Intapp's Permission-Aware AI Targets Law Firm Compliance Gap

A compliance problem disguised as an AI opportunity

Intapp, a NASDAQ-traded software company serving professional services firms, began when a patent attorney complained his law firm couldn't manage data integration. That early 2000s gripe evolved into a business now valued by public markets — and a competitive position the company believes insulates it from larger AI vendors.

The firm launched Celeste, its agentic AI platform, in February and made it generally available in mid-July. According to CEO John Hall, the product addresses a structural problem generic AI tools create inside regulated firms: they answer questions accurately but ignore whether the person asking is authorized to receive that information.

Why it matters

Large language models excel at surfacing information but lack native understanding of organizational permission structures. For law firms handling confidential client matters, investment banks managing deal flow, or accounting firms navigating audit walls, an AI that tells the truth to the wrong person creates liability. Intapp's thesis is that two decades of compliance infrastructure and firm-specific data create switching costs horizontal AI providers can't overcome by adding permission layers after the fact.

The unusual path to a public offering

Intapp never raised venture capital, an anomaly for Silicon Valley software companies. The firm's first outside funding came with its 2021 IPO. That bootstrap history reflects both the narrow focus of its initial product and the referral-driven growth model that followed.

After licensing its data integration tool to that first law firm client, Intapp hired the firm's chief information officer as its inaugural salesperson. He introduced the product across his professional network, and the company expanded through word-of-mouth in a tightly connected industry.

Hall characterizes Intapp's customer base as "large partnership firms" — a category spanning law firms, accounting practices, investment banks, private equity shops, and consultancies. He estimates the addressable market at roughly $4 trillion in annual U.S. revenue, though the segment's regulatory complexity keeps most general-purpose software vendors at arm's length.

Permission as product differentiation

Intapp positions Celeste as distinct from practice-focused AI tools like Harvey and Legora, which assist with legal work itself, and from enterprise assistants like ChatGPT Enterprise. The company partners with Harvey rather than competing directly.

The technical challenge Celeste addresses is granular access control at scale. Inside a 3,000-person law firm, document visibility varies by client relationship, conflict checks, practice group, and regulatory requirements. Hall argues that building the AI model is straightforward; encoding which employees may access which information, and maintaining an audit trail, is not.

Generic AI platforms lack the firm-specific data and compliance frameworks Intapp accumulated over 20 years serving this vertical. According to Hall, horizontal systems "tell the truth" but can't determine whether a given user should receive a particular answer within the firm's governance structure.

The details were first reported by AI Watch in an article examining Intapp's approach to governed AI for professional services firms.