United StatesProfessional services2026 – ongoingCaylent
Caylent Accelerate, an agentic system for database migration and cloud operations
Caylent cut engineering hours by 90% on a 2,500-procedure database migration with a Claude agent system
Problem
Moving legacy commercial databases to open-source ones on AWS meant translating thousands of stored procedures by hand, at roughly four hours each. Migration assessments took months of dependency mapping, so many customers put modernization off.
Approach
Caylent, an AWS consultancy, built Caylent Accelerate, which runs hundreds of Claude instances in parallel inside each customer's AWS account through Amazon Bedrock. A first harness proved brittle, so the team rebuilt it on the Claude Agent SDK. Specialized agents parse inventories, translate dialects, validate and summarize, Haiku handles high-volume parsing and Sonnet handles synthesis, and engineers check edge cases.
Results
Takeaway
Part of the saving came from finding that about 1,850 of the 2,500 procedures in one migration were dead code that needed no translation.
Categories
- Industry
- Professional services
- Business function
- IT and R&DProduction and quality
- Innovation type
- Process innovationProduct and service innovation
- AI technology
- AI agentsGenerative AI
- Adoption
- Built in-house
- Evidence
- Vendor case study
Sources
Source summaries
Anthropic describes Caylent Accelerate, built on the Claude Agent SDK, and its use in customer database migrations. It reports a 90% cut in engineering hours on one 2,500-procedure migration, 58% to 82% faster migrations in four other engagements and a $600,000 annual licensing saving. The page carries no date; the Internet Archive first saved it on July 21, 2026.
Full details are in the original sources.
