Build shelf
I'm an investigations attorney building AI-powered tools, workflows, automations, and deployment patterns for legal work. The rule behind the shelf: test what exists, build what doesn't.
Attorney workInvestigations, litigation, discovery, billing, governance, legal tech adoption.
Firm deploymentClaude Projects, plugins, skills, governance-ready instruction sets.
Build judgmentScope, review points, source visibility, auditability, and readiness gates.
Four builds on video. Watch them work before you read a word about them.
A local-first document finder for solo and small-firm lawyers: matter-aware search over a read-only local index, trust signals for extraction quality, OCR risk, and duplicates, and a review queue for weak extractions. Built because enterprise search tools do not run local-first on a lawyer's own machine.
A platform that gives legal teams a live view of their technology stack: intake, approvals, adoption, compliance gates, ROI claimed versus measured, and renewals. Built because buy-vs-build decisions deserve better records than scattered spreadsheets. Node and SQLite, with audited writes and read-only API seams.
A Supreme Court oral-argument simulator where an advocate argues out loud and a Justice questions back, grounded in real transcripts, Justice profiles, and replay coaching.
A single-file application that checks billing narratives against client-specific guidelines, with attorney self-service review, secretary bulk audit, file upload, matter switching, and rule extraction.
One pattern runs through all of them: find repeated legal work, make the workflow explicit, build the tool or automation layer, and design for attorney review.
A library of law-firm-ready Claude Project instruction sets for teams that cannot install custom skills or plugins but still need structured legal workflows inside Claude.
A work-tracking system for a 30-attorney investigations team, built with Microsoft Forms, Excel, and calendar automation. It replaced ad hoc status emails with a weekly capacity spreadsheet that partners use for staffing decisions.
An AI skill that generates defensible, matter-specific ESI discovery protocols for litigation and investigations, covering preservation, collection, processing, review design, TAR, GenAI-assisted review, privilege logs, and production. Built because no vendor makes protocol drafting matter-specific.
View on GitHubA React application for law firm knowledge managers to evaluate, classify, and approve new legal technology tools, with AI-powered analysis of vendor documents.
View on GitHubA browser tool that reviews and corrects timesheet entries against firm style guidelines: formatting rules, consolidation detection, and before/after comparisons. No installation, runs entirely in the browser.
View on GitHubA Python program that extracts structured background information and reasoning from court decisions using AI, and outputs the data to a professional Excel spreadsheet.
View on GitHubA Python program that analyzes court filings to extract parties, court details, procedural history, and litigation impact, with structured Excel output.
View on GitHubA curated collection of practical AI prompts for attorneys, organized by practice area and ready to copy-paste.
View on GitHubEvery build on this shelf started as a complaint or a curiosity. A partner, an associate, or a paralegal wished out loud that something was easier, so I built it, inside a locked-down firm environment. Or I saw a tool worth studying and built my own version, fitted to my practice group. I'm looking for a role that works the same way: listen for the friction, judge whether to buy or build, then train the team on whatever ships. Legal engineering, innovation, AI enablement, solutions. Remote first preferred.