Aleksandra Polak, Assistant General Counsel at Billtrust, in USA, discusses why in-house legal teams should move beyond governing AI to actively shaping the tools and systems that will define how businesses operate and scale.
AI is transforming how in-house legal teams work. It drafts, reviews, summarises, and explains. It’s entering workflows across contract management, compliance, and operations.
Many General Counsel draft AI guidelines, assess compliance risks, and interpret new regulation.
But if GCs only govern how AI is used, they’ll miss their biggest opportunity in a decade.
There are two layers where lawyers should get involved:
1. Helping design AI-powered systems and products for the business.
2. Building AI agents to augment legal work itself.
Both layers matter. One shapes how the company uses AI. The other reshapes how legal delivers value.
When legal stays in a policy and oversight role, it becomes reactive. Guidance gets written after tools are deployed. Issues are raised once something goes wrong.
But AI isn’t like any previous technology. It doesn’t just process data – it reflects the reasoning, biases, and priorities of whoever trained it.
If legal isn’t part of those design conversations, someone else defines what “legal judgment” looks like inside the system.
AI doesn’t just need to be governed. It needs to be guided by people who understand the intersection of law, business, and ethics.
The fiare notyer of opportunity lies outside the legal function, in the AI systems your company builds or buys.
As organisations embed AI into products and processes, questions multiply:
• What data can we use for training?
• How do we manage transparency and bias?
• Where do we draw the line between assistance and automation?
These aren’t purely technical questions. They’re questions of governance, risk, and ethics – exactly where GCs add value.
Lawyers who join those design discussions early can shape the company’s approach to AI before it’s coded into the product.
Being part of that process isn’t about writing policies but about architecting responsibility into the systems themselves.
“AI doesn’t just need to be governed. It needs to be guided by people who understand the intersection of law, business, and ethics.”
The second layer is closer to home: using AI to augment the legal team’s own work.
For years, legal tech promised a faster way to get repetitive work done. AI takes that further. It doesn’t just automate; it amplifies.
The idea is not to replace lawyers. It’s to build AI agents that extend them – their tone, risk appetite, and reasoning style.
Instead of reviewing the hundredth 3rd-party NDA or DPA this month, why not design an agent that:
• Conducts first reviews of third-party documents
• Summarises contract economics
• Handles recurring compliance tasks
Legal teams should think less “case by case,” more “workflow and product.” Just as SaaS companies do not build a custom product for every customer, legal teams should not build from scratch each time.
With AI, we can design our own “AI twin” trained on our playbooks, templates, and risk philosophy.
For decades, lawyers held an advantage: knowledge others didn’t have – caps, carve-outs, indemnities, you name it.
AI has that knowledge now. It can draft, compare, and summarise faster than any associate.
Knowledge is the baseline.
Judgment is the differentiator.
The lawyers who thrive in the AI era will be those who apply context, blending regulation, risk, and commercial sense – and teaching systems to do the same.
That’s what shaping AI really means: making sure it reflects your version of sound judgment, not a generic one.
AI will handle the predictable, standardised work:
• NDAs and DPAs
• Policy templates and checklists
• First-pass contract reviews
It struggles with:
• Ambiguity
• Context-heavy decisions
• Poorly drafted or inconsistent laws
Lawyers should be redrawing that line now, deciding which tasks can be automated and which must stay human.
Some teams are already creating “AI responsibility matrices” to clarify who owns what, when review is required, and how outputs are validated for accuracy and bias.
The goal isn’t replacement. It’s reallocation and freeing time for the work that truly requires judgment.
Buying AI tools offers speed. Vendor products are polished and proven. But they’re trained on generic data, built for the average customer.
When it comes to AI for legal work, I support build.
All off-the-shelf legal AI tools are built on the same large language models – ChatGPT, Claude, Gemini, or similar. Each still needs to be trained, configured, and aligned to your company’s specific templates, tone, and risk thresholds.
If you have to do that work anyway, you might as well build your own.
Building in-house delivers:
• Control: you know exactly what data the system uses and how it reasons.
• Precision: your outputs reflect your policies, not generic “best practice.”
With strong internal data, custom AI agents can be cheaper, faster, and better aligned than buying another platform that will always need retrofitting.
The future legal team isn’t just an AI user but an AI builder.
GCs who limit themselves to policy and compliance oversight will always be reacting to technology. Those who help design it will shape how their organisations make decisions.
AI doesn’t remove the need for judgment.
It magnifies the impact of good judgment.
The opportunity for today’s GC is clear:
• Help design AI-powered systems that define how the company operates.
• Build AI agents that redefine how legal delivers value.
The lawyers who do both won’t be replaced by technology.
They’ll be the ones building the future of responsible, scalable decision-making.
Be part of a growing global community committed to advancing in-house legal leadership.
Alexandre Verrien, former Legal Director France & Europe at Dell Technologies, examines why in-house legal teams often lose influence despite growing capability and argues that...
Every GC eventually meets a moment where the “right” answer isn’t obvious – or popular. Alexander Shevchenko, Chief Legal Officer at WebPay, shares the mindset...
Learn more about How does a GC make the decisions no one else wants to?
