Spellbook's Legal AI: Drafting Contracts in Word, and Operationalizing It Across a Legal Team
Contracts are where a huge share of legal work actually happens, and most of it is slow. A lawyer opens a draft, reviews it clause by clause against a mental model of the firm's positions, marks it up, hunts for precedent, and rewrites language they have written a hundred times before. The judgment is valuable. The mechanical drafting and first-pass review around it are not, and they consume enormous amounts of expensive time.
A company called Spellbook has built AI aimed directly at that work. Its copilot drafts and reviews contracts inside Microsoft Word, where lawyers already live, and it has grown into one of the most widely adopted legal AI tools in the market. This is a deep look at what Spellbook does, how far it has scaled, and, just as important, what it actually takes to turn it from an individual productivity boost into firm-wide value. That second part is where legal AI investments quietly succeed or fail.
The problem Spellbook set out to solve
Legal drafting has a productivity ceiling set by how fast a human can write and check language. Much of a transactional lawyer's day is not novel reasoning. It is redrafting standard clauses, comparing a counterparty's paper against the firm's preferred positions, and searching old matters for the precedent that already solved this problem. The expertise sits idle while the expert does repetitive work, and the volume of contracts a team can turn around is capped by that friction.
Spellbook was built to remove it. The company began life as Rally, founded in 2018 in St. John's, Newfoundland, by CEO Scott Stevenson and his co-founders, before launching Spellbook in 2022 as one of the first generative AI contract-drafting tools. The design choice that made it spread was deceptively simple: rather than ask lawyers to learn a new platform, it lives inside Word as a copilot, powered by GPT-4 and other models fine-tuned on legal data, drafting and reviewing up to several times faster than manual work.
What Spellbook actually does
Spellbook is no longer a single feature. It has grown into a suite that covers the contract lifecycle, and understanding the pieces matters for understanding the adoption work later. According to the Spellbook product, the core capabilities include:
- Review that redlines contracts and flags risk directly in Word against the organization's standards.
- Draft that creates clauses and full contracts from scratch or from a saved precedent library.
- Compare that benchmarks a contract against thousands of similar industry agreements.
- Ask that answers complex contract questions with citations.
- Playbooks that encode a legal team's standards into reusable review frameworks.
- Associate, an AI agent for multi-document, coordinated drafting and review workflows.
- ACM (Autonomous Contract Management), an end-to-end system for intake, review, and storage and indexing.
The through-line is that Spellbook is trying to move up from a drafting aid toward a system that manages contracts end to end. It is enterprise-serious about the plumbing too, with SOC 2 Type II, SSO, zero data retention from its LLM providers, and support for GDPR, CCPA, and HIPAA across 80-plus countries.
Who benefits, and the proof in adoption
The direct beneficiaries are clear. Transactional lawyers get their time back for judgment instead of redrafting. In-house teams clear contract backlogs faster. Firms get more throughput from the same headcount, and more consistent positions across every deal.
The evidence is in the adoption. Spellbook reports more than 4,500 legal teams using the product, a 4.7-star G2 rating, and customers that include Dropbox, eBay, Asics, Fender, Crocs, and Franklin Templeton, alongside names like Nestlé and law firm Kennedys. In March 2026 it became the exclusive AI drafting partner for the Canadian Bar Association's roughly 40,000 members. One user summed up the individual value bluntly: it helps them bill an extra hour a day, maybe more.
The money and the momentum
Spellbook has moved from scrappy startup to category consolidator quickly. In October 2025 it raised a $50 million Series B led by Keith Rabois at Khosla Ventures, with participation from Threshold Ventures and existing investors, at a $350 million post-money valuation. Coverage from LawSites and Artificial Lawyer framed the round as fuel to expand from drafting into full contract review and management.
Then, in March 2026, the company added a $40 million debt facility from RBCx, the innovation banking arm of Royal Bank of Canada, explicitly to fund acquisitions in a consolidating legal AI market. That is a company playing offense, buying its way to a broader platform rather than defending a single feature.
Why this matters beyond legal
Spellbook is a clean example of a pattern across enterprise AI. The tools that win are not the general-purpose assistants. They are the vertical systems that go narrow, learn the rules and language of one high-value workflow, meet the professional inside the software they already use, and earn trust by showing their sources. Contract work is a near-perfect target: high volume, expensive expertise, document-heavy, and full of repeatable patterns.
It also points at the real bottleneck in enterprise AI, which is not the model. It is turning a capable tool into an organizational habit. An individual lawyer downloading a Word add-in is easy. A 200-lawyer firm or a global in-house team actually standardizing on it, with shared playbooks and consistent positions, is a different and much harder thing. That distinction, between a tool that individuals use and a capability an organization operates on, is the same shift reshaping how work gets delivered everywhere: away from renting effort and toward buying outcomes that are measured and proven. It is a theme we have written about as the future of work.
