Watch webinar on-demand
What Large Language Models Can — and Cannot — Do
Large language models have made it possible to work with unstructured legal text in ways that were impractical just a few years ago. Contracts are, by nature, dense and highly variable documents. LLMs are well suited to tasks that involve reading, interpreting, and restructuring that kind of content at scale.
In practice, this translates to capabilities like:
- Contract summarisation — producing a structured overview of key terms, obligations, and dates without requiring manual review of the full document
- Metadata extraction — identifying and pulling structured data points from uploaded or archived contracts
- Natural language search — finding contracts in your archive by asking questions in plain language, rather than building manual filters
Precisely has already shipped natural language archive search as a live feature. Users can type a question into the archive and Precisely generates structured metadata filters automatically — without sending any contract content to an external model.
What LLMs cannot reliably do is exercise legal judgement. They can flag potential issues, but they cannot assess risk the way a qualified lawyer can. They can compare clauses, but they cannot account for the commercial context behind a negotiation. May and Oscar were clear on this in the webinar: AI is a tool for acceleration, not a substitute for expertise.
Why Data Security Matters More Than You Might Expect
When evaluating any AI tool for contract workflows, the question of where your data goes is not a secondary concern — it is fundamental.
The majority of AI infrastructure is hosted in the United States or China. For businesses operating under EU data protection law, that creates an immediate compliance question. GDPR requires that personal data is processed lawfully and that transfers outside the EU are subject to appropriate safeguards. Contracts frequently contain personal data. Using an AI tool that routes that content through non-EU infrastructure may put you in breach of your obligations.
Precisely's approach is straightforward. All data is hosted in Europe. When AI features process queries — such as translating a natural language question into archive filter logic — they do so without exposing the underlying contract content, metadata values, or party information to any external model. The AI handles the logic. Your data stays within your environment.
For legal teams evaluating contract management software, this distinction matters. It is worth verifying the same point with any vendor you assess.
The EU AI Act: What It Means in Practice
The European Union's AI Act represents the most significant regulatory framework for artificial intelligence to date. It establishes risk categories for AI systems and places corresponding obligations on both developers and the businesses that deploy them.
For teams using AI in contract workflows, the practical implications include greater expectations around transparency — knowing what an AI tool is doing and on what basis — and accountability for the outputs it produces. The webinar covered how these requirements are likely to shape enterprise AI adoption over the coming years, and why building good governance habits now is a sensible position regardless of where regulation ultimately lands.
Precisely's product decisions reflect this direction. Features are designed to keep users informed and in control. AI suggestions — such as metadata validation recommendations — are presented for review and can be accepted, modified, or dismissed. Nothing is applied automatically without user sign-off.
Precisely's AI Roadmap: Where Things Are Heading
May and Oscar shared Precisely's thinking on how AI will develop within the platform. The current focus is on areas where AI adds clear, measurable value without introducing risk.
Metadata extraction and validation. When contracts are uploaded or completed, AI can suggest metadata values based on the document content. Users review and confirm before anything is saved. This keeps your contract archive accurate and reduces the manual effort of data entry — without removing human oversight.
Natural language search. Already live, this feature allows users to query the archive in plain language and receive structured filter results. It removes the need to know exactly which metadata fields to search, making contract retrieval faster for the whole team.
Contract summarisation. Providing structured overviews of key terms and obligations, so reviewers can orient themselves quickly before working through the detail.
Looking further ahead, the roadmap includes deeper support for contract review workflows, risk flagging, and more intelligent assistance throughout the drafting and negotiation process. The consistent principle is the same throughout: AI that augments the user's judgement, not AI that operates independently of it.
A Practical Starting Point for AI Adoption
The webinar closed with guidance for teams beginning to evaluate or introduce AI into their contract processes. A few principles stood out.
Start with low-risk, high-value tasks. Metadata extraction and archive search are sensible entry points. The stakes are lower than AI-assisted drafting, and the efficiency gains are immediate and tangible.
Audit your vendor's data handling. Before enabling any AI feature, understand where your data goes and what it is used for. Ask vendors directly. Check their documentation. For EU-based organisations, data residency is not optional.
Keep humans in the loop on substantive decisions. AI can prepare, summarise, and flag — but final decisions on contract terms, approvals, and risk assessments should remain with qualified people.
Build governance habits early. As the AI Act takes effect, organisations with clear internal policies on AI use will be significantly better placed than those starting from scratch.
For teams already using Precisely, AI features are available to enable in Organisation Settings. For teams still evaluating options, our platform comparison covers how Precisely stacks up against other CLM tools across security, compliance, and AI capability.
Frequently Asked Questions
Is the webinar free to watch?Yes. Enter your email above and you will receive immediate access to the full recording. No sales call required.
Do I need to be a Precisely customer to watch?No. The webinar is open to anyone with an interest in AI and contract management. Whether you are evaluating CLM software or already using a platform, the content is relevant.
What AI features does Precisely currently offer?Precisely currently offers natural language archive search and AI-powered metadata validation, both available to organisations with AI features enabled in Organisation Settings. Further capabilities are in development — see our product changelog for the latest updates.
How does Precisely ensure GDPR compliance when using AI?All Precisely data is hosted in Europe. When AI features are used, only the query logic is processed by the model. No contract content, metadata values, party names, or document titles are ever sent to an external system. All filtering and validation happens within your own environment.
