Legal Tech Innovation Consulting @ Titans | Head of Digital @ BamLegal

Peter Duffy

Nicola Shaver

CEO and Co-founder, Legaltech Hub | Author | Innovation, AI, Legaltech Leader, Advisor, Investor | LLB, MBA | Fastcase 50, 2021 & 2024, ABA Women of Legaltech, 2022 | Adjunct Professor

I’m an executive leader in legal AI and innovation, recognised by Fastcase 50 (2021, 2024), ILTA (Innovative Leader of the Year), ABA Women of Legal Tech, Relativity (AI Visionary), and as a Fellow of the College of Law Practice Management.

I work at the intersection of law, technology, and business, helping firms, corporate legal teams, and legaltech companies turn AI and legal technology from buzzwords into practical, sustainable change.

As the CEO and co-founder of Legaltech Hub, I lead a team dedicated to providing the most comprehensive, high-quality data and insights on legal technology and generative AI globally. Our platform helps law firms, corporate legal departments, and vendors make informed decisions with market reports, tools, and resources they can trust.

In addition to running Legaltech Hub, I advise law firms on AI strategy, legal technology evaluation, and strategic decision-making, and I work with legaltech companies of all sizes -- from early-stage startups to mature players -- on product strategy, roadmap, and how to succeed in selling to law firms. I’ve had the privilege of serving on the advisory boards of several innovative legaltech companies and investing in others.

I’m passionate about sharing knowledge and inspiring change. I teach a class on AI and legal technology as an adjunct professor at Cardozo Law School, and I frequently speak as a keynote at conferences across the US, Canada, Europe, the UK, Australia, and New Zealand. I’m also a prolific writer on generative AI, legaltech, and innovation in law, and my book, "The Handbook for Legal Innovation", was published in 2023.

A former practicing lawyer and Global Managing Director of Knowledge & Practice Innovation at Paul Hastings, as well as an innovation leader in other top-tier firms and Fortune 500 environments, I know firsthand the challenges and opportunities facing the legal industry today.

My mission is simple: to help the system of law work better for everyone -- through advising, educating, mentoring, and connecting people across the industry.

CEO, Co-Founder

The definitive resource for legal technology - access the most comprehensive directory of commercial global legal technology with a powerful search allowing both law firms and legal departments to easily find the tools they need to support their practice. In 2022, Legaltech Hub is evolving into a full insights and analysis platform for legal technology, with practical guidance for all sectors of the market.

Access Legaltech Hub here: https://www.legaltechnologyhub.com/

Three law firms are part of the current Y Combinator winter 2026 cohort.

That tells us something about how AI-native firms are operating in our ecosystem - increasingly like product-led growth companies that happen to deliver regulated services. The only previous YC law firm was Atrium.

Between 2024 and 2026, 20 AI-native and AI-first law firms have come to market. In 2025, this activity accelerated, and this year, we’ve already seen five more launch out of stealth: General Legal, Arcline, LegalOS, JustPoint Law
and FairPlay Law.

Some of these are run by experienced and successful former tech founders (from Thomson Reuters, Casetext, and Pangea3). JustPoint and Eudia Counsel secured Arizona ABS approval before launch. Many have launched with as venture-backed startups with licensed law practices.

These are not just “AI-enhanced boutiques” anymore. They are operating as competitors to both tech companies and traditional firms.

The question is, will that be enough to challenge either of those verticals?
Can they win market share from incumbent law firms?
Can they scale like high-growth tech companies?

Or do they get pulled back into the gravity of partnership economics and regulatory constraint?

In my latest article, I map the emerging AI-native law firm ecosystem and examine it through the lens of Clayton Christensen’s Innovator’s Dilemma, exploring whether these firms true disruptors attacking from below with new production systems, or sustaining innovations dressed in AI language?

It's a topic that should matter to all of us. If these models prove viable, they don’t just change firm competition. They change the production architecture of legal services. And once that shifts, everything else follows.

👉 Link to full article in comments.

#law #lawfirms #lawyers #AI #AIinLaw

Nicola (Nikki) Shaver

Legal AI & Innovation Executive | CEO, Legaltech Hub | Former Global Managing Director of Knowledge & Innovation (Paul Hastings) | Adjunct Professor | Advisor & Investor to Legal Tech

In 2026, lawyers trust #genAI for complex legal drafting.

That was not the case in 2024, when we started collecting adoption data by use case. 

