The university is rethinking how contracts are understood and governed—moving away from static records towards living intelligence that helps people make confident, timely decisions. Built on AWS, the approach combines AI‑driven classification with human judgement to reduce manual effort, standardise processes, and establish a strong foundation for procurement insight.
A turning point in how the university works with its agreements.
Like many large institutions, the university has built up an extensive library of agreements over time. Standard templates guide how new contracts are created and awarded, while tens of thousands of historic and active agreements reflect how those templates have been adapted, amended, or departed from in practice. Gaining visibility into where those changes occur, which clauses are routinely modified, and where risk tends to sit has traditionally required time‑consuming manual review across multiple systems and formats.
An earlier phase of work had already delivered real progress. Contract storage was standardised, key metadata was extracted from around 25,000 agreements, and teams were better able to answer basic questions about expiries and supplier relationships. That work demonstrated two things. First, there was clear value in treating contracts as data, not just documents. Second, it opened the door to a deeper level of insight—one that could group contracts by their closest template, identify how and where clauses diverge, and put those insights directly into the hands of the people who need them.
This was not about replacing existing systems. Records repositories remain essential for compliance, but they are not designed to classify, compare, or interact with contracts at scale. The university's ambition was to complement that environment with contract intelligence—extracting key facts, surfacing obligations, linking related agreements and supporting documents, and enabling people to ask natural‑language questions and receive clear, auditable answers. With a solid digital foundation in place, the timing was right to move forward.
An AWS foundation where AI accelerates work and people guide outcomes
The next phase is anchored on AWS, turning a static archive into a dynamic, searchable source of insight. The programme progresses in deliberate steps, pairing machine intelligence with human oversight to build trust and ensure outcomes remain aligned with policy.
The first phase focuses on large‑scale mapping. Using a retrieval‑augmented approach with re‑ranking, the solution analyses each contract's structure and content to determine its closest matching template. This provides a clear, evidence‑based view of where the contract estate aligns with standard terms and where it diverges. The workflow is automated end to end, with exception handling and audit trails, delivering high mapping accuracy while keeping validation work light but meaningful for reviewers.
The second phase moves to clause‑level comparison against the matched template. Here, AI is used to speed up the identification of differences, maintain transparency around why a match was made, and ensure people remain in control of decisions. The focus is not on experimentation for its own sake, but on repeatability—making it easy for legal, procurement, and category teams to see which clauses are frequently amended, which changes introduce risk, and where templates or guidance should be updated to reduce friction in future negotiations.
Sitting above this analysis is an experience designed around how people work. The experience layer brings chat, search, clause review, and analysis together in a single interface. Users can move seamlessly from a question—such as "What's our position on termination for convenience in supplier X's contracts?"—to the relevant documents and structured data without switching tools. The design also supports practical integration with procurement analytics, such as spend data, so contractual and commercial insight can be considered together when decisions matter.
Just as importantly, the platform is designed to be operated over the long term. A defined service model covers monitoring, logging, model tuning, and reporting, allowing classification quality and response accuracy to improve over time. Rather than a one‑off uplift, the university is building a sustainable capability—using Microsoft as a stable foundation for application hosting, security, and observability, alongside a governed, human‑in‑the‑loop process that strengthens outcomes as adoption grows.
From one-off insight to an ongoing advantage
The most immediate benefit is speed to clarity. Questions that once required searching through folders can now be answered in minutes, with supporting evidence—how closely a contract aligns to a template, which clauses depart from the standard, when key dates fall, and how obligations apply across a supplier's portfolio of agreements. This doesn't replace expert judgement; it preserves it for where it adds the most value, shifting effort from manual search to focused analysis.
The second benefit is consistency. By mapping the contract estate against a library of around 350 templates and examining clause‑level variation, the platform exposes patterns that were previously anecdotal. Templates that are consistently amended can be updated to better reflect how the university contracts. Clauses that regularly trigger negotiation can be refined or supported with clearer guidance. Governance becomes data‑led—grounded in real behaviour, not just documented intent.
The third benefit is foresight. With contract intelligence in place, the university gains flexibility. Linking spend and performance data into the experience layer connects commercial reality with contractual position, helping teams see when supplier performance or rate compliance should prompt action—and which contractual provisions apply. Over time, the same architecture can support adjacent use cases, such as targeted amendments, proactive communications when regulations change, or automated checks that flag exceptions before they become delays.
Underlying all of this is a simple principle: technology should help people do their best work. By building on Microsoft and committing to a managed service model that monitors, tunes, and explains outcomes, the university is creating a durable capability that supports security, builds trust, and compounds value over time. Records repositories continue to do what they do best—preserve and evidence. Contract intelligence now does what is required—help people understand, manage, and act.
This is transformation on the university's terms: practical steps, visible benefits, and a clear path forward. It starts with understanding what exists and comparing what matters. It continues by putting insight in front of the people making decisions every day. And it prepares the institution for a future where contracts are not just stored, but actively used to deliver better outcomes for students, staff, suppliers, and the wider community.
