For law firms, AI conflicts clearance should be a competitive advantage. Instead, a conflicts analyst at a mid-size firm spends hours clearing a single matter. She reconstructs the corporate tree by hand. She chases intake forms. She cross-references Outside Counsel Guideline (OCG) spreadsheets that may or may not reflect the latest client amendments. By the time the matter is cleared, the firm has done the work correctly. The client has been waiting since Monday.
Across town, a competitor cleared the same matter before lunch.
The analyst didn’t do anything wrong.
The job that got rewritten
Every step of that process needs to happen. The question is who, or what, does it.
Reconstructing a corporate tree is data assembly. Chasing intake completions is project coordination. Cross-referencing spreadsheets against outside counsel guidelines is version control. The analyst was hired to find the thing that isn’t obvious, weigh the risk, and make the call. Somewhere along the way, the platform’s limitations rewrote the job description. And because every firm’s platform had the same limitations, the industry normalised it.
Why layering generic AI makes it worse
A general-purpose model doesn’t know what a conflict check is. It doesn’t understand the difference between a corporate affiliate and a conflicting party, what an ethical wall requires, or how matter relationships sit inside a law firm. It moves faster, and faster results built on incomplete domain knowledge carry real consequences.
The EU AI Act is the first and most comprehensive standalone legislation specifically governing AI and regulatory development in AI governance globally. Other jurisdictions are watching it and responding in their own way: some are developing dedicated frameworks, others are slotting AI obligations into existing legislation. Australia is a case in point, its Privacy Act reforms, coming into effect December 2026, will require firms to disclose where AI makes or substantially assists in decisions that affect individuals. The instruments differ. The direction doesn’t. Wherever you operate, the accountability sits with the firm deploying the AI.
There is also the difference between personal productivity and firm productivity. A general-purpose AI can help an individual analyst work faster. It cannot enforce the ethical walls, jurisdictional rules, and cross-practice governance obligations that a conflicts decision requires, particularly for firms operating across regions or practice groups where those obligations differ.
Most firms discover this after the pilot. MIT’s 2025 research found 95% of enterprise AI initiatives deliver zero measurable return. S&P Global puts the abandonment rate at 42% of all AI projects in 2025, more than double the year before. The firms that stalled didn’t fail because the technology let them down. They entered production without the foundation it requires: governance that maps to how the firm actually operates, and a data model that understands what a matter, an ethical wall, and a cross-practice conflict actually mean.
What the shift actually looks like
The compliance leaders pulling ahead have made a specific architectural choice: AI handles the assembly in a governed environment, humans handle the judgment with full context.
By the time a question reaches the analyst, the corporate tree is already mapped, the search has already run, and results are triaged by relationship type and risk level. What used to take a morning takes minutes, not because the analyst works faster, but because the preparation is done before they open the matter.
Client obligations follow the same logic. Instead of manually cataloguing OCGs against a spreadsheet that lags every client update by weeks, obligations are surfaced automatically and tracked consistently across matters.
This is also the answer to the question every compliance leader eventually asks: what happens when the AI gets it wrong? For intake, conflicts, and terms, full autonomy isn’t the right model. The right design keeps humans accountable for decisions while removing the manual preparation that was making those decisions slow.
Intapp Compliance with Celeste is built on that design. AI agents run across intake, conflicts clearance, and client obligations management using a data model built specifically for how law firms structure matters, relationships, and risk, inside existing workflows, so the firm doesn’t have to choose between modernising and keeping the team operational while it does.
What C-level conversations on this tell me
The compliance leader who makes this case internally needs language their managing partner will act on. When I sit with managing partners and COOs, the conversation usually moves quickly from “does it work” to three bigger questions.
Speed and capacity move together. Faster clearance means more matters through the same window. But the more durable gain is that analysts shift from data preparation to judgment work. The firm gets both: throughput goes up, and the quality of the analysis improves because the analyst is doing what they were actually hired to do.
Intake quality compounds downstream. When AI handles intake follow-up automatically, flagging incomplete submissions before they hit the queue, searches run on cleaner data. Fewer matters reopened. Fewer escalations. Fewer write-offs. Fewer billing disputes. That downstream effect often exceeds the speed gain and it’s where the hard savings show up.
When compliance runs cleanly, the firm has a platform to grow from, not a queue to manage. Compliance teams currently absorb demand spikes by stretching permanent staff or bringing in temporary cover. When the platform handles the preparation work, that pressure has somewhere to go. And when compliance data is reliable and real-time, managing partners can see which practice groups have capacity, where intake is creating friction, and where the firm has room to grow without adding risk. Managing partners who run compliance on a modern platform stop asking how to manage demand. They start asking how much more the firm can take on.
The transition is the real test
The concern most compliance leaders carry into this conversation isn’t whether the technology works. It’s whether the people and the data will move with it.
Data migration is the visible risk. Conflicts systems carry years of matter history and obligation records, a migration that loses fidelity, even briefly, creates the exposure it was meant to prevent. What runs in parallel, what the rollback plan looks like. These are the questions a production-ready platform should answer without hesitation.
Change management is equally high on the list. Analysts adopt a new platform when it fits how they actually work and feels more intuitive to use. The firms that get this right map both analysts’ and firm workflow, demonstrate where the friction disappears, and build the transition around that. Intapp Compliance with Celeste is built to answer all of this before the migration begins.
The compounding cost of waiting
The capability gap between firms running AI-native compliance and firms still on legacy platforms widens every quarter. A competitor clearing the same matter in hours rather than days doesn’t just win that matter faster. Over time, they accept more work, deepen client relationships, and make the case to their own managing partners that compliance is a growth function, not a cost centre.
See how Intapp Compliance with Celeste helps your firm accept more business, faster — without adding headcount or compromising risk.