AI-Assisted Contract Review for a Global Licensing Portfolio

The company is one of the world's largest pharmaceutical manufacturers, headquartered in continental Europe with research facilities, manufacturing plants, and commercial operations spanning more than sixty countries. Its product portfolio includes branded prescription drugs, biosimilars, generics, and consumer health products, distributed through a web of licensing agreements that has been built, amended, extended, and renegotiated over more than two decades. The licensing portfolio — over three hundred active agreements governing the development, manufacture, and distribution of products across global markets — is the legal infrastructure on which much of the company's revenue depends.

No single person in the organisation understood the full portfolio. This was not a failure of competence but a consequence of scale and history. The agreements had been negotiated by different teams, in different regions, at different times, under the supervision of different general counsels. The European licensing team managed agreements with partners across the EU and UK. The Asia-Pacific team handled agreements with manufacturers in India, China, and Southeast Asia. The Americas team oversaw licensing relationships across North and South America. Each team had its own filing conventions, its own contract management practices, and its own institutional memory. Cross-regional visibility was limited to what could be assembled manually — and manually assembling a view of three hundred agreements was, as the head of legal operations put it, 'a project, not a query.'

The company had long suspected that the portfolio contained inconsistencies. Licensing agreements negotiated fifteen years ago under different standard terms coexisted with agreements signed last year under current templates. Some agreements had been amended multiple times, with each amendment modifying different provisions. Territorial overlaps were a known risk: the company had experienced at least one dispute in which two licensees claimed rights to the same market under different agreements. Change-of-control provisions varied widely — some agreements gave the counterparty termination rights upon a change of control, others required only notification, and others were silent on the subject entirely.

A manual review of the full portfolio had been proposed and rejected multiple times over the previous five years. The scope was simply too large. The head of legal operations estimated that a thorough manual review — reading each agreement, extracting key terms, comparing them against current standards, and flagging deviations — would require a team of six to eight contract lawyers working full-time for four to six months. The cost, including external counsel rates for specialised pharmaceutical licensing expertise, was estimated at well over a million euros. And the result would be a static snapshot: by the time the review was complete, new agreements would have been signed and existing ones amended, and the analysis would already be falling out of date.

The head of legal operations had been following developments in AI-assisted contract analysis for several years. She had attended demonstrations of various contract review platforms and found most of them impressive on simple contracts but inadequate for the complexity of pharmaceutical licensing agreements. These are not standard commercial contracts — they contain development milestones, royalty escalation clauses, territory-specific sublicensing rights, regulatory approval obligations, pharmacovigilance requirements, and intellectual property provisions that interact with each other in ways that even experienced licensing lawyers sometimes struggle to parse. She was sceptical that any AI tool could handle this complexity meaningfully.

Her scepticism was tempered by pragmatism. The manual alternative was impractical, and the status quo — not knowing what was in the portfolio — carried risks that the company could no longer afford to ignore. The company was in active discussions about a potential acquisition by a larger pharmaceutical group, and the acquiring company's due diligence team would inevitably ask about the licensing portfolio's change-of-control provisions. Not being able to answer that question quickly and accurately was, the general counsel observed, 'a problem that has moved from theoretical to urgent.'

The team selected JuristIQ after a competitive evaluation that included three other AI contract analysis platforms. The evaluation was rigorous: each platform was given a test set of twenty licensing agreements — chosen to represent the full range of complexity in the portfolio — and asked to extract key terms and identify deviations from the company's current standard clause library. JuristIQ's performance was not perfect, but it was substantively better than the alternatives on the most complex agreements, particularly in its handling of multi-layered territorial provisions and conditional royalty structures. The licensing lawyers who reviewed the test results noted that JuristIQ's extractions read like the work of a competent junior lawyer, not a keyword search engine.

The full portfolio review began with a two-week configuration phase. The company's legal operations team worked with Juristic to build a clause library reflecting the company's current standard terms for each major provision: territory, royalty rates, development milestones, termination rights, change-of-control, audit rights, sublicensing, pharmacovigilance, intellectual property ownership, and regulatory obligations. Each standard clause was defined with acceptable variations and deviation thresholds, allowing JuristIQ to distinguish between minor language differences and substantive departures from the company's preferred terms.

