Internal assistants
Agents that retrieve context, call internal APIs, draft work, and route exceptions to humans.
AI agent builds
Build AI workflows that do useful work, not just demos.
I help teams design and implement agentic workflows that connect models to tools, documents, APIs, approvals, and ordinary application code.
Common work
Workflow design, prototype builds, API/tool integration, guardrails, and evaluation loops.
Platforms
AWS Bedrock, agent frameworks, .NET, Python, browser automation, and API backends.
Outcome
A narrow working system plus a clear path to production hardening.
Good fit projects
Document workflows
Extraction, classification, summarization, review queues, and system-of-record updates.
Operational automation
Model-assisted workflows for support, reporting, intake, research, QA, or data cleanup.
What I would inspect first
Workflow shape
- Who uses it and what decision it supports.
- Which steps need model judgment and which should be normal code.
- Where humans approve, correct, or audit the result.
Production risks
- Tool permissions and data access.
- Cost per run and expected volume.
- Failure modes, logging, evaluation, and fallback behavior.
Start here
Request reviewDescribe the workflow you want the agent to handle.
Include the current tools, data sources, and what a successful first version would do.