

Breaking the Cloud Barrier: How DBOS Transforms Application Development
Dec 19, 2024
Chan Li and Peter Kraft, co-founders of DBOS, Inc., discuss their groundbreaking serverless platform designed for highly reliable applications based on Postgres. They tackle challenges in error handling for stateful applications, introducing innovative solutions like an AI refund agent. The duo explains their transition from academic research to commercial product, focusing on enhancing cloud developer experience through improved data management and programming support. They also delve into DBOS's integration with OpenTelemetry and its promising advancements in AI applications.
AI Snips
Chapters
Transcript
Episode notes
AI Refund Agent Demo
- Qian Li demonstrated DBOS with an AI refund agent.
- This agent interacts with real-time systems, escalates complex refunds to humans, and maintains state across interruptions.
Agent Development Challenges
- Building stateful, asynchronous agents with existing tools requires complex integrations across multiple services like AWS Lambda, SQS, and Kafka.
- This leads to code bloat and maintenance challenges.
Shopify Integration Success
- An early DBOS user successfully built a Shopify integration in two days.
- Previously, they struggled for two months using AWS Lambda, Step Functions, and EventBridge.