Maintainers Roadmap
summary
Announces a conversational API in Dapr 1.15 that does batch inferencing and plugs to AWS Bedrock OpenAI Anthropic Mistral and Hugging Face with per-provider model support plus built-in PII scrubbing and in-memory caching with TTL to avoid repeated LLM calls. Workflow engine becomes scalable to scale to millions of workflows with Java Python Go C# JS/TS Rust and upcoming PHP SDKs. Scheduler becomes default for actor reminders with automatic migration. Open proposals include list keys delayed pubsub authorization and search while jobs and distributed lock APIs advance toward stable.
AI-written summary · about
chapters (12)
- 0:00Dapper’s roadmap intro
- 0:42Dapper’s conversational AI API
- 3:28LLM security and reliability
- 4:52Stable workflows and SDK
- 6:58Scheduler and actor improvements
- 8:01Dapper’s state management APIs
- 9:27State store and PubSub proposals
- 10:50Authorization and search APIs
- 12:15Maintainer priorities beyond 115
- 13:18Jobs and distributed lock APIs
- 14:43Bulk PubSub docs and stability
- 15:27Closing remarks and Q&A
related talks
45mMaintainer panel
45mOpening Remarks and Roadmap Update
09mDriving AI Innovation the Open Way
26mWelcome to the AI Jungle! Now What?
16mKeynote: Accelerating AI/ML Development with Open Source Software
58mAI in Reality Fireside Chat: Enterprise AI & Open‑Source Innovation
16mKeynote - State of Bazel and Announcements
15mOpenAssistant's Opensource LLMs Saga