The team then uploaded the full portfolio — 312 agreements, including all amendments and side letters — into Juristic. JuristIQ processed the documents over a period of three days, analysing each agreement against the clause library and producing a structured output for every provision in every agreement. The results were presented in a dashboard interface where the legal team could filter by clause type, deviation severity, region, counterparty, and business unit. Each flagged deviation was linked to the specific language in the agreement, allowing lawyers to click through from the dashboard to the relevant paragraph without searching through the document manually.

The findings were more significant than anyone had anticipated. JuristIQ identified forty-seven agreements with non-standard change-of-control provisions — the exact question the acquiring company's due diligence team would ask. Of these, nineteen contained provisions that gave the counterparty an outright termination right upon a change of control, which could put material revenue streams at risk in an acquisition scenario. The remaining twenty-eight had various intermediate provisions — consent requirements, renegotiation triggers, or enhanced termination rights that activated only if the acquiring company was a competitor of the licensee.

Beyond the change-of-control findings, JuristIQ flagged twenty-three agreements where audit rights had been narrowed beyond the company's standard terms — in some cases limiting the company's ability to verify royalty calculations to once every three years rather than annually. Twelve agreements contained territorial restrictions that appeared to conflict with more recently signed deals in overlapping markets, a finding that prompted an immediate review by the regional licensing teams. Eight agreements contained development milestone provisions that had been triggered but not formally acknowledged, creating potential disputes about milestone payments that had not been invoiced.

The legal team spent four weeks reviewing and validating JuristIQ's findings. The review confirmed the vast majority of the AI's extractions and flagged deviations — the false positive rate was approximately eight percent, primarily in cases where the AI had flagged stylistic differences as substantive deviations. The licensing lawyers who conducted the validation were, by their own account, impressed. One senior lawyer, who had been vocally sceptical of the AI approach, noted that the tool had identified three territorial overlaps that she had not been aware of despite having managed the relevant regional portfolio for over a decade. 'It is not that I was negligent,' she observed. 'It is that the volume makes it impossible to hold the full picture in your head. The tool holds the full picture.'

The practical impact of the review was immediate. The company's M&A team received a comprehensive briefing on the change-of-control provisions across the portfolio, enabling them to model the potential impact on licensing revenue under various acquisition scenarios. The nineteen agreements with outright termination rights were prioritised for renegotiation, and the legal team began approaching counterparties to discuss amendments — conversations that, in several cases, the counterparties welcomed because they too had been uncertain about the implications of their own provisions.

The territorial overlaps identified by JuristIQ led to three internal restructuring discussions and one renegotiation with an external licensee. The narrowed audit rights were flagged to the finance team, which adjusted its royalty verification schedule for the affected agreements. The untriggered milestone provisions resulted in the company issuing invoices for approximately four million euros in development milestone payments that had been earned but not billed — a direct financial recovery attributable to the review.

The head of legal operations estimated that the AI-assisted review accomplished in six weeks what would have taken six months manually — and at a fraction of the cost. But she is careful to note that the time and cost savings, while substantial, were not the most important outcome. 'The real value,' she explained, 'is that we now know what is in our portfolio. For the first time in the company's history, we can answer questions about our licensing agreements at a portfolio level, not just one agreement at a time. That changes how we make decisions.'

The portfolio analysis has since become an annual exercise, with each review building on the baseline established by JuristIQ's initial pass. New agreements are analysed as they are signed, and the dashboard is updated continuously. The legal operations team runs quarterly reports on portfolio-level metrics — average royalty rates by region, prevalence of non-standard provisions by clause type, upcoming termination dates — that inform both the legal team's priorities and the commercial team's negotiation strategies.

The general counsel, reflecting on the project at a company leadership meeting, described it in terms that resonated beyond the legal department: 'For twenty years, our licensing portfolio was a filing cabinet. We put agreements in, and we took them out when we needed them. JuristIQ turned the filing cabinet into a managed asset. We can see it, analyse it, and make strategic decisions about it. That is a fundamentally different relationship with our own contracts, and it has made us a better-run company.'

The company is now exploring the use of JuristIQ for its clinical trial agreement portfolio — a separate body of approximately one hundred and fifty agreements with contract research organisations and clinical sites — where similar complexity and similar risks of hidden inconsistencies are believed to exist. The head of legal operations expects the second portfolio review to proceed faster than the first, building on the clause library and analytical framework already established. 'The infrastructure is in place,' she noted. 'Now we just need to point it at the next problem.'