Legal tech in law firms and in-house teams: The wall is coming down

For years, the legal technology market organised itself in a peculiar way. Law firms on one side, in-house legal teams on the other. Both bought legal tech, but rarely the same products. Legal research, document management and eDiscovery crossed over. Almost everything else split; the platforms were built around workflows pointing in opposite directions.

The split made sense, because the two sides were buying for different outcomes. A firm sells expert time to external clients, so its technology existed to capture that time and protect the margin on it. Success meant utilisation and realisation. 

An in-house team is a business-enabler with one client - the business - and an ever-increasing queue of internal requests. The outcome it needs is the business moving safely with high velocity with as few lawyer hours as possible. Its work also doesn't end at signing: obligations and renewals run on for years. So technology that focuses on this side historically existed to route the volume and keep track of what happens after the deal closes.

Firm tech made expert hours more valuable, whereas in-house tech aimed to make fewer of those hours necessary.

That split is now breaking at one layer and hardening at another. The models are shared. The products built on them are starting to split again along the old line.

Claude, Gemini and ChatGPT are used by law firms and in-house teams alike, often through enterprise deployments with no legal-specific wrapper. A partner drafting a memo and a general counsel stress-testing a contract position can be working in the same model, through the same interface. 

This matters because previous generations of legal technology were integration plays. They connected existing workflows, automated document flows and surfaced existing data in more useful formats. The intelligence, such as it was, lived mostly in the rules and structured data that humans had built around the system. The model itself was not the product. 

But that is changing. Frontier models can take on work that used to need a lawyer at every step. Of course, the output still needs real oversight, but it doesn't need rebuilding from scratch. When the same capability serves an associate and in-house counsel doing comparable work, the case for two separate product markets gets thin.

This play is most visible at the platform layer. Harvey launched as a law firm tool and is now a fixture in in-house environments. Luminance built its reputation in law firm due diligence and has expanded steadily into CLM and legal operations. Legora sells one platform to both, but is now building segment-specific products on top of it. The companies themselves rarely frame this as convergence; they frame it as expansion.

This is partly commercial. Law firms and in-house teams are both large buyers and serving only one side of the relationship is a constraint, not a strategy. But it is also a product decision, because the underlying tasks are not as different as the market structure implied. Contract analysis looks similar whether the lawyer doing it sits in a firm or a legal department. So does due diligence, drafting, regulatory review, issue spotting. The divide was shrinking before the legal tech vendors had fully registered it.

The legal AI tools may be shared, but how law firms and in-house departments use it in their workflows are completely different.

Law firms operate around matters: defined engagements with starts, ends, billing codes and client reporting. In-house legal teams operate around continuous flows of work with no natural beginning, internal stakeholders instead of paying clients and obligations that run across time rather than project. These are not just organisational differences. They reflect genuinely different ways of thinking about how legal work is structured.

An in-house team using a platform built around matter logic will encounter assumptions that do not map onto how their work flows. A law firm using a platform built around operational throughput will find itself reaching for structures the product did not anticipate. Capability has converged faster than workflow design has. The platforms are still working out what a genuinely shared model of legal work looks like when you are not starting from an assumption about which institutional context it sits inside.

October 2026 showed how vendors are answering that. In the same week, Legora and Harvey each launched a product called Intake. Each is a front door for in-house teams that captures requests, asks for missing context, routes work to the right lawyer and drafts a first pass from the team’s own playbooks. Legora’s is live in Slack. Harvey’s is in early access. Both are built for legal departments, not firms. The engine is shared and the product on top of it is splitting along the old line. 

None of this is new. The previous generation of legal tech didn't just segment the market, but encoded the split. Firms bought practice management and billing built around engagements with a paying client. Legal departments bought the mirror image: matter management and e-billing built to control spend on those same engagements. Same matter, opposite ends of the invoice.

Contract lifecycle management (CLM) is the clearest case. These systems were built entirely around in-house legal operations: intake portals, approval workflows, obligation tracking, renewal calendars. Law firms have no native use of CLM, but at times offer it as a managed service to clients.

