LegalTech: what are startups building now?
In our Legal Tech market deck, you will find everything you need to understand the market
SUMMARY
LegalTech startups are building AI execution layers that can take in a legal matter, use trusted legal and company knowledge, complete multi-step work, and hand lawyers the parts that still require judgment.
The funding boom is unusually concentrated around companies trying to become infrastructure rather than narrow tools. Harvey and Legora alone account for roughly $1.1 billion of recent financing, while seven disclosed rounds in the article add up to more than $1.4 billion.
The copilot is becoming a feature inside a larger system. The more ambitious products now plan tasks, call tools, move across research and drafting, interact with systems of record, and return something close to a finished work product.
LegalTech is also splitting by workflow ownership. Harvey and Legora are going broad, while Ironclad, Supio, EvenUp, DeepJudge and others are building much deeper products around contracts, litigation, knowledge or another high-value legal domain.
Two of the most valuable control points sit outside the classic act of legal analysis. The “legal front door” captures work before a lawyer touches it, while contract-intelligence platforms keep working after signature by tracking obligations, renewals and operational actions.
The stronger moat increasingly comes from context rather than access to a frontier model. Firm precedents, negotiation history, matter data, playbooks, internal judgment and deep integrations can keep improving even when the underlying model changes.
Legal research is being pulled into the execution stack for a simple reason: an autonomous workflow is only useful if the law behind it can be checked. That pushes citations, authoritative databases, retrieval and human approval closer to the core product.
AI-native law firms are the most disruptive business-model experiment in the market. By combining software with licensed lawyers and selling finished work, fixed-fee or outcome-based providers can keep more of the productivity gain created by automation.
The labor effect is likely to appear first as fewer human hours per matter, especially in junior work such as review, first drafts, research synthesis, chronology building and routine diligence. That also creates a training problem for firms, because those tasks have historically been how younger lawyers learned judgment.
The clearest pattern across the market is that LegalTech is moving from isolated tools toward systems that control more of the legal workflow end to end. And yes, the market is messier than the label “legal AI” suggests: the real fight is over workflow, data, distribution and responsibility for the final legal outcome.
This market map, featured in our Legal Tech market deck, highlights top companies and startups in the Legal Tech market
LegalTech: what are startups building now?
Why is LegalTech suddenly attracting so much money?
LegalTech is attracting exceptional amounts of capital because startups are no longer pitching AI mainly as a faster way to write legal documents; they are trying to own the infrastructure through which legal work gets done. We are seeing investors fund platforms that can ingest a matter, understand a firm's knowledge, perform multi-step work, produce a deliverable and increasingly trigger the next action without forcing a lawyer to assemble the workflow manually.
The concentration is striking. Legal-tech startups have attracted roughly $2 billion of venture investment so far this year, according to reporting by Business Insider. Harvey's latest $550 million round and Legora's $550 million Series D alone represent about $1.1 billion, or roughly 55% of that total. Add Norm's $120 million Series C, Wordsmith's $70 million Series B, Crosby's $60 million Series B, Ivo's $55 million Series B and Checkbox's $23 million Series A, and those seven disclosed rounds account for more than $1.4 billion.
Harvey went from an $11 billion valuation earlier this year to $15.5 billion in its latest financing, while Legora moved from $1.8 billion in late 2025 to $5.55 billion only a few months later. Investors are betting that a few platforms could capture part of the legal-services market itself rather than merely sell another productivity tool to lawyers.
| Startup | Recent disclosed financing | Valuation |
|---|---|---|
| Harvey | $550M | $15.5B |
| Legora | $550M | $5.55B |
| Norm | $120M | $1.2B |
| Wordsmith | $70M | Not disclosed |
| Crosby | $60M | Not disclosed |
| Ivo | $55M | Not disclosed |
| Checkbox | $23M | Not disclosed |
Are LegalTech startups still building AI copilots?
LegalTech startups are moving quickly beyond the copilot model toward agents that execute complete legal workflows. Drafting, summarizing and answering questions remain important, but they are increasingly features inside much larger systems rather than the whole product.
