AI Readiness for In-House Legal Teams: From Pilot to Operations

Most in-house legal teams have already tried AI. According to FTI Consulting and Relativity’s General Counsel Report, 87% of general counsel now report generative AI use within their teams, compared with just 44% a year earlier. It’s rarely a lack of willingness to experiment that’s missing. What’s rare is already having a plan for what happens after the pilot.

8am’s 2026 Legal Industry Report found that nearly seven in ten legal professionals now use generative AI for work, while more than half of their organisations provide zero formal training. That gap, between individual adoption and organisational readiness, is where most legal AI initiatives quietly stall: someone on the team is using ChatGPT for research, a handful of contracts have gone through a trial tool, and nobody can say with confidence what data went where.

This is a practical readiness check, not a tool comparison. The question that matters here: is your team ready to use AI not just in individual chat sessions, but at the process level, as automation? If you’re looking for a side-by-side of specific platforms, see our comparison of legal automation tools instead.

What an AI readiness check for in-house legal teams should cover

“Readiness” stays abstract until it’s translated into something concrete: something a legal operations manager can actually use as a reference point. Four steps to start:

An honest audit of where things actually stand. Before evaluating a single new tool, find out what your team is already using, even informally. You’ll almost always find tools in use in one team that nobody else knows about, and data flows nobody signed off on. This audit matters for everything that follows.

A clear governance policy. It doesn’t need to be exhaustive. It needs to state which tools are approved, what data can and can’t be used with them, who reviews outputs before they reach a client or business unit, and what happens when something goes wrong. It doesn’t need to be perfect from day one either; better to keep refining it than to have no governance at all. Clarity is what matters here.

Two workflows, not twenty. The most common mistake is trying to automate everything at once. Pick two high-volume, rule-based processes; contract review and NDA generation are the usual starting points because the rules and review criteria are clear for most cases and the current manual effort is easy to measure. Our Legal Use Case Canvas is a free one-page framework for structuring exactly that choice of use cases.

A yardstick for measuring success. Measure a baseline before you start automating. If you can’t say how long a contract review takes today, or how many NDAs your team generates a month and at what time cost, you have nothing to compare the automated version against. Take the time to do this.

Why a workshop is worth it

Management and staff alike are often tired of hearing the word “workshop.” Too many of those hours have been wasted on someone talking at a room. But when it comes to your team’s AI readiness, a well-run workshop is worth real value, whether it’s led externally or organised with internal experts. In a few hours, you get everyone who matters around one table. Consider inviting other departments too, even just for part of it. The goal: work through the four points above:

• Map which tools are already in use
• Identify two or three initial processes for automation
• Capture baseline numbers you’ll use to measure the automation’s success later
• Involve everyone affected, and take the fear of AI and job loss off the table

How to choose AI tools for an in-house legal team

Once you’ve identified the team’s biggest pain points and the first automation opportunities, tool selection gets easier too. The answers to the following questions are what matter:

Is the mechanism behind it visible? If a tool classifies a contract as low risk, can you see exactly which criteria led to that classification? If it flags a clause for review, can you trace the reasoning? A tool that can only say its output “generally works well” is not something you can explain to a regulator or a client.

Is it hosted within the EU or EEA? This is a baseline GDPR requirement for handling client and company data. Confirm where the provider actually hosts data rather than assuming a general privacy policy covers it. Also check for recognised certifications such as ISO 27001.

Can your own team customise and adjust it? If every process or regulatory change needs a developer ticket with the vendor, the software gathers dust fast. No-code platforms that let legal staff build and refine automations directly remove that dependency; see how this plays out in practice in our breakdown of legal intake automation.

The same logic applies to choosing an AI assistant over a general-purpose chatbot: does it operate inside defined boundaries, or is it a free-roaming agent that happens to be good at sounding confident? For a regulated function, the boundary matters more than the fluency. Our comparison of e! by Lexemo against Microsoft Copilot walks through exactly what sets a legal automation tool apart, for teams already working inside Microsoft 365.

First automation steps: contract review or legal intake are the natural starting points

NDA generation and contract review are the standard starting points for AI and automation in legal teams, and for good reason: they’re repetitive, the review criteria are clear, and volume is usually high enough to show results within weeks. Compliance checks and routine risk assessments follow the same pattern and can be built as guided workflows that other departments, like sales and HR, use directly, while legal keeps control over every template and every approval or sign-off. If intake and routing themselves are the bottleneck, before a contract is even reviewed, that’s worth solving first; our guide to legal intake automation covers that specific workflow.

