Our session at the Buying Legal Conference landed late on day two, a busy stretch with plenty pulling people in other directions. It turned out to be one of our favorite conversations of the week, so we’re sharing the ideas that stood out.
Priori CEO and Co-Founder Basha Rubin moderated a panel with Allyson Sullivan, U.S. Associate Director of Legal Operations at Ferring Pharmaceuticals, and Deborah Koplovitz, Vice President and Associate General Counsel at Allstate. The session centered on a question that comes up in nearly every client conversation right now: If AI is changing how legal work gets done, how do you determine what that work should cost?
1. The paradox is real
Big Law demand grew 4.2% in the first half of 2026, well above a historical norm of about 1.5%, while revenue grew 11.7% over the same period.¹ Everyone expects AI to change the economics of legal work, yet almost no one can show where it has.
Waiting for the market to sort that out on its own (“hope chess,” as it came up on the panel) isn’t much of a plan. The departments making progress are building their own evidence.
2. Invoices can reveal more than compliance
Invoice review today centers on compliance: approved timekeepers, correct rates, and proper task codes. The same invoices can also show whether work was efficient, properly staffed, or AI-assisted.
The answers live in billing narratives, the free-text descriptions every timekeeper writes a little differently. LLMs can now read years of them and pull out structured signals, such as what drove the cost and whether seniority fit the work. One panelist built a working analysis agent in about an hour and a half.
3. Scope is a data asset
RFPs already capture scope in detail: complexity, expected discovery volume, local counsel needs, and the assumptions behind the budget. Connected to the invoices that follow, that record tells you whether a budget overrun is a rate problem or a scope problem.
Picture two matters that each come in 30% over budget. In one, scope held, so the conversation is about rates and efficiency. In the other, custodians tripled and a third jurisdiction surfaced, so the conversation is about scope. Structured intake shows you which one you’re having, and every RFP becomes a row in your own benchmark database.
4. Should cost starts with did cost
Your own eBilling history is the most defensible starting point for pricing the next matter. Map billing narratives to consistent units of work, match comparable matters on complexity and volume, and benchmark the actual cost of each unit.
One adjustment matters most: past invoices reflect how work was done before AI, so each unit of work should be recalibrated for what AI now makes possible.
Get the slides
The slide deck from our Buying Legal Conference session covers each point in more depth, including the full framework for moving from did cost to should cost and five ways to bring AI into your firm relationships. Fill out a quick form and we’ll send it your way.
Working through any of this with your own team? We’d love to show you how other legal departments are putting these ideas into practice.
¹ Citi Global Wealth at Work Law Firm Group, first-half 2026 law firm financial results (August 2026).
