{"schema_version":"newruntime-agent-readable-v0.2","type":"post","stable_id":"post:stripe-kai-knowledge-agent-platform","slug":"stripe-kai-knowledge-agent-platform","title":"Stripe Builds A Shared Agent Platform For Knowledge Work","description":"Stripe's Kai combines surface-agnostic APIs, domain-owned AgentStudio assets, per-session sandboxes, long-horizon state, and more than 1,000 internal tools and skills.","retrieval_nugget":"Stripe's Kai combines surface-agnostic APIs, domain-owned AgentStudio assets, per-session sandboxes, long-horizon state, and more than 1,000 internal tools and skills. Stripe's Knowledge AI Platform, Kai, is a useful counterexample to the idea that knowledge work needs thousands of independent micro-agents. Before Kai, Stripe employees had built more than 4,000 workflow-specific agents. Prompt quality varied, ownership fragmented, and maintenance became difficult.","status":"published","published_at":"2026-08-01","updated_at":"2026-08-01","record_date":"2026-08-01","date_kind":"published_at","topics":["knowledge-work","agent-platforms","skills","enterprise-ai"],"source_urls":["https://stripe.dev/blog/meet-stripes-knowledge-ai-platform"],"visuals":[{"id":"stripe-kai-knowledge-agent-platform","kind":"editorial-diagram","role":"hero","src":"https://newruntime.com/images/posts/stripe-kai-knowledge-agent-platform.webp","alt":"Hand-drawn three-layer knowledge agent platform with many work surfaces, a domain-owner control plane, and secure per-session execution connected to a large skill and tool library.","caption":"Stripe separates work surfaces, domain-owned expertise, and secure execution so one agent platform can serve diverse knowledge workflows.","credit":"New Runtime synthesis from Stripe Engineering","source_url":"https://stripe.dev/blog/meet-stripes-knowledge-ai-platform","generated_with":"gemini-3.1-flash-image","width":1600,"height":900,"legend":[{"label":"Surfaces","description":"Web, Slack, browser extensions, and internal applications share the same agent APIs."},{"label":"Domain control","description":"Experts own skills, tool selections, custom agents, quality signals, and expected outputs."},{"label":"Execution","description":"Per-session sandboxes, virtual filesystems, orchestration, and access controls support long work safely."}]}],"routes":{"html":"https://newruntime.com/posts/stripe-kai-knowledge-agent-platform/","markdown":"https://newruntime.com/posts/stripe-kai-knowledge-agent-platform.md","json":"https://newruntime.com/posts/stripe-kai-knowledge-agent-platform.json"},"source_format":"markdown","next_reads":[{"type":"topic","path":"/topics/enterprise-ai/","reason":"Explore the enterprise ai topic hub.","url":"https://newruntime.com/topics/enterprise-ai/","title":"Enterprise AI - New Runtime","media_type":"text/html"},{"type":"topic","path":"/topics/knowledge-work/","reason":"Explore the knowledge work topic hub.","url":"https://newruntime.com/topics/knowledge-work/","title":"Knowledge Work - New Runtime","media_type":"text/html"},{"type":"related_material","path":"/posts/drskill-agent-loadout-audit/","reason":"Shares skills.","url":"https://newruntime.com/posts/drskill-agent-loadout-audit/","title":"Dr. Skill Audits What An Agent Loads Before It Works","media_type":"text/html"},{"type":"related_material","path":"/posts/genkit-agent-skills-progressive-disclosure/","reason":"Shares skills.","url":"https://newruntime.com/posts/genkit-agent-skills-progressive-disclosure/","title":"Genkit Adds Progressive Disclosure For Agent Skills","media_type":"text/html"},{"type":"related_material","path":"/posts/glean-ai-consulting-operating-model/","reason":"Shares knowledge work.","url":"https://newruntime.com/posts/glean-ai-consulting-operating-model/","title":"AI Changes Consulting When Firm Knowledge Becomes A Working System","media_type":"text/html"}]}