Quick summary
Signing a license is a win, but the revenue depends on what happens next: getting every term, milestone, and obligation into the system of record. Agreement AI in Evolve reads an executed agreement and extracts its parties, dates, and financial and non-financial terms for staff to review. Here’s how it works, what it catches, and how it connects to licensee compliance through the Partner Portal.
Every executed license gets its moment. The signatures land, congratulations go out, and somebody probably brings cake. Well, let’s not all start high-fiving just yet.
The agreement then goes into a queue. A 17-page PDF waits for someone to turn it into the dates, dollars, and obligations a tech transfer office’s system can track. That handoff is where revenue quietly slips away. The deal was fine. The milestone buried in section 3.1(b)(ii) just never made it into the record.
Agreement AI in Evolve closes that gap. Here’s what it does, how it works, and why it matters long after the cake is gone.
Where Agreement Entry Breaks Down
The trouble usually starts after the deal is done. Over the years, tech transfer offices have told us about the same four problems:
- Finding the terms. Someone has to read a 20-page document and decide which terms belong in the system.
- Losing context. The person entering terms might not have been part of the negotiation. They didn’t see the back-and-forth, so they work from the document alone.
- Missing triggers. Non-financial milestones often drive financial obligations. If the milestone isn’t in the system, nobody knows payment is due.
- Missing revenue. That’s where the first three problems end up. Signing an agreement is a win. Compliance is what brings in the revenue afterward.
How Agreement AI Works
A user uploads the executed agreement, and Agreement AI reads it. It pulls the title and parties, then extracts the terms. Each term lands side by side with its source, so users can review it, edit it, and decide what goes into the record. Extraction runs in the background and takes about two minutes for a typical agreement. Each document shows its status, and a partial or failed extraction can be retried.
The review step is deliberate. Agreement AI uses a human-in-the-loop approach: the AI finds the terms, but a real person makes the call. It’s there to help you do the job, not to make the final judgment.
The title and parties aren’t the hard part. The terms are, and Agreement AI captures four details that make a term usable:
- Recurrence: whether a term repeats annually, quarterly, or on some other schedule.
- Start dates: the anchor for all the date math agreements love, such as “60 days after the first anniversary of the effective date.”
- Section and clause references: where each term came from, including nested sub-clauses rolled up into one reference.
- Clause text: the actual language, so nobody has to copy and paste paragraphs by hand.
What It Pulls From a Real License
Take a typical 17-page exclusive license. Here’s what Agreement AI pulls from it:
| Term | What the Agreement Says | What Lands in the Record |
|---|---|---|
| License maintenance fee | Section 4.3: $150,000 within 60 days after the first anniversary of the effective date, and every anniversary after | A recurring annual payment with its start date and clause text |
| Milestone payments | Section 4.4: a table of events, each paid within 30 days of occurring | A separate term for each milestone, read straight from the table |
| Royalty | 3.5% on net sales in the first tier | One term per tier, with the parent section and sub-clause combined |
| Equity | Shares equal to 5% of issued and outstanding stock, due on the effective date | A non-financial term to confirm the office received evidence of equity |
| Company financing | Section 3.1(a): raise a set amount by the third anniversary | A compliance term to monitor |
| Sublicensing | 25% of all sublicense income | A revenue term tied to sublicense activity |
| Progress report | Section 3.1(b)(ii): a written report within 60 days after the start of each calendar year | An annual reporting obligation with a start date |
The progress report shows why section references matter. Agreement AI pulled from three nested levels and rolled them into one reference. When users need to know where a term came from, the answer is right there.
Handling the Messy Parts
Real agreements don’t stay tidy. Here’s how Agreement AI handles the tough cases:
- Stepped payments. If a licensee pays one amount in year one, another at a milestone, and a third annually after that, Agreement AI creates three terms: two one-time and one recurring.
- Complex constructions. When a term sits several levels below its main clause, Agreement AI pulls both and wraps them together.
- Tables. Milestone schedules in table form become individual terms.
- Missing section numbers. If Agreement AI can’t find a section number, it says so, which is a useful flag on its own.
Built for Confidential Data
Tech transfer agreements are full of confidential information, so data protection is critical. TTOs keep control over which AI models operate within Evolve and the safeguards around them. Your confidential data never leaves the institution’s tenant and never trains an external model. And because Evolve is a Salesforce OEM product built for tech transfer, it inherits Salesforce’s security model, including role-based permissions that control who can use which AI functions.
From Signed to Compliant
Agreement AI gets the terms in. The Evolve Partner Portal keeps them moving.
Licensees can log in to see their invoices and obligations and complete required reports right in the portal. Offices can set up a standard set of questions for a type of obligation, such as an annual progress report, and licensees answer them online. Nobody emails a Word document, and nobody in the office retypes it.
The same goes for products. When licensees report products built on the institution’s technology, along with their sales, that data flows into Evolve for royalty calculations and iEdison reporting.
Catch Every Term
Agreement AI extracts the terms from your executed PDFs in minutes for your team to review. Less data entry, more time managing the deal.
Agreement AI is one of four features in Evolve AI, the AI add-on module for Evolve. These run natively in Evolve, on the records your office already keeps, and each one produces a first draft that your staff review, edit, and act on.
- Agreement AI reads an executed agreement and suggests its title, effective date, parties, and terms for review.
- Assessments AI generates a first assessment of a new technology’s patentability and commercial potential on the asset record, so staff have a starting point at triage.
- Grant Guidance AI generates commercialization guidance for researchers from an award’s abstract, so your office can engage faculty while the research is still under way, before a disclosure exists.
- Potential Partner AI identifies companies that could license or partner on a technology your office is marketing, and saves the list on the asset record.
All four leverage the AI model your institution already licenses (e.g., OpenAI, Google Vertex AI, Amazon Bedrock), under your own provider agreement. Your data is never used to train Evolve AI and adheres to your security rules.
Signing the license is the start. Agreement AI gets its terms into Evolve in minutes, ready for your team to review.
If you want to see Agreement AI in action, you can book a demo now.
Frequently Asked Questions
What is Agreement AI?
Agreement AI is an Evolve capability that reads an executed agreement and extracts its title, parties, dates, and financial and non-financial terms, so users review terms instead of keying them in by hand.
What else is included in Evolve AI?
Evolve AI includes four features: Agreement AI, Assessments AI, Grant Guidance AI, and Potential Partner AI. Each runs inside Evolve on your institution’s own AI model, and each produces a draft for staff to review. New features added to Evolve AI are included in the add-on when they’re released.
Does Agreement AI enter terms automatically?
No. It uses a human-in-the-loop approach. Agreement AI finds the terms and shows each one beside its source, and users decide what goes into the agreement record.
What kinds of terms can Agreement AI extract?
It extracts financial terms such as license maintenance fees, milestone payments, royalties, and sublicense income, plus non-financial terms such as equity, financing obligations, and progress reports. Each term includes its recurrence, start date, section reference, and clause text.
Can Agreement AI handle tiered or stepped payments?
Yes. A stepped payment becomes separate terms. For example, a first-year amount, a milestone amount, and an ongoing annual amount become two one-time terms and one recurring term.
Is agreement data safe with Agreement AI?
Yes. Customers control which AI models operate within Evolve, and your confidential information never leaves your IT-approved systems or trains an external model. Evolve also inherits Salesforce's security model, including role-based permissions.
How does Agreement AI support licensee compliance?
Agreement AI gets every term and milestone into Evolve. The Partner Portal then lets licensees see their invoices and obligations and submit reports in the app, so obligations don't slip after the agreement is signed.