Legora provides one of the clearest examples. Its early product was presented as collaborative AI for lawyers; its newer aOS is explicitly described as an agentic operating system that moves from matter intake through research, drafting, review and final delivery. Its Agent can plan a task, choose tools, execute the steps, evaluate its own work and return a finished product for human approval.
Harvey is moving in the same direction. Earlier this year it said customers were already running more than 25,000 custom agents, with agents handling high-volume and increasingly complex legal work from beginning to end. Supio has taken the model even further inside plaintiff law: its agent operates across an individual case, an attorney's caseload and the entire firm, while its intake agent can conduct prospective-client conversations, score leads and push structured information into the case-management system.
A copilot waits for the next prompt. These systems increasingly perform the next steps themselves.
If you want more recent data on this point, please see our latest Legal Tech market report.
As this chart shows, and as featured in our Legal Tech market deck, search interest in Legal Tech has been growing steadily
What are LegalTech startups actually building?
LegalTech startups are currently converging on a handful of distinct products: general legal operating systems, contract automation, legal intake and operations, litigation platforms, proprietary knowledge systems, regulatory agents and AI-native law firms. Calling all of them "legal AI" hides how differently they intend to make money.
Harvey and Legora want to become broad execution layers across many practices. Wordsmith, Checkbox and Eudia are attacking the corporate legal department from its front door. Ironclad and Ivo are concentrating on contracts. EvenUp and Supio are building around plaintiff litigation. DeepJudge is attacking institutional knowledge. Norm is combining regulatory software with an actual law firm. Crosby, LegalOS, Vector Legal and Manifest OS go further by selling legal services themselves.
The shared pattern is expansion across the matter lifecycle: upstream into intake and downstream into execution, records, reporting and service delivery.
| What startups are building | Representative companies | What they are trying to own |
|---|---|---|
| General legal AI platforms | Harvey, Legora | Research-to-delivery workflow |
| In-house legal operating systems | Wordsmith, Eudia, Checkbox | Legal requests and internal workflows |
| Contract intelligence | Ironclad, Ivo, Spellbook | Contract lifecycle and obligations |
| Litigation platforms | Supio, EvenUp, Eve | Case lifecycle and case evidence |
| Knowledge infrastructure | DeepJudge | Firm knowledge and precedent |
| AI-native legal services | Crosby, Norm, LegalOS, Manifest OS | The actual legal service |
| Specialist workflows | Patlytics, Solve Intelligence | High-value domains such as patents |
Are Harvey and Legora building the same legal operating system?
Harvey and Legora are converging on the same strategic prize: becoming the operating layer for sophisticated legal work. Harvey is pushing deeper into models, agents and customer-specific intelligence, while Legora is emphasizing orchestration across existing legal systems and collaborative workflows.
Harvey already operates at exceptional commercial scale. The company says 80% of Am Law 100 firms use Harvey, alongside five of the Fortune 10. Recent reporting puts it at roughly 200,000 users and about $350 million in annual recurring revenue. Its valuation has risen to $15.5 billion after its latest $550 million financing.
Its product strategy is also expanding downward into the AI stack. Harvey has deployed custom agents, embedded legal engineers inside customers and begun building domain-specific model capability. Harvey Tenet, its first post-trained open-weight model, reduces its dependence on simply wrapping OpenAI or Anthropic. Harvey LAB and its Legal Agent Benchmark show that evaluation is becoming part of the platform too.
Legora is taking a slightly different route. It says its platform has grown beyond 1,500 law firms and corporate legal teams and more than 100,000 legal professionals across over 50 markets. Earlier this year it reported surpassing $100 million of ARR less than 18 months after general launch, up from roughly $1 million. Its valuation rose from $1.8 billion to $5.55 billion within months.
Legora has launched an agentic operating system, acquired agent startup Walter AI and litigation-intelligence company Wexler, and built integrations with iManage, NetDocuments, Box, Docusign, Ironclad and Intapp. Rather than replacing those systems, it is trying to orchestrate work across them.
Harvey is putting more effort into owning the intelligence stack itself, while Legora is leaning harder into becoming the control plane across the legal stack.
