AI for US mid-market law firms
Founded 2017. Eight years of client work, eighteen months operating Casetrack in production, and one fixed-price engagement model that ships in eight weeks.
AppRocket has been a working consulting firm since 2017. The team here today is the team that spent eighteen months operating Casetrack, a production legal-software product used by real attorneys handling real matters. We learned by doing what evaluation rigor, attorney-trust onboarding, and integration brittleness actually cost in the field.
We sunset Casetrack in 2024. The honest reason is that the lessons it taught us were earning their keep faster as services engagements than they were as a single product. Today we build production AI for mid-market law firms of 50 to 500 attorneys, using four templates we have shipped before: intake triage, doc review assist, conflict-check resolution, and billing reconciliation.
The firm is led by Qasim Zafar. The core team is six people working as a distributed team headquartered in Palo Alto, California, plus two specialist engineering-partner firms (one focused on evaluations and RAG, the other on integration and UI). That structure lets a small core deliver enterprise-grade work at boutique speed.
Eight years of client work across mid-market services firms, NASA Artemis, US legal, fintech, and healthcare. The production AI portfolio is what we anchor on today.
Three fixed-scope tiers, one pricing page, no qualifying call disguised as a deck review.
$15,000 · 2 weeks
We map which workstreams are worth automating, which are not, and what each one would cost to ship. The deliverable is a plan you can act on with us, with another vendor, or in-house.
Learn more→$80–250K · ~8 weeks
Production AI tested against a fixed evaluation harness from week one. Four templates we have shipped before: intake triage, doc review assist, conflict-check resolution, billing reconciliation.
Learn more→$8,000/mo · ongoing
The operating work that keeps production AI in production: eval drift, model swaps, incident response, and quarterly capability upgrades.
Learn more→AppRocket is led by Qasim Zafar. Stanford MS&E, Sequoia InSITE Fellow, NASA Artemis subcontractor, and the operator who ran Casetrack in production for eighteen months. Read the full bio →
Qasim Zafar
Founder & CEO
Stanford MS&E
Management Science & Engineering researcher
Sequoia InSITE Fellow
Sequoia Capital fellowship for early-stage founders
NASA Artemis subcontractor
Product manager — Lunar Gateway program
Savaree → Careem exit
Founding-team operator experience
Operator of Casetrack
18 months running production legal software (sunset 2024)
Stanford MS&E researcher and Sequoia InSITE Fellow who founded AppRocket in 2017. Built production AI for NASA Artemis (Lunar Gateway), digitized law firms end-to-end, and ran Casetrack — a production legal-software product used by real attorneys — for eighteen months before sunsetting it in 2024 to focus on services.
A six-person distributed team, plus two engineering-partner firms. Click through for full bios.
Four operating principles that show up in the codebase, the evaluations, and the engagement model.
We publish what worked and what didn’t. The Casetrack retrospective is on this site as primary-source research, with the failures included.
A six-person core team plus two engineering-partner firms. We staff the work the engagement actually needs, not the headcount an agency template would bill.
The evaluation harness is the deliverable. Model choice comes after we have agreed on how to measure correctness, latency, and the cost of being wrong in front of a paying customer.
The founder shows up in architecture review and the production push, not on the kickoff slide and the steering-committee call. That is the bar mid-market firms are paying for.
Because every implementation is a launch. The work that pays the bills is built, deployed, and operated, not pitched. That is the bar the name commits us to.
Headquartered in Palo Alto, California. We work as a distributed team across time zones.
Thirty minutes, no deck required. We will tell you the right place to start.