But this year, not only are genAI tools where a lot of lawyers do their drafting, those solutions are starting to edge out more traditional document automation workflows. 

The most used solutions for legal drafting at law firms this year (out of a survey size of 130 firms) include Harvey, Thomson Reuters Contract Express,  Thomson Reuters CoCounsel, Definely, DraftWise, Clearbrief.ai, Legora, Litera Office & Dragons, Avvoka, LexisNexis Lexis Create+, Document Drafter.

Question for firms: do we still need two solutions for drafting (doc automation + smart drafting / AI drafting), or have we reached the point where these have sufficiently converged that only one is necessary?

In our analysis of this use case we look at the way market share has shifted from 2024 - 2026, looking not just at which providers are currently winning but also which ones have real momentum building around their gaining of market share. 

This data is from the annual SKILLS-LTH genAI Use Case Survey for 2026. The data just dropped and we'll be sharing our analysis one use case at time, each Friday after Oz Benamram and the SKILLS.law community host an online session discussing that use case. Reach out to Oz to register for those and follow along at Legaltech Hub to see our full analyses and understand how the market is shifting.

Full use case analysis report linked in comments. See that for momentum charts.

#legaltech #AI #lawyers #Law

Most law firms are asking the wrong question about AI.

They’re comparing tools, evaluating vendors, and debating use cases.
But after watching dozens of rollouts, here’s what actually determines success:

Who owns the workflow?

If you can’t answer that in five seconds, your AI initiative is already fragile.

No owner = no quality threshold
No owner = no adoption discipline
No owner = no authority to change behavior

The projects that succeed - litigation evidence summaries, due diligence review - are practice-owned. They sit inside a clear production system.

The ones that stall - intake, cross-functional processes, knowledge capture - don’t fail because they’re complex. They fail because they’re unowned.

The biggest misconception right now is that AI will clean up messy workflows and broken data.

It won’t. AI doesn’t fix disorder. It amplifies it.

Clear, accountable workflows scale. Diffuse ownership collapses.

AI implementation is workflow implementation. Most firms aren’t ready to admit that yet.

👉 Full analysis inked in comments.

#legal #law #legaltech #AI #lawfirms

Something big IS happening.
The question isn’t whether - it’s how fast.

Last week, Matt Shumer compared our collective complacency about AI to February 2020, when most people still thought COVID wouldn’t fundamentally change their lives. Just like many professionals are still treating AI as a side conversation.

Shumer predicts that within 1–5 years, up to 50% of white-collar jobs could be displaced by AI.

If that sounds extreme, it's intended to. Shumer's point is that there's real urgency here, and we're not heeding it.

In a public response on substack, Gary Marcus pushed back, arguing that today’s systems still struggle with reliability, long-horizon reasoning, and independent execution. They require oversight. They break and hallucinate, especially during long, complex workflows.

He’s right about current limitations.

But here’s the thing: even if AI isn’t fully replacing knowledge workers in the next year, it's already restructuring how knowledge work gets done.

One strong operator with AI can now outperform small teams without it.
Junior work is compressing. Learning curves are shifting.

And yet in many professional environments, especially law, the dominant question remains: “Is this overhyped?”

That's the wrong question.

The better one is:
What happens to our talent models, training pathways, and pricing structures if AI becomes deeply embedded, even without or before full automation?

Shumer may be early. Marcus may be right about today. But neither argument justifies complacency.

The real responsibility, especially for those of us in knowledge industries, is not to debate ourselves into comfort. It’s to prepare.
➡️ Prepare our teams.
➡️ Prepare incoming juniors.
➡️ Prepare our children.

Because whether this shift happens in two years or ten, the future of white-collar work will be shaped by people who know how to work with AI, not by those arguing about whether it will matter.

I'm linking both pieces in comments below, for those who haven't read them.

Law firms think they're innovating with AI. Most are standing still.

Even the ones that have genuinely succeeded at deploying genAI and driving adoption.

Over the past two years - skeptics notwithstanding - AI has fundamentally changed how legal work gets done. That's real. But here's the problem: it's changed it for almost everyone in exactly the same way.

Drafting is faster. Research is quicker. Summarization is instant. Transcripts are assumed. Review is more consistent. Data is accessible and queryable.

That's meaningful progress. But because most firms are using the same tools, it's also now table stakes.

When everyone optimizes the same work with the same tools, optimization stops being a differentiator.

