a16z: AI agents' task length doubles every 7 months as Databricks bets on agents' system of record
_AustinCalvert_ · x · 2026-10-28
a16z's latest Profiles of the Future piece (Ben Horowitz & Michael McGuiness) argues agents and new database tech will reshape the firm.
Key points:
- The length of software tasks AI agents complete with 50% success has doubled roughly every 7 months since 2019 (METR data); recent models run ahead of trend, going from 30 seconds of human work in 2022 to a full workday today (Claude Opus 4.6 at 12 hours)
- Yet companies still don't trust AI to update Salesforce records. Databricks CEO Ali Ghodsi: "AI does not have an intelligence problem. It has a context problem."
- His answer: build a "system of record for agents" giving them reliable enterprise data and decision context
- Two radical changes ahead: agentic automation ("describe a task, automate it") and frontline employee empowerment—if organizations and their database tech evolve to support it
- Databricks has surpassed a $7B run rate growing 80% y/y, called Silicon Valley's most under-the-radar story
Related event: a16z Essay: Databricks Passes $7B Revenue as Agent Queries Overtake Humans(5 posts)→
More from coding & agent
- ZCode offers 300M free GLM-5.3-Flash tokens daily, with a quirky claiming window — AccBalanced · 2026-10-28
- Dev ships 3-month project in 2.5 weeks by letting Codex self-verify in a loop — kmeanskaran · 2026-10-28
- MLE-World: a learned world model predicts ML experiment outcomes to cut agent costs — Brian Ko · 2026-10-28
- Marc Andreessen on Replit: the dawn of the 'self-driving company' — amasad · 2026-10-28
- Accent-induced bug: speech-to-text turns "Claude Instance" into "cloud instance" for a coding agent — francesc · 2026-10-28
- Harrison Chase on self-improving agents: LangChain aims to be the model-agnostic all-in-one platform — hwchase17 · 2026-10-28