Governance you can prove

Regulation around AI use in legal contexts is tightening across the EU, and governance can’t be something you bolt on right before an audit. If you want the detail on how the EU AI Act’s risk tiers work, we cover that separately in building trust in AI: the EU AI Act’s risk-based approach. For AI readiness, the practical point is simpler: it’s easier to build governance in from the start. In the context of an AI tool, that means role-based access, a full audit trail for every AI decision, and clear limits on what data any AI agent can reach.

For EU-based legal teams, this isn’t optional guidance, it’s binding law. Since 2 February 2025, Article 4 of the EU AI Act has required both providers and deployers to ensure their staff have a sufficient level of AI literacy, with training and awareness measures expected to be documented. Since 2 August 2026, Article 26 also requires deployers of high-risk AI systems to keep automatically generated usage logs for at least six months, maintain technical documentation and risk assessments, and be able to show human oversight to regulators on request (see the European Commission’s regulatory timeline for the full schedule).

A minimal but defensible setup covers three things: an inventory of which AI systems are in use, role-based usage policies, and a training register that records who attended, what was covered, when, and the outcome, for every AI literacy session.

A team that made this transition

SRUV, a European arbitration body handling passenger rights disputes across travel and transport, moved from manual case handling to a governed AI workflow using e! by Lexemo. Their team translated existing arbitration expertise into visual decision trees, connected to existing systems through APIs and built with the flexibility to switch between AI models rather than being locked into one provider. The result: case processing time dropped from 5–7 minutes to under 1 minute per case, across 45,000 cases a year, with every decision step audit-ready. Non-technical staff used AutoMate, a conversational workflow builder, to design and refine the logic themselves, without writing code, and have since built over 400 Legal Bots independently. That’s the difference between individual efficiency and organisational readiness: a single lawyer using ChatGPT gets faster at one task, while a governed workflow like this one runs the same way for everyone who touches the process.

Where to start your team’s AI readiness this week

If a new tool is already on the table, take a step back and look at the bigger picture first: is the groundwork for AI readiness actually in place? Is there a current list of who’s using which tools? Where are the biggest pain points in the team right now, and where is capacity being tied up unnecessarily? Do you already have a clear AI usage policy, even an imperfect one is better than none? Do you have clear criteria documented for buying further tools, for example: EU hosting, audit trails, human oversight, integration options with your existing tool landscape, and visible decision logic? The team that has defined its own AI readiness moves faster, more strategically, and with results that last.

Frequently Asked Questions

What does an AI readiness check actually cover for an in-house legal team?

An AI readiness check for an in-house legal team covers four steps: an audit of which AI tools are already in informal use, a clear governance policy covering data handling and human review, two or three high-volume workflows selected for structured automation, and a way to measure baseline performance before automating. It is an organisational readiness question, not a tool-shopping question.

Why is a workshop worth running for an in-house legal team’s AI readiness?

A well-run workshop gets everyone who matters around one table in a few hours. It maps which AI tools are already in use, identifies two or three workflows with the clearest automation potential, and captures baseline numbers you’ll use to measure success later. Just as important: it involves everyone affected and takes the fear of AI and job loss off the table.

What AI tools should an in-house legal team actually use?

An in-house legal team should prioritise AI tools whose underlying mechanism is visible and traceable, that are hosted within the EU or EEA and hold recognised certifications such as ISO 27001, and that support role-based access with a full audit trail. Tools that only offer a general-purpose chat interface are harder to govern than tools built around defined, auditable workflows.

How do you choose an AI assistant for an in-house legal team instead of a general-purpose chatbot?

Choosing an AI assistant for an in-house legal team means checking three things: does it work inside defined workflow boundaries rather than as an open-ended agent, can every decision step be traced and explained, and can legal staff adjust the logic without a developer ticket? A tool that cannot explain its reasoning is a liability, no matter how fluent it sounds.

Can in-house legal teams automate contract work with AI, or does every contract still need a lawyer?

In-house legal teams can automate the repetitive parts of contract work with AI: NDA generation, standard clause checks, and routing to the right approver. Contracts that fall outside defined parameters, such as unusual terms or high-value deals, should still route to a lawyer. The automation qualifies and pre-structures the request; it does not replace judgment on genuinely complex matters.