If you want more recent data on this point, please see our latest Legal Tech market report.
This chart, featured in our Legal Tech market deck, illustrates yearly venture capital funding for Legal Tech startups
Are contracts still the biggest opportunity in LegalTech?
Contracts remain one of LegalTech's biggest commercial opportunities, and startups are no longer satisfied with automating redlines. The category is moving toward contract intelligence: using agreements as structured operational data that can trigger actions across sales, procurement, finance and legal.
Ironclad illustrates how far that transition has progressed. The company surpassed $200 million in ARR and says its platform has processed more than two billion contracts. Its newer AI agents can identify obligations, renewal risks or commercial issues and then push users toward an action instead of simply answering questions about the document.
Ivo raised $55 million specifically around AI contract work for in-house teams. Spellbook, which started with an AI contract-review tool inside Microsoft Word, said when announcing its latest financing that it had reached roughly 4,000 law firms and in-house teams across 80 countries and expected revenue to triple over the year.
Contract startups began by helping lawyers read agreements. Now they are trying to own the workflow before signature, during negotiation and after the deal is done.
Why is the "legal front door" becoming a startup category?
The legal front door is becoming a real LegalTech category because the bottleneck inside many corporate legal departments starts before a lawyer does any legal analysis. Requests arrive through email, Slack, forms and meetings; lawyers manually triage them; routine questions consume expert time; and management has little structured data showing what the department actually does.
Wordsmith is building directly around that problem. The company describes its product as an inbox that can receive legal requests, perform work and return matters to lawyers ready for review. It says revenue grew more than 14-fold over twelve months and that more than 500 companies now use the platform.
Checkbox raised $23 million around a similar thesis: become the "front door" through which internal legal work enters the department, then automate routing, intake and repeatable workflows. Eudia's unified workspace combines an assistant, specialized agents and what it calls Expert Digital Twins intended to capture the judgment of experienced lawyers.
The product opportunity is to automate the queue before expensive legal judgment is required.
This chart, featured in our Legal Tech market deck, looks at Clio’s strategy in Legal Tech
Are LegalTech startups replacing legal point solutions with one giant platform?
LegalTech startups are consolidating some point solutions, yet the market is currently settling around connected platforms rather than one application replacing everything. The fastest-growing companies are integrating aggressively with systems that already contain legal data because those systems are deeply embedded in how firms work.
Legora connects with iManage and NetDocuments for document management, Box for enterprise content, Intapp for professional-services workflows, Ironclad and Docusign for contracts, and DeepJudge for institutional knowledge. Harvey has partnered with Ironclad around contracts and PacerPro around litigation docket information. Supio connects plaintiff-law workflows to Thomson Reuters' Westlaw legal research.
The architecture emerging is layered: systems of record remain in place, while AI platforms sit above them to retrieve information and coordinate tasks.
Are proprietary legal knowledge and models becoming the real moat?
Proprietary legal knowledge is becoming a stronger LegalTech moat than access to any single language model, although model customization still matters. Precedents, previous negotiations, playbooks, client preferences, matter histories and experienced lawyers' judgment are being converted into machine-readable context that agents can reuse.
DeepJudge is built almost entirely around that problem: helping lawyers search and apply institutional knowledge that historically sat buried inside document-management systems. Legora has made governed institutional knowledge a central part of its integrations. Harvey's customer-specific agents likewise encode workflows and expertise rather than relying on a generic model response.
Eudia makes the thesis particularly explicit with its Expert Digital Twins, which are designed to preserve how an organization's strongest experts reason about recurring problems. Supio says its plaintiff-law system learns from a firm's cases and applies that accumulated knowledge across its caseload.
Harvey's decision to post-train an open-weight model adds another layer. EvenUp promotes its own Piai system for plaintiff law, while other startups use several frontier models and route tasks toward whichever performs best.
The more defensible asset is the feedback loop around real legal work: precedents, approved outputs, negotiations, case outcomes and internal decision patterns that remain valuable when the underlying model changes.
If you want more recent data on this point, please see our latest Legal Tech market report.