The next competitive frontier isn't better AI adoption. It's Horizon 3 innovation - offerings that don't look like traditional legal services at all:
🔥  Legal expertise embedded directly into client workflows 
🔥  Continuous legal capability, not episodic engagements 
🔥  Outcomes and systems, not matters and hours 
🔥  Prevention and prediction, not just response

The firms that will struggle here aren't short on ambition or technology. They're constrained by structures purpose-built to make all work look the same - same leverage models, same billing logic, same incentive systems.

That structure is the real barrier. And most firms aren't talking about it.

In my latest article, I explore what Horizon 3 actually looks like in law, and offer a practical diagnostic firms can use to assess whether they're structurally capable of doing this work at all.

Link in comments.

#AI #genAI #Law #differentiation #lawyers

Great evening out with the Definely crew and a select group of law firm leaders. 

Also, a lovely walk in Central Park with Nnamdi Emelifeonwu  - why? Watch this space… something new coming soon.  

#legaltech

No alternative text description for this image

There’s a term we should start using more deliberately: Hallucination Tax.

It’s the hidden cost of AI.

Every minute spent verifying output, every citation double-checked, every “let me just confirm this.”

That’s tax.

The promise of AI is time compression. But if verification expands with output, the gains disappear.

Good tools reduce hallucinations.
Great tools reduce hallucinations and reduce verification time.

When evaluating AI, firms shouldn’t just ask, “How accurate is it?”

They should ask: How much verification will this require at scale?

That’s the real cost curve.

#legaltech #AI #lawfirms

Most people completely misread Anthropic’s legal plug-in launch.

This wasn’t about entering legaltech. It was about owning it.

Not by building end-user products, but by becoming the layer everything else runs on.

While vendors feared Anthropic was competing with them, Anthropic was working on making themselves unavoidable, becoming the obvious starting point for every legal tech startup.

That’s the game now. Not features. Not use cases. Not even users.

👉 Control of the stack.

Because if you own the model layer, you don’t need to win every deal; you just need everyone else to build on you.

The market thought this was competition.
It wasn’t - it was distribution.

Check out full article in comments.

#legaltech #AI #law #lawyers #LLMs

Legaltech CEOs & founders - this is for you.

Legaltech Hub is hosting LTH Velocity, a free, closed-door session designed specifically for the people building legaltech companies.

No buyers. No pitching. No noise.

Just honest conversation about what actually matters right now:
selling into law firms
👉 what’s working (and failing)
👉 paths to growth and exit
👉 the challenges no one talks about publicly

There are very few spaces in this market where founders can speak openly with peers. This is one of them.

We’ll then roll straight into LegalEve, where many of the top innovation leaders, CIOs, and buyers are.

First: real talk with peers
Then: real access to buyers

It’s free. It’s small. And it’s designed to give you an edge in this market.

Registration link for both events in the comments.

p.s. Even though it it BONKERS and not our branding, I couldn't resist including this movie poster ad for Velocity...

AI Vocabulary Day: Context Window vs. Retrieval Depth

Two terms that are often conflated. Not the same thing - and in legal AI, the difference is the ballgame.

Context window = how much text the model can see at once.
➡️ A property of the model, set at training/deployment
➡️ Measured in tokens
➡️ Bigger doesn't mean smarter - more text can actually dilute reasoning (yes, really)

Retrieval depth = how much relevant information the system can find and surface before answering.
➡️ A property of system design, not the model itself
➡️ Depends on retrieval quality, ranking, and orchestration
➡️ Determines whether the system can reason across a complex matter — or just a single document

In practice: Large context window + poor retrieval = noise dressed up as capability 

Strong retrieval + modest context window = focused, usable signal

(The ceiling matters too: retrieval can't fully compensate for a window that's genuinely too small. But most legal AI failures aren't hitting that ceiling -they're failing long before it.)

Architecture matters more than scale. That's why:
- Bigger models don't automatically fix legal AI
- Tools that shine in demos sometimes collapse on real matters
- A 1M-token context window is not a retrieval strategy

If you're evaluating legal AI, stop leading with: "How big is the context window?"

Start asking:  
"How does the system decide what goes into it?" 
"What happens when the relevant information spans dozens of documents?"

The second question is where most vendors go quiet.

#AI #legaltech #RAG #law #AIVocabulary

In 2024, almost no one was using GenAI for timelines and chronologies.

In 2026, that has changed.

In our latest LTH–SKILLS survey (n=130 firms), we’re seeing the early formation of a new AI workflow category, and a structural divide in how firms are approaching it.

