What the community has built

AI-powered web app that edits PDF documents through natural language instructions. An agent interprets what you want changed and handles the structural manipulation — no manual editing required.

Nolan HurlburtLegal AI Strategist

The Noroboto family of attacks, originally developed by Drew Miller of Tritium Legal and the LQ RED TEAM, rewrites the TrueType glyph-cmap of an embedded font so that ASCII codepoints in the content stream map to glyphs for arbitrary other characters. The visible page renders correctly because the font's cmap draws the intended glyph; the extracted text yields Private Use Area noise because the codepoints in the underlying text are PUA values. The original public release applied to DOCX documents only. I added a PDF extension to Noroboto which consists of a single new code path in noroboto.py that, given an arbitrary input PDF, obfuscates it. You can try it here: https://noroboto.io. Only total-document obfuscation is shipped publicly. Surgical (partial-targeted) and replacement variants are deliberately withheld for ethical reasons, until a mitigation system is broadly deployed at the ingestion layer.

Alexios vdSKAssociate General Counsel

Any lawyer can transform her know-how into a composable, modular skill which can be published to the Dealroom marketplace or downloaded to a local deployment of the solution.

Sergio MaldonadoTriple-admitted lawyer with a programming background, twice-exited entrepreneur

Lexometrica Ground Truth is an independent LLM leaderboard built on a closed, static dataset of 30 highly complex cases derived from real Russian court practice. We discard standard memorization metrics to test real legal intelligence within the IRAC (Issue, Rule, Application, Conclusion) logical framework: evaluating the models' ability to identify hidden problems (issue-spotting), apply relevant norms to facts (rule-application), and draw accurate conclusions.

Nikita PolyakovLegalTech Architect (ex-Litigator)

Sonomos is a local-first privacy layer for professionals who use AI tools but can’t afford to expose sensitive data. Its two core features — Dagger, which detects sensitive information in real time with a traffic-light overlay, and Cloak, which masks that data before it ever leaves your machine — sit between the user and whatever AI tool they’re typing into. No cloud processing, no trust required. Built for lawyers, financial advisors, and anyone in a regulated industry who needs to use AI without violating client confidentiality. Currently running pilot programs with law firms.

Zachary BrennerFounder & JD/MBA Candidate

Glyph is a Chrome extension that puts §, ¶, ©, ™, and 26 other symbols lawyers actually use right next to whatever text field you’re typing in — Clio, Gmail, Google Docs, wherever. One click to insert. No alt codes, no character map, no breaking your flow.​​​​​​​​​​​​​​​​

Zachary BrennerFounder & JD/MBA Candidate

My first attempt at "solving" OCR. The idea was that I would load up an unreasonable number of OCR engines and have a judge determine which extraction should win. It is an interesting start, but performance was inconsistent and better extraction models appeared shortly after.

Erin FraserTrusted advisor to private clients

Corporate legal teams overpay law firms every month. Rate overcharges, duplicate entries, non-compliant line items — most go unnoticed. Invoice Checker fixes that: 📄 Upload any invoice (PDF, DOCX, image) 🔍 24 automated checks — rate variances, duplicates, seniority mismatches, scope creep ⚖️ Cross-referenced against your panel rates and engagement letters ✅ Confirm, waive, or dispute each flagged item 📧 Auto-generates a PDF report and draft dispute email 🔒 Your data never leaves your infrastructure (BYOT — bring your own API key) Open source · Self-hosted · MIT licence

Nicolás PaniguttiLegal Ops | Intrapreneur

Tritium is the next-generation word processor for legal--the legal IDE.

Drew MillerFounder

An AI-powered operating system for in-house legal teams. Centralizes escalations, playbooks, and knowledge, and gets smarter every week from the work the team does. How it works Slack /escalate command kicks off any matter — no new tool to learn for requesters AI triages every matter: classifies practice area, sets priority, routes to the right attorney Pre-analysis runs before a lawyer touches the matter — pulls relevant playbooks, knowledge base articles, prior similar matters, and counterparty memory into a synthesized brief Drafting workspace composed from a versioned clause library; export to Notion or Google Drive Matter-scoped Copilot has the full context (playbooks, KB, prior matters) and can be invoked anywhere The self-learning loop: Seven weekly cron jobs mine the team's actual decisions to update the institutional brain: Rejection mining clusters lawyer disagreements and proposes new playbooks or routing patches Playbook-edit mining compares matched playbooks against final lawyer-shipped output and proposes targeted amendments Conflict detection scans for duplicate canonical clauses, rule priority collisions, and near-duplicate playbooks Every accepted lawyer edit becomes the next version — the corpus compounds Architecture: Stack: Next.js · tRPC · Drizzle ORM · PostgreSQL · Railway Voyage-Law-2 embeddings across knowledge, playbooks, templates, rules, and execution patterns — global ⌘K semantic search Natural-language rule engine (SLA · routing · triage · playbook triggers) with a four-state lifecycle: draft → shadow → active → archived Codified statutory-research workflow with citation verification — designed to eliminate the hallucinations of pure-LLM legal research Integrations Slack (trigger), Salesforce (matter context), Notion (canonical store), Google Drive, GitHub, Anthropic API (Copilot · analysis) Status: proof of concept, deployed and operational.