This chart, featured in our Legal Tech market deck, illustrates yearly funding for Legal Tech startups
Are litigation startups building something different from corporate LegalTech?
Litigation startups are developing a much more case-centric form of LegalTech because lawsuits generate an unusually rich combination of evidence, chronology, medical records, depositions, legal research and financial outcomes. That lets specialized companies build deeper workflows than a horizontal legal assistant can easily reproduce.
Supio has expanded from analyzing records into an agent covering intake through case resolution. The company says ARR has increased 17-fold since emerging from stealth, and its platform now includes intake, medical-record workflows, legal research and case-level automation.
EvenUp has taken a similar vertical approach to personal-injury law. Its platform uses information drawn from hundreds of thousands of injury cases and millions of medical records to generate demand packages and other case work. Its $150 million financing valued the company above $2 billion.
The repetition matters: plaintiff firms handle large portfolios of cases with recurring document structures and measurable outcomes, giving specialist systems unusually rich feedback loops.
Are startups building better legal research engines?
Legal research is becoming infrastructure inside broader LegalTech products rather than remaining a standalone search box. Research databases, matter context and workflow execution are merging because an agent cannot safely complete legal work if it cannot reliably verify the law behind its conclusions.
Clio's $1 billion acquisition of vLex is the clearest structural evidence. Clio historically owned practice management; vLex brought a massive legal-information database and the Vincent AI platform. Combining them puts authoritative law, matter information and firm operations inside one system.
Specialist AI companies are also connecting themselves to incumbent research libraries. Supio has integrated Westlaw into plaintiff-case workflows, while Harvey has connected to PacerPro for live docket intelligence. Legora has progressively added primary legal materials alongside private firm knowledge.
A Stanford evaluation published in 2025 found that leading proprietary AI legal research systems still hallucinated more than 17% of the time under the researchers' testing methodology. Verified legal research therefore has to sit inside the execution workflow, not beside it.
This chart, featured in our Legal Tech market deck, compares the main business model options for Legal Tech SaaS platforms
Are AI-native law firms starting to replace traditional firms and the billable hour?
Yes. AI-native law firms are beginning to compete directly for legal-services revenue, and their economics make them more capable of challenging hourly billing than conventional LegalTech software. They combine agents with licensed lawyers, sell the finished legal outcome and can keep more of the productivity gain when automation reduces the work required.
Crosby is explicitly structured as a vertically integrated AI-powered law firm. It combines AI agents with licensed lawyers, accepts responsibility for finished work and sells services rather than software seats. Its focus began with contract review, but it is already experimenting with agents that model counterparties, negotiate and let clients interact directly with work in progress. The company raised a $60 million Series B.
Norm has taken the model into enterprise legal and regulatory work. After building regulatory AI software, it launched Norm Law, employs attorneys to supervise agents and charges customers based on outcomes rather than simply reselling attorney hours. Its $120 million Series C valued the company at $1.2 billion.
LegalOS is doing this in immigration. It says its agents were designed using 12,000 successful visa petitions and can help produce filing-ready applications in as little as 48 hours, with licensed attorneys reviewing and signing the submissions. Vector Legal is applying the model to startup corporate law, while Manifest OS raised $60 million to build technology-backed law firms around fixed and outcome-based pricing.
The pricing advantage follows directly. If automation turns five hours of work into thirty minutes, an hourly firm loses billable inventory while a fixed-price firm expands its margin. Crosby sells productized legal work, Norm charges based on outcomes, and Manifest OS is explicitly building around fixed and outcomes-based pricing. Eudia has reported customers shifting from billed hours toward billed outcomes, including a case in which Duracell reported a 50% reduction in contracting costs.
Thomson Reuters' survey of legal professionals found that 71% of in-house lawyers expect outside firms to change their commercial models as AI use grows, while only 28% of firms said they had actually changed pricing in response.
| Model | Example | Customer buys |
|---|---|---|
| Legal software | Harvey, Legora | Technology access |
| Workflow platform | Ironclad, Wordsmith | Automated legal process |
| AI-enabled service | Eudia | Technology plus managed expertise |
| AI-native law firm | Crosby, Norm, LegalOS | Finished legal work |
If you want more recent data on this point, please see our latest Legal Tech market report.