Here’s the signal:
- One broad GenAI platform is pulling ahead in live adoption (~28%)
 - A second tier of vendors holds meaningful share
 - A widening field of litigation-focused point solutions is gaining evaluation momentum

But the real story isn’t who’s winning. It’s why firms are choosing differently.

👉 For litigation-heavy practices, timelines aren’t just outputs.
They are systems of record - living, evidentiary infrastructures requiring traceability, version control, and defensibility.

👉 For lighter-use environments, general GenAI may be appropriate.

Because this use case isn’t just a tooling question; it’s a workflow architecture decision. And those decisions compound.

In this week’s analysis, we break down:
 – Live vs pilot vs consideration shifts
 – Year-over-year redistribution of attention
 – Where momentum is building (even if live usage hasn’t caught up yet)
 – What the divergence between broad platforms and purpose-built tools tells us about the future of this category

Timelines may look like a narrow use case. They aren’t. They sit at the core of how firms transform unstructured facts into structured narrative, which is the heart of litigation, investigations, and due diligence.

Full breakdown now available here:https://lnkd.in/eVWqqk_y

#legaltech #AI #lawfirms #GenAI

MCP doesn’t just change how AI systems connect; it also changes where risk and accountability live.

Most governance frameworks in legal are built for a world where:
- tools were self-contained
- workflows lived inside products
- vendors defined responsibility boundaries

MCP breaks all three assumptions. When orchestration moves out of individual tools and into shared layers, governance can no longer follow ownership; it has to follow orchestration.

That introduces new challenges firms aren’t talking about enough yet:
⚡ blurred trust boundaries across tools and models
⚡ expanded attack surfaces created by interoperability
⚡ model and workflow drift that erodes reliability over time
⚡ the need for observability across multi-system AI workflows

These are concrete, structural risks that most firms aren't prepared for.

In the latest article in our MCP series, I outline why MCP acts as a governance stress test, and provide a checklist of the questions law firms and legal teams should be asking now, before composable AI becomes deeply embedded in day-to-day practice.

Article linked below.

#legaltech #AI #AIgovernance #legal #law

AI vocabulary day: Embeddings

Think of them as a map of similarity:
- Contracts that “feel” alike end up close together
- Concepts that are unrelated end up far apart

This is why, in a RAG system, weak results are often caused by poor embeddings, not the underlying model.

But here’s the part that really matters:
🔥 Embeddings decide what the AI pays attention to.

Who controls embeddings?
- the vendor, if embeddings are fixed and opaque
- the firm, if embeddings can be tuned, retrained, or swapped
- the platform, if embeddings are shared across tools

Why should lawyers care? Why should you care, gentle reader, if you are not building with AI?

Embeddings determine which documents are retrieved. They influence what context is included or missed. They shape the AI’s understanding of your language, clients, and precedent.

Two systems using the same model can give very different answers because they’re using different embeddings.

So when evaluating legal AI, it’s worth asking:
 ➡️ Who controls the embeddings?
 ➡️ Can they be adapted to our data and practice areas?
 ➡️ Are they optimized for legal meaning, or generic text?

Embeddings are not just plumbing. They’re judgment, encoded.

#AI #LegalAI #RAG #AIVocabulary

How will standards like MCP affect the corporate legal market?

Part 3 of our series on MCP is out today, written by the always engaging Samridhi Jain and the brilliant Joshua Duplantier of EY.

In this article, they explore how MCP:
- affects visibility into the business,
- reduces lift around integrations,
- accelerates tool consolidation, 
- improves data availability, and
- provides operational clarity for in-house legal.

At LTH, we're dedicated to breaking down new concepts and developments critical for lawyers and legal teams into accessible learning materials that will help you level up your AI maturity.

Access the full article (and our previous two MCP articles) here: https://lnkd.in/g3R4BF6x

#legaltech #AI #GenAI #MCP #lawyers

Open article: How MCP Is Changing the Way Corporate Legal Departments Operate by legaltechnologyhub.com

How MCP Is Changing the Way Corporate Legal Departments Operate

If you want better AI outputs, show examples - don’t just describe what you want.

One good example often does more than a paragraph of explanation.

#AI #AIadoption

taplio

The Battle

Peter Duffy

Nicola Shaver

Followers

Posts

Engagements

Likes

Performance ScoreNicola Shaver's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Peter Duffy

Nicola Shaver

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Nicola Shaver's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreNicola Shaver's LinkedIn Performance