evaluations - A detailed audit of SWE-Bench Pro estimates roughly 30% of tasks are broken—an automated pipeline flagged 200 (27.4%) and human annotators found 249 (34.1%)—primarily due to overly strict tests, underspecified prompts, low-coverage tests, and misleading prompts. #7 𝕏 Cognition launched SWE-1.7, their most capable model yet, scoring within a few points of top frontier models at a fraction of the cost and running at 1000 tok/s. They report that their refined RL training recipe continues to deliver scaling gains. #8 𝕏 Ali Ghodsi ran an in-house evaluation on his 3,000-engineer, multi-cloud codebase and found that simply swapping inference harnesses can halve AI costs while maintaining quality, with GLM 5.2 emerging as a top performer.
Related
Claude's coding-focused workflow and command environment. The guide referenced here covers commands, memory, skills, subagents, MCP, hooks, plugins, checkpoints, and CLI usage.
OpenAI's coding-focused model/product, referenced here as part of the Google Cloud partnership. Important for PMs interested in code generation, developer workflows, and model distribution.
An open-source AI coding agent with a CLI, desktop app, and built-in agents for planning and building. It is relevant for coding workflows and codebase planning.
A Google model referenced as the benchmarked target for Harness² improvements. The issue reports gains on WorkBuddy-Bench and JobBench when its harness is optimized.
Vercel’s AI infrastructure layer for routing and managing model usage. The newsletter notes strong token volume growth on the gateway.
Stay updated on Ali Ghodsi
Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.
Subscribe Free