Is in-house LegalTech becoming more important than law-firm LegalTech?
In-house LegalTech is becoming at least as strategically important as law-firm AI because corporate legal departments control both their own workload and billions of dollars of external legal spending. Startups that automate internal work can therefore create value twice: by reducing internal effort and by preventing matters from ever reaching an outside firm.
Wordsmith's entire operating-system thesis is built around keeping more work inside the company. Eudia combines AI with legal expertise for large enterprise departments. Checkbox automates requests flowing from business employees to legal. Ironclad connects contract data across legal, procurement, finance and sales.
Thomson Reuters found that 61% of corporate legal professionals face some or significant internal pressure to adopt AI faster. Another 32% said they had already reconsidered, or expected to reconsider within twelve months, relationships with external firms that fail to demonstrate AI-enabled value.
Some LegalTech vendors are therefore helping clients avoid outside legal spend, not merely helping law firms work faster.
This chart, featured in our Legal Tech market deck, breaks down revenue across customer segments in the Legal Tech market
Are general AI companies becoming a threat to LegalTech startups?
General AI companies are now a serious competitive threat to LegalTech startups, which is why legal-specific workflow and data have become so important. The risk is already visible: Google has launched Gemini Enterprise for Legal, while OpenAI has been moving deeper into legal workflows and integrations.
A generic assistant can already summarize documents, draft clauses, compare text and perform basic research. As frontier models improve, the amount of differentiation available from simply prompting a model with "act like a lawyer" approaches zero.
That pushes LegalTech startups toward harder layers. Harvey is developing models, evaluations, agents and embedded legal engineering. Legora is connecting private knowledge, document systems and execution tools. Ironclad owns contract workflow and a repository built around billions of agreements. Supio and EvenUp specialize around case structures and domain-specific datasets.
The defensible layer is increasingly customer context, regulated workflows, legal-content licenses, matter-level integrations and distribution inside firms.
Is accuracy still the biggest obstacle to autonomous legal AI?
Accuracy remains a hard constraint, but the problem has evolved from "does the model hallucinate?" to "can the entire workflow be trusted?" A legal agent might retrieve the correct cases and still apply the wrong standard, miss an exception, use outdated company policy or make a judgment that no experienced attorney would approve.
The Stanford research showing hallucination rates above 17% for tested professional legal research systems demonstrated why retrieval alone cannot eliminate the issue. Newer systems increasingly respond with citations, source links and explicit human-review stages, but traceability does not automatically make the underlying reasoning correct.
That is why current LegalTech architectures retain lawyers at decision points. Legora's agents return control where judgment is required. Supio describes attorney oversight and approval. Crosby and LegalOS use licensed attorneys to approve finished legal work. Google's general counsel has similarly argued that AI can automate substantial legal work without replacing human judgment.
For now, autonomy is expanding around the lawyer rather than removing the lawyer from the highest-risk decisions.
This chart, featured in our Legal Tech market deck, shows how AI contract review platform technology has evolved over time
Will LegalTech reduce the number of lawyers?
LegalTech is much more likely to reduce the number of human hours required per matter than eliminate lawyers outright. That distinction matters because automation can shrink some jobs even while total demand for legal work remains large.
The tasks most exposed today are precisely the ones junior lawyers, paralegals and legal-operations staff have traditionally performed: document review, first drafts, research synthesis, chronology building, intake, contract comparison and routine diligence. An agent that completes those steps in minutes removes real labor demand even when a lawyer still approves the answer.
Thomson Reuters found that 48% of legal professionals worry about the effect of AI on the development of independent judgment. Its respondents estimated that the path toward trusted professional judgment could lengthen by almost two years for lawyers if traditional training work disappears.
The sharper risk is the erosion of the junior work through which firms historically trained senior lawyers.
What is the most important LegalTech product being built now?
The most important LegalTech product currently being built is an AI execution layer that can understand a matter, access trusted information, use an organization's own knowledge, perform multi-step legal work and know when a lawyer must intervene. Nearly every major startup trajectory is pointing toward some version of that architecture.
