An enterprise AI copilot that turns legal and procurement knowledge into actionable workflows for Q&A, RFP creation, agreement generation, and supplier review.

LangGraphDSPyFastAPIPostgreSQL/RDSDynamoDBS3AWS Bedrock KBAzure OpenAILiteLLMpython-docx

2 expert copilots

ProcLegEnterprise assistant

Procurement assistantGrounded in 12 approved sources

What are our approved payment terms for strategic suppliers?

ProcLeg

Strategic suppliers default to Net 45. Net 60 requires Finance approval and a documented working-capital exception.

Procurement policyTerms playbook

Proposal comparison3 proposals attached

Compare Acme and Northstar. Which offer is stronger?

Acme is the recommended option

It is 8% lower on total cost with comparable delivery risk.

Acme91Northstar86

View evidenceDownload comparison

Legal assistantMSA v4 attached

Review this agreement and flag non-standard clauses.

I found 2 clauses for review

~~one year~~ three years retention

~~fees paid~~ 2x annual fees liability

Accept changes Download DOCX

Ask ProcLeg anything...

PROCLEG / LIVE

Procurement + legal intelligence workspace

Explore all labs

Lab Project

RevFlow AI

An AI-assisted revenue workflow platform that turns scattered account, opportunity, and activity data into structured workflows and next-best actions.

AI AgentsWorkflow EngineStructured OutputsRBACAudit Trail

View project View code Live demo

RevFlow workspace command center — Contract-to-revenue workspace

RevFlow billable customer accounts — Contract-to-revenue workspace

Read all essays

Field note 0113 min read

Building Agentic Applications: Workflow, Agent, or Something Between?

A decision guide to deterministic workflows, tool-using agents, and bounded agentic workflows—and why most production systems need less autonomy than expected.

Jul 16, 2026Open essay

Field note 0214 min read

Production RAG, Part C: Guardrails, Evaluation, and Observability

How to measure grounded answers, contain failures, trace behavior, and operate RAG safely—the part of the system that prepares for inevitable mistakes.

Jun 18, 2026Bring forward

Field note 0314 min read

Production RAG, Part B: Retrieval, Filtering, and Reranking

A retrieval pipeline that balances semantic relevance, exact terms, permissions, latency, and context quality—and the staged ranking problem behind a single search call.

May 14, 2026Bring forward

Field note 0415 min read

Production RAG, Part A: Ingestion, Chunking, and Embeddings

How to turn proprietary enterprise knowledge into retrievable, permission-aware evidence—and why production RAG often fails before a prompt is written.

Apr 09, 2026Bring forward

Field note 0513 min read

Building in the Cloud: AWS, GCP, Azure, and the Tradeoffs Between Them

How to select a cloud ecosystem and a deployment model using workload and organizational constraints.

Dec 17, 2025Bring forward

Field note 0614 min read

Designing Frontend and Backend Architecture That Can Evolve

A practical framework for boundaries, APIs, deployment shapes, and choosing technology without chasing trends.

Oct 08, 2025Bring forward

0106

Drag, swipe, or use the arrow keys to explore

Skills / constellation scan

One connected domain at a time.

AI & agentsConnected craft

Agent workflows, retrieval systems, evaluation, and document intelligence.

LangGraphRAGLLM systems

Auto scan 1 / 5

Skills / connected, not collected

The interesting work happens between the tools.

The connecting ideaSystems builder

Useful products emerge when every layer understands the others.

Let's build something interesting.

Whether it's AI systems, internal tools, or backend architecture, I'm always up for meaningful work.

Get in touch

coffee + good conversations

Engineering,made tangible.

Follow what sparks your curiosity.

Work

Lab

Essays

A featured case study from production.

ProcLeg Procurement and Legal AI Assistant

Experiments, side quests, and passion projects.