How a Legal Tech Firm Won Back 7-Figure ARR

A legal tech team encoded its private document corpus into a RAG research assistant on Azure — shipping beta in 90 days and pulling 7-figure net new ARR by month six.

7-figure ARR

Net new ARR at general availability

4–6 months

Implementation Time

Csongor Barabasi

Founder & CEO

Bonsai Labs

Connect ↓

The Challenge

A legal technology company with over five years in market faced displacement as the legal AI boom arrived. Rivals like Harvey AI and Legora raised large rounds and captured headlines, siphoning AI talent away from established players. The client’s product was not AI-powered, leaving them unable to attract the engineering talent needed to compete. Without a credible AI product in market quickly, the company risked losing customer trust, market share, and access to future capital.

What They Built

Bonsai Labs joined the client as an embedded AI strike team with a mandate to ship a proprietary legal research assistant — built on the client’s private document dataset — in three months. The team ran four stages: deep domain immersion with in-house lawyers, a privacy-first Microsoft Azure infrastructure build, evaluation dataset co-creation with legal staff, and tight weekly sprint cycles benchmarked against accuracy targets. What surprised the team was how decisive first-answer correctness became: lawyers abandon a tool the moment its first response is wrong, so engineering obsession on that single KPI unlocked user trust faster than any feature. The product reached beta in three months with paying customers, hit general availability at month six, and the resulting traction unlocked a new funding round.

Bonsai Labs embedded with the client as an AI strike team with a three-month mandate to ship a proprietary legal research assistant. The engagement ran in four stages. First, deep domain immersion: the team worked alongside in-house lawyers to understand how legal research actually functions, what questions practitioners ask, and what accuracy requirements would determine trust. Second, infrastructure: a privacy-first build on Microsoft Azure was designed to keep the client's proprietary document dataset secure while enabling OpenAI-powered retrieval. Third, evaluation dataset co-creation: rather than relying on generic benchmarks, legal staff helped build the evaluation set to reflect real practitioner queries. Fourth, tight weekly sprints benchmarked against a single KPI: first-answer correctness. The team discovered that lawyers abandon a tool instantly if the first response is wrong — making that metric the primary engineering focus. Beta launched at month three with paying customers. General availability hit at month six. The resulting traction unlocked a new funding round.

AI Role

Not shared

Infrastructure

Microsoft Azure (privacy-first cloud infrastructure) • OpenAI (LLM for legal research and retrieval) • Client's private document dataset (proprietary training/retrieval corpus) • Custom evaluation dataset (co-created with legal staff) • RAG (retrieval-augmented generation) architecture

Integration Points

Client private document dataset → Azure-hosted vector store → RAG retrieval layer • Practitioner legal queries → OpenAI LLM → first-answer legal research responses • Responses → evaluation dataset benchmarking → weekly accuracy sprint reviews • Validated product → beta (month 3) → general availability (month 6) → funding unlock

Impact

7-Figure ARR at Month Six

The AI-powered legal research assistant generated seven-figure net new ARR by its general availability launch — revenue that did not exist six months earlier and was entirely attributable to the new product.

Beta Live in 90 Days

Despite operating in a high-security, privacy-sensitive legal environment with millions of documents to index, the team reached beta with paying customers in exactly three months — the original mandate.

New Funding Round Unlocked

The successful AI product launch provided the commercial and credibility signal needed for the company to close a new funding round, repositioning it as a serious competitor in the legal AI market alongside Harvey and Legora.

Technology Utilized

Implementation Complexity

Revenue Growth

Time Savings

Not shared

Best Fit For

Mid-market B2B software companies — particularly PE- or VC-backed — with an established customer base facing pressure from AI-native competitors who need to ship a credible AI product fast without deep internal engineering resources.

Csongor Barabasi

Founder & CEO

Bonsai Labs

Founder & CEO, Bonsai Labs. Builds and deploys AI products for PE firms and B2B software portcos. Ex-ML engineer. Delivers rapid PoCs and enterprise-grade, secure AI systems.

Connect with the Expert

Industry:

Legal & Compliance

Technology & Software

Business Function:

Product & Engineering

Legal & Compliance

Company Size:

51-250

Project Cost:

Cost not disclosed

Ownership:

Venture-Backed

Organization Type:

Private Company

AI Pattern:

Knowledge Management & Search (RAG)

AI-Accelerated Custom Software

Value Type:

Revenue Growth

Time Savings

AI Model:

Frequently Asked Questions

How did a mid-sized legal technology company win back seven-figure ARR with a RAG research assistant?

The mid-sized legal technology company brought in an embedded AI strike team to ship a proprietary legal research assistant built on its private document corpus in three months. The team ran domain immersion with lawyers, a privacy-first Azure build, co-created an evaluation dataset, and focused weekly sprints on first-answer correctness. The product reached beta in 90 days and pulled seven-figure net new ARR by its month-six general-availability launch.

What AI tools and models did the legal technology company use?

The solution used knowledge management and search (RAG) with AI-accelerated custom software. It was built on a privacy-first Microsoft Azure infrastructure with OpenAI-powered retrieval over the client's private legal document dataset.

What results did the legal technology company achieve?

Three outcomes: seven-figure net new ARR by the month-six general-availability launch, a beta with paying customers live in 90 days despite indexing millions of documents in a high-security environment, and a new funding round unlocked by the product's traction.

How long did the engagement take?

Time to results was in the 6–12 month range overall. Beta launched at month three (the original mandate) and general availability hit at month six.

Who is this RAG research assistant approach best for?

Mid-market B2B software companies — particularly PE- or VC-backed — with an established customer base facing pressure from AI-native competitors, that need to ship a credible AI product fast without deep internal engineering resources.