PDI Brew: Amazon Bedrock and Lambda auto-deploy internal apps from text
Non-developers describe a tool and get a secured AWS web app in seconds
PDI Technologies, a 40-year-old firm serving 200,000 customer locations, created PDI Brew to eliminate the backlog of small internal tools that never get built. Traditional delivery—repositories, pipelines, auth, hosting—stalls shipping-cost calculators and dashboards for months. PDI Brew flips that: an employee describes the tool, and a two-agent system handles the rest. The planning agent interviews the user, refines the request, and outputs a structured manifest, using either a skill inside an AI assistant like Claude or a Bedrock model invocation. Then a provisioning agent running on AWS Lambda decomposes that manifest, classifies the workload, and orchestrates creation of downstream resources—API Gateway, DynamoDB, S3, CloudFront, and IAM—within a single request.
Security is built-in, not bolted on. Every generated app inherits enterprise SSO via Entra ID, least-privilege scoped IAM, HTTPS, and centralized observability, leaving no insecure path. The serverless, scale-to-zero architecture keeps hundreds of apps nearly free when idle. For AI, PDI Brew provides a governed gateway to Amazon Bedrock, exposing chat, summarization, and classification with guardrails, quotas, and full audit trails—so no app author ever touches a model endpoint or API key. The full architecture demonstrates how agentic provisioning patterns can safely hand over software delivery to non-technical users.
- PDI Brew's planning agent emits a structured manifest; a Lambda provisioning agent creates all AWS resources in one request
- Generated apps include SSO, HTTPS, scoped IAM, and centralized observability by default
- In-app AI via Amazon Bedrock is governed with guardrails, quotas, and audit trails—zero API keys for app authors
Why It Matters
Clears the internal-tool backlog: business users ship secure, serverless apps without DevOps, and AI stays centrally controlled.