Downloading the tool is easy. Operationalizing it is the project
Spellbook's Word-native design means an individual lawyer can be productive in minutes. That is a genuine strength, and it is also where teams underestimate the work. Getting firm-wide value, the kind that shows up in throughput and consistency across every matter, is a project with several distinct pieces, none of which the software does for you.
1. Playbook and precedent configuration. Spellbook is only as good as the standards you give it. Encoding your firm's or legal team's positions, fallback provisions, and preferred language into Playbooks, and building out the precedent library it drafts from, is careful, expert work. Done well, every lawyer drafts from the same firm-approved positions. Skipped, everyone gets generic output and quietly reverts to their own templates.
2. Precedent and knowledge organization. The Compare, Ask, and precedent features are only as strong as the contract history behind them. Getting real value means organizing and connecting the team's existing agreements and matter history so the AI is reasoning over your knowledge, not just the public corpus. That is a data and knowledge-management effort.
3. Integration into the document and contract stack. For anything beyond individual drafting, value depends on Spellbook connecting to the systems the team actually runs on: the document management system, the contract lifecycle management platform, the intake process. The move toward Autonomous Contract Management only pays off when that integration is genuinely built, not demoed.
4. Firm-wide rollout and lawyer change management. This is the piece that decides success. Lawyers are trained skeptics being asked to trust AI on their craft, and adoption that stops at a few enthusiasts never changes the economics. Getting a whole team to standardize takes training, champions, a rollout sequenced to visible wins, and a feedback loop so lawyers' corrections improve the shared playbooks. The difference between 10 percent and 90 percent adoption is entirely here.
5. Measurement and continuous improvement. The value case has to be proven in the team's own numbers. Baseline drafting and review time, turnaround, and consistency before, measure them after on a defined workflow, and keep tuning the playbooks. Without that loop, you cannot prove ROI to the partners or the general counsel who approved the spend.
Why teams stall, and it is not the software
None of those five require a better model. The bottleneck is capacity and know-how, not capability. Standardizing a legal team on an AI platform is a cross-functional effort spanning legal, knowledge management, and IT, and most teams do not have people who understand both legal workflows and AI rollout sitting idle to run it. A fast-scaling vendor keeps its own people focused on product and the biggest accounts. The lawyers are busy lawyering. So the tool lands as a handful of power users, the firm-wide value never materializes, and the investment underdelivers.
This is not a Spellbook problem or a legal problem. It is the defining problem of enterprise AI, and it is why the old ways of adding capacity fall short. Hiring permanent staff for a one-time rollout is slow and expensive, and traditional outsourcing or offshoring rents bodies by the month rather than delivering an outcome.
Closing the gap faster
Teams that win with legal AI treat operationalizing it as the real project and resource it properly. That usually means a delivery layer that sits behind the vendor's product and the legal team and owns the last mile: building the playbooks and precedent libraries, organizing the contract knowledge base, integrating the document and CLM systems, running the lawyer change management, and standing up the measurement loop that proves ROI.
This is exactly the work a Virtual Delivery Center is built to deliver. AiDOOS runs elastic delivery pods that plug in alongside a product like Spellbook and the legal team's own people, take on the configuration and adoption work, and turn a promising tool into a firm-wide capability. The vendor keeps its focus on the product. The team gets to real, measured value without building a permanent function for a one-time push. And because the work is scoped as delivered outcomes rather than bodies rented by the month, you pay for progress, not effort.
The sensible way to start is small and provable: pick one contract type or one practice group, build the playbooks, get the team genuinely standardized on it, measure the before and after, and use that proof to roll out across the organization.
The bottom line
Spellbook has taken the slow, repetitive core of contract work and rebuilt it around AI that drafts and reviews inside the tools lawyers already use, and its funding and customer base show the market agrees. The technology is ready.
The question every legal leader should ask is not whether the AI drafts well. It is whether the team will actually operationalize it, standardize on it, and prove the value, or whether it will stay a clever add-in a few people use. Answer that well, and the throughput and consistency gains are real. Leave it to chance, and even a category leader like Spellbook becomes another tool that impressed in the demo and never changed the numbers. The model is no longer the hard part. Adoption is, and that is a solvable problem.
If you are rolling out Spellbook or any complex enterprise AI, AiDOOS provides the delivery pods that get it operationalized and proving ROI. To explore the product itself, visit Spellbook. For more on where enterprise delivery is heading, read our North Atlantic Briefing.