For years, the threat to CLM vendors looked distant, because legal AI platforms were law firm tools. That’s no longer true. The platforms that won the firms are now selling directly to legal departments and they’re taking pieces of the CLM stack with them. 

Take contract redlining. A firm associate reviewing a counterparty's MSA and an in-house counsel handling an incoming NDA can each load a playbook and fallback positions into Spellbook or Harvey inside Word and get speedy redlines, with comments explaining each change. The only difference is whose playbook it is. For the firm, it belongs to the client and for the in-house team, it belongs to the company.

Obligation tracking follows the same pattern through tabular extraction. A firm running diligence on a deal and a legal department auditing its own executed agreements can both drop a batch of contracts into Harvey Vault or Legora's Tabular Review. Each gets a grid with one row per contract and columns for renewal dates, notice periods, payment terms and audit rights.

Intake goes further. Harvey’s version routes contract work into its own Contract Intelligence product. That looks less like a workaround to CLM and more like a bid to become it.

The legacy vendors built their moats around the assumption that the two markets would stay separate. Legal AI platforms are making that assumption look increasingly fragile.

Firms have long operated on the basis that they knew what work required what expertise, how long it took and what it was worth. Clients accepted that framing because they lacked a credible alternative. An in-house team using the same platform as their outside counsel starts to develop real intuitions about how long document review should take, what contract analysis actually involves and where human judgment adds value versus where it is reconstructing work the platform could have done faster. Those intuitions enter conversations about scope, budgets and billing. They change what clients feel entitled to ask.

The bigger shift is in information, more than fees. When both parties in a professional services relationship are working within the same technological frame, the knowledge gap that historically structured that relationship starts to close. What gets renegotiated is not just cost. It is the definition of the service itself.

The law firm model has always depended on a specific learning architecture. Junior lawyers do structured work - research, drafting, review - and develop judgment through repetition and feedback. That judgment eventually becomes what clients pay senior rates for. It works because the early work is meaningful in both output and developmental terms.

If a platform handles first-pass contract review as well as a junior associate, the question is not just what the associate does instead. It is how the associate develops the judgment to supervise and refine the platform's work without having done the underlying task enough times to know what it feels like to get it wrong. Supervision without prior experience is a different skill and the profession has not yet worked out how to build it.

In-house teams face an adjacent version and it’s no longer hypothetical. Legora’s Intake can complete pre-approved routine requests end-to-end and send the answer straight back to the business. Harvey’s platform resolves routine requests and drafts the rest. A lawyer reviewing that output is working on what to remove from the material. That’s often efficient. It can also slowly wear away the close engagement that makes oversight mean something. The risk is that lawyers stop knowing what they’re looking for.

Traditional legal technology came with a defined responsibility map. The vendor supplied the tool, the organisation implemented it and the lawyer remained professionally accountable for the work product. That held because the tool was doing something discrete and bounded: retrieving a document, surfacing a clause, flagging a deadline.

Platforms operating across both environments are doing something more continuous. They shape how work is initiated, scoped, analysed and delivered. The lawyer remains accountable, but the platform is increasingly embedded in the chain of decisions that leads to the output the lawyer signs off on. When something goes wrong, the question of where the error entered the process is genuinely difficult to answer. 

The legal profession has governance frameworks developed over a very long time for very good reasons. The question is not whether they apply here. It is what they look like when the assistance is this deeply embedded in the work itself.

The fault line that organised the legal technology market for decades was never natural. It was a product of how technology was built, how professional services relationships were structured and how the economics of legal work had historically developed. None of that was permanent. It was contingent.

What the current generation of Legal AI platforms is doing, more than anything else, is making that contingency visible. The harder questions - who captures the value that gets displaced, how expertise develops in this new architecture, where professional accountability sits when a platform is this embedded in legal work - are not going to be resolved by the technology itself. They require the profession to make active choices about what it is actually for.

Those conversations are beginning, but they're running well behind what's shifting. The sharpest one: if clients learn what the work takes and juniors stop doing it, who gets paid for the judgment that's left? 

Shreya Vajpei is the Founder of Indian LegalTech Network.

Vaishali Gopal is Senior Advisor, Indian LegalTech Network and Founder of Future Relics Consulting.