Live webinar — Wed 3 June 2026

Under the Hood: How agentic AI actually does contract work (reliably)

87% of in-house legal teams report using AI. Only 7% say it's actually reduced their outside-counsel spend. The adoption story is over. The impact story hasn't started — and that gap is the defining problem for in-house legal in 2026.

Martin Lukac CTO at Flank

Date Wed 3 June 2026

Time 4pm BST · 5pm CEST · 11am ET

Format 45 min talk + 15 min Q&A

Save your seat

Can't make it live? Register anyway — we'll send the recording.

Event intake

Sign up in under a minute

A few fields and you're in. We'll send the calendar invite, joining link, and the recording afterwards.

01 — The session

Not the vendor demo. The architecture underneath.

In this 60-minute session, Martin Lukac, CTO at Flank, goes under the hood of how agentic AI actually does contract work — reliably.

How a language model gets wrapped with the right context, tools, guardrails, and supervision to make it reliable enough to review, redline, and run playbooks across thousands of contracts without a lawyer checking every line.

Built for general counsel, legal operations leads, and in-house lawyers who've sat through enough demos to want to understand what's actually happening underneath.

02 — You'll leave with

A working mental model — not another vendor pitch

A working mental model of how agentic AI is built.
Why the last 18 months of model improvements changed what's possible.
Where the 87% vs 7% impact gap actually comes from.
The four questions every legal team should ask before buying AI.
What the in-house operating model looks like when AI handles the routine contract work.

03 — The frame

The gap between adoption and impact

The defining problem for in-house legal in 2026 isn't whether to adopt AI. It's why adoption hasn't moved the outside-counsel line.

87%

of GCs report AI use — up from 44% last year, 20% in 2023

say AI has actually reduced their outside-counsel spend

78%

of legal departments mandated to "do AI" without dedicated funding

10–50×

larger services budget vs. the £100–300k software line

Sources: FTI / Relativity GC Report, March 2026. In-house benchmarking.

04 — The takeaway

The four questions to ask before buying anything labelled "AI"

Most of what gets labelled "AI" makes a lawyer faster on one step of one workflow. That's useful. It's not the same as software that does the work. These four questions separate the two.

Does it do the work — or just make a lawyer faster?

Does it run on your playbooks — or generic training data?

Does it handle the full workflow — or one step?

Who supervises the output?

05 — Your speaker

Martin Lukac, CTO at Flank

Martin runs the team building Flank's agents — the ones in production at FT, Bolt, DeepL, Mural, and TravelPerk handling NDAs, DPAs, and the long tail of routine contract work. He's the person who sees what breaks, what scales, and where the architecture matters more than the model.

Wed 3 June 2026 · 4pm BST

Save your seat. 45-minute talk, 15-minute live Q&A. Can't make it? Register anyway — we'll send the recording.

Under the Hood: How agentic AI actually does contract work (reliably)

Save your seat

Sign up in under a minute

Not the vendor demo. The architecture underneath.

A working mental model — not another vendor pitch

The gap between adoption and impact

The four questions to ask before buying anything labelled "AI"

Martin Lukac, CTO at Flank

The Intake

Outsource legal work to supervised agents