Marco CrocettiSenior Counsel

A Claude Code agentic workflow that automates the full lifecycle of a blog post — from pitch to published. Built for Alt + Counsel, it runs five phases: pitch, write, review, post, and verify. The review phase uses three synthetic personas (a legal technologist, a solo corporate lawyer, and a lawyer-coder) to audit drafts before anything goes live. The thing that writes the blog is itself a project worth showing.

Hou Fu AngSenior Legal Counsel

license-comply is an open-source CLI tool that scans Python projects for dependency license risks, going beyond simple license detection to provide risk ratings, plain-English explanations, and remediation steps tailored to your specific project type. It features a customizable policy engine, multiple output formats, CI integration, and an optional AI-powered executive summary. Published on PyPI and licensed under Apache 2.0, it's built to make open-source compliance accessible to both developers and legal teams.

Sam ClearwaterSenior Vice President & Counsel - AI

Identify which source sentence a piece of text most likely originated from, using lightweight text similarity metrics. No embeddings or API calls required at runtime — just fast, deterministic string matching that works in the browser or on the server. Designed for RAG citation highlighting, contract playbook matching, plagiarism detection, source attribution, and any other text provenance task.

Yu Chou TeoGovernment Lawyer

This is an MCP server. You give an AI agent a knowledge document and a questionnaire. The agent reads both, matches answers to questions, and writes them directly into the original Word, Excel, or PDF. Tested with Claude Code and Gemini CLI.

Oscar GC is a single-binary Electron application for in-house legal work. Lawyers open a practice area, create a matter, and work with a scoped AI agent that knows the area, has access only to that matter's documents, and is paired with vendored in-house legal skill libraries plus a contract-redline service. The app runs locally on Linux,.

Intelligent Q&A platform that routes questions to domain experts, then converts their answers into automation rules so the same question never needs a human twice. Compounds over time — the more experts answer, the more the system handles on its own.

AI-powered legal query optimization platform that builds institutional memory from uploaded documents, tracks knowledge expiration with a “Good Until Date” system, and generates precisely-scoped requests for outside counsel — cutting 30–50% of unnecessary spend by surfacing what the department already knows.

A host-independent OOXML reconciliation engine built specifically to solve one of the hardest problems in legal tech: programmatic track changes. This NPM package safely converts AI-generated or programmatic edits into valid, native Microsoft Word redlines (w:ins/w:del). It handles complex edge cases like list generation, table diffing, and comment injection entirely dependency-free, allowing developers to build robust document automation tools across Node.js, browsers, and Deno.

Microsoft Word Add-in that integrates Gemini AI directly into the editing environment with full Track Changes support. Bring-your-own-key architecture — documents route straight to Google’s API, never through a middleman. Handles edge cases like nested lists and table formatting where most AI tools fall short.

I built an agentic-first legal IDE from scratch: Python/FastAPI backend, SvelteKit frontend, PostgreSQL, and Docker; model-agnostic agentic loop with tool-calling, subagents, multi-step orchestration, and human-in-the-loop review; hybrid BM25 and pgvector RAG retrieval; binary diff document versioning; 6-layer privilege-based permission model.

Matteo BürglerTMT Associate · Programmer

Detects the active window every 60 seconds — documents, emails, research, calls — and logs automatically. Step away and forget to manually pause? No problem. If there's no keyboard or mouse input for 2+ minutes, Ghost Logger automatically flags the entry as (Idle) — so idle time never sneaks into billable hours. All data saved locally. No cloud, no subscription. User can open the CSV log in Excel anytime for billing or review.

Sze Yao TanState Counsel

Billing review tools exist, and some of them are good. But they're typically designed for legal departments with hundreds of firms and millions in annual spend. Plus: law is a human business! I don't want to mediate my relationships with counsel through a third-party platform. I don't want a dashboard. (I really despise dashboards.) I want to understand what I'm paying for, I want to work more harmoniously with our partner firms, and I want to have an informed conversation about it when the numbers don't look right. So I set out to build a tool to help with that: Eden. A Loom video showing Eden's capabilities is here: https://www.loom.com/share/e063a1080a874b348dfb20b5eaf24dd8

Andrew CookeChief Legal Officer, Perk