Intapp Built a $578 Million Business on the Work Nobody at a Law Firm Wants to Do

The first version of Intapp was not an AI coworker. It was a box. In 2003, the young company - then called Tsunami Software - installed a dedicated Integration Appliance at the law firm Loeb & Loeb. Its task was brutally practical: keep the same data synchronized across several systems. Nobody mistook it for magic. It simply stopped important information from drifting apart.

That modest job contains the blueprint for everything Intapp became. Professional firms do not sell shoes or subscriptions; they sell judgment. Their inventory walks out of the elevator each evening. Their best business often arrives through a partner's relationships. And before a new matter can start, somebody must check conflicts, client terms, access rights, independence rules and, eventually, whether every billable six minutes made it onto an invoice.

Horizontal software can be configured to handle all that. Intapp's proposition is that it should not have to learn the vocabulary from scratch. The company builds software around the nouns and anxieties of law, accounting, consulting, investment banking, private capital and real assets: matters, mandates, funds, limited partners, lateral hires, ethical walls and outside-counsel guidelines.

A tiny company picks a narrow door

Thad Jampol, Dan Harsell and Vera Newman started Tsunami in December 2000, just as the dot-com mood curdled. They had no lush launch story. Within six months, they had a customer. Two years later they shipped their no-code integration product. When Don Coleman traveled to a Florida trade show in 2003, a board member supplied frequent-flyer miles. There was still no money for a booth, so the team stood in somebody else's.

The first thing that failed, in other words, was not the product. It was the usual startup apparatus: travel budget, booth, polish. The team worked around it because the client problem was specific enough to demo. Loeb & Loeb needed one data set to remain consistent across multiple systems. The appliance did that job, and its name - Integration Appliance, shortened to IntApp - eventually renamed the company.

The founders began with integration, then moved closer to the decisions that make professional firms peculiar. Walls arrived in 2007 to restrict sensitive information. Time followed in 2008. Intake and Conflicts launched in 2014. The sequence mattered: each product encoded another piece of how a regulated, partner-led firm actually operates.

The moat is not the prompt. It is knowing that one colleague may see a matter another may not.

DealCloud made the back office face the market

Intapp's defining expansion came in 2018, when it paid $123.3 million in cash for DealCloud. The acquired company had been built by people from private equity who disliked the generic tools available for tracking deals and relationships. It gave Intapp a front-office system: not merely approving and billing work, but sourcing it, mapping who knows whom, managing a pipeline and remembering why a promising conversation went quiet.

DealCloud is often called a CRM, but the useful distinction is its data model. A private-equity firm thinks in funds, intermediaries, portfolio companies, deals and investors. A law firm thinks in clients, matters, partners, practices and experience. DealCloud connects these objects and captures emails and meetings, then overlays market data, relationship scores, workflows and reporting. Hamilton Lane says it replaced 28 systems and processes with the platform. Taylor Wessing made adoption unusually plain: if the data was not in DealCloud, its managing partner told staff, it did not exist.

The rest of the portfolio wraps around that lifecycle. Intake gathers new-business information. Conflicts searches for incompatible interests. Terms extracts and enforces client obligations. Walls controls access. Collaboration and Workspaces organize documents and email around Microsoft 365. Time finds billable activity. Billstream catches billing-rule problems before an invoice annoys a client. In 2025, the roughly $50.9 million net-cash purchase of TermSheet extended the pattern into real-estate investing; Intapp launched the combined Intapp Properties product in 2026.

The cloud turn - and what it cost

For years, Intapp sold both on-premises and cloud software. In 2020 it changed its mind about straddling the two and committed to selling cloud solutions exclusively. That choice made the business easier to update, connect and measure, but migrations are expensive in attention. The customer must clean old data, settle process arguments, configure permissions, integrate email and finance systems, train reluctant partners and decide which spreadsheet is finally allowed to die.

By June 2026, cloud ARR was $495.7 million, up 29 percent in a year, and cloud represented 84 percent of total ARR. Total ARR reached $590.5 million. More than 1,400 clients spent at least $50,000 annually; 142 spent more than $1 million. Subscription depth is the model: sell directly, usually under one- to three-year terms, charge by users and modules, and expand as a customer connects more of its business.

Celeste is the old strategy wearing an AI badge

Celeste, generally available since July 2026, is Intapp's bid to move from systems people operate to an agent that operates parts of the firm. It can prepare meeting briefs, screen incoming deals, begin a conflict check, update records and produce documents. The pitch avoids the blank-chat-box problem. Celeste inherits a firm's definitions, permissions, relationship graph and configured methods, then runs repeatable playbooks with human approval where risk enters.

This is why the company's old products suddenly look contemporary. An AI agent inside a law firm cannot simply retrieve the best answer. It has to retrieve the best answer the requesting person is permitted to see. It must respect an ethical wall, material nonpublic information and a client's confidentiality terms. Intapp has been selling that control layer since before “agent” became a software noun.

Partnerships widen the aperture. Microsoft supplies Azure, Microsoft 365, Copilot integration and distribution; Intapp says the companies co-sold eight of its 10 largest fiscal 2026 deals. KPMG brings implementation and risk expertise. Anthropic supplies Claude models. Harvey connects legal-work agents. Moody's, MSCI, PitchBook and FactSet contribute market and screening data. Snowflake helps carry analytics. Intapp does not need to own every model or dataset if it controls the context and policy between them.

What buyers can copy - and where the fit gets awkward

The results customers publish are concrete enough to resist brochure language. Mintz says Billstream cut time spent editing bills by 35 to 40 percent. FMI Capital Advisors reports saving more than eight hours a week preparing transaction bidder lists. Alterna Equity Partners says qualified deal flow rose 50 percent. Gerald Edelman implemented DealCloud in four to six weeks after data cleanup, dashboard configuration and training.

Those examples also disclose the condition: software only compresses a workflow once a firm agrees what the workflow is. A beautiful relationship graph fed by stale contacts remains a beautiful picture of the past. An ethical wall is useful because people trust the permissions beneath it. A CRM becomes institutional memory only when partners stop treating it as clerical punishment.

1. Start with expensive specificity. Pick a workflow where generic software creates manual work or real risk.
2. Encode the nouns. Build the data model around how customers describe their world, not around your database tables.
3. Land beside daily work. Email, calendars, documents and finance systems are where adoption is won.
4. Make governance productive. Controls should accelerate safe work, not merely record why it stopped.
5. Earn the platform. Intapp expanded from one integration problem into adjacent workflows over decades; the suite followed trust.

This approach will not work everywhere. A 20-person shop with a clean spreadsheet and simple pipeline may find the configuration, governance and buying process excessive. A firm unwilling to clean data or enforce adoption can waste a sophisticated system. Local competitors may understand regional language and rules better. And broad platforms such as Salesforce and SAP can appeal to companies that value one general stack across many divisions.

Intapp also remains a growth company with tension in the accounts. Fiscal 2026 produced $108.6 million in non-GAAP operating income but a $40.1 million GAAP operating loss. The company spent $275.2 million repurchasing shares during the year. It must keep cloud growth high, turn agents into paid usage, integrate acquisitions and defend its niche against larger horizontal vendors and sharper point solutions.

Still, Intapp's market position is unusually coherent. It sits between general enterprise platforms and single-purpose legal or deal tools, offering a connected operating layer for firms whose product is expertise. The company began by keeping records synchronized. Now it wants an AI coworker to act on those records. The distance between the appliance and the agent is 25 years; the underlying sales pitch has barely moved: your firm is strange in important ways, and the software should know that before the first meeting.