The Contract Was Hiding in Plain Sight

The most expensive sentence in a company may already exist. It is sitting on page 47 of a supplier agreement, beneath a heading nobody remembers, inside a folder owned by someone who left in March. It might promise a rebate. It might trigger an automatic renewal. It might cap liability at precisely the moment everyone assumed it did not. The sentence is not secret. It is merely hard to find, which in a sufficiently large organization is almost the same thing.

Workday Contract Management is built around this small bureaucratic tragedy. Its raw material is the agreement after the champagne, the DocuSign notification, and the relieved email saying the deal is done. Traditional contract software helped put that agreement in a digital cabinet. The technology Workday acquired with Evisort tries to turn the cabinet into a database.

The search box came before the empire

Jerry Ting had the germ of the idea as an undergraduate. At Harvard Law School, it sharpened. Lawyers kept hunting for clauses, precedents, dates, and party names by hand. Ting originally imagined a litigation tool, but conversations with professors, alumni, and working legal teams pointed him toward transactional work. In 2016 he joined forces with fellow law student Jake Sussman and MIT-trained data scientist Amine Anoun. The name Evisort was a contraction of “Evidence Sort.”

The founders did something refreshingly unmagical: they contacted more than 200 corporate general counsel. The answers narrowed a grand ambition into an ordinary complaint. Companies had software that stored contracts but did not understand them. People still typed effective dates and clause names into spreadsheets. Search was poor. Institutional memory depended on whoever happened to know which folder to open.

“What’s the point of a management system that’s just storing the document but not managing it?”Memme Onwudiwe, founding team member

That observation became the wedge. Evisort trained models on contract language so an uploaded agreement could become structured information without a paralegal first tagging every field. It connected to places companies already kept files, including SharePoint, Box, Google Drive, Dropbox, and Salesforce. The file could stay put while the data moved. That matters because enterprise software projects often fail at the point where a sensible new system demands an heroic old-system migration.

Two products, one pile of promises

Today Workday sells two related products. Contract Intelligence is the reading layer: a repository, extraction engine, search tool, natural-language question interface, custom AI model builder, and dashboard system. A company with an older CLM can put this layer on top. Contract Lifecycle Management adds the making of agreements: request forms, approved templates, clause libraries, drafting, version comparison, approvals, e-signature, obligation tracking, renewal, and reporting.

The distinction is useful. Some organizations do not need another workflow transformation. They need to know what is inside 80,000 documents by next quarter. Others need the entire route from “please hire this vendor” to “we should renegotiate before November.” Workday’s pitch is that the same intelligence should travel through both.

The users are not only lawyers. Procurement searches for discounts and supplier obligations. Finance looks for payment schedules. Sales wants fewer stalled approvals. HR has employment agreements. Compliance needs evidence. M&A teams need to know which counterparties can object when ownership changes. The competitive set includes Icertis, Ironclad, DocuSign CLM, LinkSquares, Sirion, ContractPodAi, Agiloft, and Conga. Workday’s particular advantage is the chance to connect words in agreements with the people, suppliers, and financial records already living in Workday.

The customer became the buyer

Evisort raised about $154.5 million across its major disclosed rounds, including a $100 million Series C in 2022. It sold subscriptions to more than 150 businesses by that year. Named customers included Microsoft, McKesson, BNY Mellon, NetApp, Vonage, Jelly Belly, and Workday. Then the case study became corporate strategy.

Workday’s procurement operation manages roughly $2 billion in addressable annual spend, about 4,000 active suppliers, and 6,000 new contracts a year. Before Evisort, its team described losing thousands of hours to searches and data cleanup. After adoption, Workday says it increased contracts under management from about 30,000 to more than 45,000, cut search time by 75 percent, and unlocked $21 million in cost avoidance and value creation over three years. The reported return was 77 times the investment.

Workday’s figures describe its own global procurement deployment. They are a case study, not a universal forecast.

In September 2024, Workday announced it would acquire Evisort. The purchase price was not disclosed. That is the honest answer to what the deal cost. The more interesting answer is what Workday believed it was buying: a way to connect unstructured contractual language to its structured finance and HR platform. By March 2025, Evisort-powered Contract Intelligence and Contract Lifecycle Management were available through Workday.

The agent needs a rulebook

The newest move is from extraction to action. Workday’s Contract Negotiation Agent can review a whole inbound agreement against a company playbook, flag language outside policy, and propose redlines. Administrators can define preferred clauses, acceptable fallbacks, and walk-away positions. In 2026, Workday expanded the idea toward full-document review, AI-assisted intake, playbook creation, bundled e-signing, amendments, renewals, and multi-document workflows.

Here is the catch: automation becomes useful only after the organization has decided what “good” means. A model cannot rescue a company that has no approved language, no owner for exceptions, and no shared view of risk. Poor scans, inconsistent permissions, and exotic document structures complicate extraction. Some adjacent Workday agents accept only particular formats and languages. Legal and financial commitments still deserve human review.

This is also where a small team should resist the glamour. If a business signs a few dozen straightforward agreements a year and one careful person can find them, enterprise CLM may create more governance than it removes. The economics improve with volume, dispersion, repetition, and expensive mistakes. A system that can inspect 450,000 documents in a day is impressive. A company with 450 documents has a different problem.

The machine can propose the redline. The company still has to know where its red lines are.

What is worth copying

The transferable lesson is not “buy more AI.” It is the sequence. Start with the dull task people repeat. Interview enough users to discover whether their complaint is local or structural. Meet the documents where they already live. Make the first result searchable and auditable. Prove value on a real corpus. Then widen the workflow.

Evisort did not begin by promising an autonomous legal department. It began with a search problem. Workday did not encounter it through a slide deck alone. Its own procurement group lived with the product, built an integration, and measured the result. The company that once sold “Google for legal contracts” now occupies a stranger position: part librarian, part traffic controller, part junior negotiator.

The real product is corporate memory with an alarm clock. It notices the promise on page 47, attaches it to the supplier, and tells the right person before November. In a world stuffed with generated language, there is something almost comic about the value of simply remembering what everyone already signed.