Harvey and Legora are pursuing it horizontally across sophisticated legal work. Wordsmith, Eudia and Checkbox are building it around corporate legal departments. Ironclad is building it around contracts. Supio and EvenUp are building it around plaintiff litigation. Crosby, Norm, Manifest OS and LegalOS are putting the same architecture inside the law firm itself.
Earlier legal software digitized individual activities. The emerging generation is trying to connect those activities into one continuously executing system.
If you want more recent data on this point, please see our latest Legal Tech market report.
In our Legal Tech market deck, we identify pain points entrepreneurs should prioritize
So what are LegalTech startups really building now?
LegalTech startups are now building the operating infrastructure for AI-mediated legal work, and the strongest evidence suggests that the market is moving decisively beyond standalone copilots. The central competition is becoming who controls the legal workflow rather than who generates the best paragraph.
The clearest winners so far fall into three groups. Harvey and Legora are racing to become broad operating systems for lawyers. Vertical platforms such as Ironclad, Supio and EvenUp are using deep workflow and proprietary domain data to own specific categories. A newer group including Crosby, Norm, LegalOS and Manifest OS is going one step further and rebuilding the law firm itself around software.
A generic legal chatbot looks increasingly difficult to defend as a standalone category because foundation-model companies can provide much of the underlying drafting and reasoning. Startups therefore need to own something harder: authoritative legal data, proprietary matter context, an end-to-end workflow, distribution inside a legal organization or the final legal service.
LegalTech is moving toward systems that perform much of the work themselves and bring lawyers in when judgment is required.
OUR METHODOLOGY
This analysis asks a simple question that becomes messy very quickly: what LegalTech startups are actually building now? We separated the market into the main places where companies are competing — funding, product architecture, workflow ownership, legal knowledge, integrations, service delivery, pricing and human review — and compared those dimensions rather than treating every company with an AI label as the same kind of business.
We prioritized recent, checkable evidence showing what companies are actually doing. That included financing rounds, product launches, acquisitions, integrations, customer adoption, ARR disclosures, workflow expansion and legal-AI evaluations. Funding and valuation were used as measures of investor conviction and scale, not as proof that a product category will win.
Companies were grouped primarily by the part of legal work they are trying to control. That gave us a consistent way to compare broad platforms such as Harvey and Legora with contract systems, litigation products, in-house legal tools, knowledge infrastructure and AI-native legal services.
Freshness was important because LegalTech is changing fast enough that a recent product launch can say more about a company's direction than the category it occupied a year earlier. We gave newer product, commercial and strategic developments more weight when they materially changed the picture, while keeping older research where it provided a useful benchmark.
We also looked for patterns that appeared across several independent types of evidence. The conclusion that LegalTech is moving toward an AI execution layer did not come from one company or one funding round; it came from the same architecture showing up across agents, integrations, institutional knowledge, contract intelligence, litigation workflows, legal research and AI-native law firms.
Key sources used for this analysis include Crunchbase on LegalTech funding and market direction, Harvey on its latest financing and adoption, Harvey on Tenet, Harvey on its Legal Agent Benchmark, Legora on its Series D, Legora on its ARR and agentic workflow expansion, Wordsmith on its Series B and customer growth, Checkbox on the AI Legal Front Door, Ironclad on agentic contract workflows, Supio on its end-to-end agentic platform, Eudia on Expert Digital Twins, Clio on its acquisition of vLex, Stanford Law School on legal-research reliability, Crosby on its AI-native law-firm model, Manifest OS on technology-backed law firms, Thomson Reuters on AI adoption, pricing and talent effects, and Google Cloud on Gemini Enterprise for Legal.
This chart, featured in our Legal Tech market deck, breaks down regional revenue across Europe, Asia, North America, Africa, and South America in the Legal Tech market
Related blog posts
- Legal AI: what’s changing now?
- Legal AI: what is getting real adoption now?
- Legal AI: what is actually working now?
- The latest funding news in the Legal Tech market
