Research, product updates, and working patterns from the teams putting legal agents into production.
AI knows contract law. It doesn't know your company, by Lorna Khemraz
The playbook had a gap. The agent found it, by Paul Lacey
If 2024 was the “Year of the Pilot”, a time of cautious experimentation, proof-of-concept projects, and learning what works, then 2025 marked the beginning of the shift.
Sort the pile by what you're doing, not what the document is called, and the right first workflow shows up, by Khaled Baroudi
The cheapest task in legal never gets done, and it's quietly the most expensive one to skip, by Taariq Ismail
The trap is checking everything at the same price, whatever the risk, by Martin Lukac
and how it could, by Paul Lacey.
How to move the work and resolve the bottleneck, by Jake Jones
What a lawyer needs before putting their name behind an agent's work, by Lorna Khemraz
..and what a Roman emperor knew about it, by Khaled Baroudi
And why the first thing an agent does is hold up a mirror, and why that's the point, by Taariq Ismail
Stop judging legal AI by the demo. Judge it by an ordinary Tuesday, by Martin Lukac
Choosing the first use case is a strategic bet, not a scoping exercise, by Paul Lacey
The model your vendor runs is a risk you can close with the paper already in front of you, by Lorna Khemraz
Flank doubles down on its agents-as-a-service model, bringing in one of the most influential operators in the field.
What separates the teams that fixed intake from the ones still drowning in it, by Jake Jones.
In-house intake processes are messy. This is especially true at the enterprise level. Vendors rarely understand quite how chaotic intake and triage is, and they sell elegant solutions that would be a perfect fit, if only the way the…
Most of your AI vendors are running on a model they do not own, and your contract with them almost certainly does not say what happens when that model changes. That gap is one of the most underpriced risks in software procurement right…
Two product categories. One actual job. Here's why that matters, by Taariq Ismail
The market sells drafting and review as separate capabilities. The work was never divided that way, and the contract workflows being deployed now make the split impossible to maintain.
Today's legal AI gives your team a better place to do the work. I think the next stage does the work for them and hands back the outcome. To make that case I have to talk, briefly, about Flank. An opinion piece by Jake Jones.
Today's tools make your lawyers faster. The next stage takes the work off their desk entirely. By Jake Jones
A few weeks ago our product designer came to me and told me he was having a crisis. Not a vague one. He sat down and said, fairly directly, that it felt like his role had changed underneath him and he was not sure what was left of it.
AI doesn't replace roles evenly - it hollows them from the middle, by Paul Lacey
A few weeks ago I ran one of the AI automations I use to speed up a chunk of my own work. It started generating immediately. The output was structured, the tone was close, the length was right. For the first forty seconds I thought: this…
The sequencing, measurement, and expertise that separate pilots from outcomes, by Lorna Khemraz
Every vendor's demo looks sharp. The difference shows up at week six, by Martin Lukac
A few weeks ago I watched a general counsel run a clause past Claude over breakfast. She had a redline from a counterparty, an indemnity carve-out that did not sit right, and forty-five minutes before her first call. She pasted the clause…
Notes from The Lawyer's GC Summit in Alicante, May 2026
Why the only thing that resolves the AI FOMO is a plan that is yours, by Lorna Khemraz
Why faster AI makes being in the room more valuable, not less, by Paul Lacey
What coding agents changed about the build vs. buy decision, and why the answer is still buy, by Martin Lukac.
52% of legal teams use AI. 7% have moved the spend number. The architecture explains the difference, by Taariq Ismail.
What an enterprise legal function should look like when headcount stops being the constraint, by Flank Co-founder Jake Jones
Output is no longer scarce. The question is what the lawyer's hour was always for. By Lorna Khemraz.
The data is there. The surface to make it useful isn't, by Paul Lacey.
Most legal ops leaders aren't asking it yet. In two years, it's the only one that will matter, by Taariq Ismail.
Legal training worked through proximity, observation, and correction. All three are breaking. Lorna Khemraz investigates.
The erosion of human endeavour by agentic automation. An exploration continued by Jake Jones.
Why most legal AI pilots solve an activity, not a need, by Taariq Ismail
Why the absence of failure in legal AI is itself a failure of ambition, by Martin Lukac
The front door is how legal AI stops being a pilot and starts being a production system, by Paul Lacey
Change management is the variable nobody budgets for, by Taariq Ismail
Why the deployments that 'failed' were never built to succeed, by Lorna Khemraz
Why 52% of legal teams adopted AI but almost none reduced costs, by Martin Lukac
A nine-year-old beats six boys with a four-move algorithm. Decades later, machines prove that even intuition is computable.
Why most legal AI pilots fail to move out of pilot stage, by Taariq Ismail
The real risk profile of scaling AI agents: the jagged edge is structural, not temporary, but recovery is measured in minutes.
The services line item that enterprise legal teams are starting to reallocate. Jake Jones explains.
The same capability that makes legal software better eventually makes it unnecessary. The question is which layer you're buying.
Why better AI doesn't mean a better model, by Martin Lukac
Legal teams can name what to automate instantly. The hard part is building the reliability and scalability infrastructure around it.
Why enterprise contract complexity is no longer a ceiling for AI agents, by Paul Lacey
How operational chaos masquerades as legal complexity, and what happens when you separate the two, by Lorna Khemraz
Every legal AI product runs on the same handful of LLMs. The staggering difference in output quality is entirely a product problem.
Enterprise contract complexity is no longer a ceiling for AI agents. The boundary is moving fast.
The week legal AI stopped being theoretical
The legal services market is three markets with different economic engines. Agents will hit law firms, ALSPs, and in-house teams very differently.
The throughput problem nobody is measuring
For the past few years, the legal industry has been in a phase of exploration. We have seen firms and in-house teams adopting Generative AI tools largely for the novelty factor – testing the waters to see what is possible.
There is a pervasive fear that autonomous AI replaces the need for legal expertise. The reality is the opposite: it clarifies the value of true expertise by stripping away the drudgery.
For decades, the economic model of the in-house legal team has been broken. It was a linear equation, but not any more.
The ultimate measure of success for a legal AI deployment in 2026 is not how many lawyers are logging into a tool. It is whether the business users even know the tool exists.
For the last two years, lawyers have been told that AI is "almost there". Yet, in private, the reality has been different.
Flank is the first enterprise-ready agentic system built for legal work. Instead of assisting lawyers; it handles the work itself: reviewing, drafting, redlining, triaging, and filing across the systems your team already uses such as Outlook, Teams, Jira, and Salesforce.
This post explores how to build aligned legal AI agents that reflect context, escalation paths, and layered decision-making - not just outputs. That's because useful legal AI isn’t just about automation. It’s about understanding how lawyers think.
Trusted by DeepL, Bolt, SumUp, TravelPerk, Axel Springer and more.