Scott Stevenson is taking Spellbook upstream of legal review, where agreements surface, context scatters and the first consequential decisions about risk and escalation are already being made.
A legal department can design a disciplined contract process and still miss the work that never enters it.
That gap—not faster redlining—is the real target of Spellbook’s move into Autonomous Contract Management.
Scott Stevenson describes the system as “setting the table” for the lawyer. In his telling, the agreement should be found, the relevant communications gathered, the language measured against company standards and the meaningful exceptions surfaced before legal is asked to decide.

That sounds like preparation, but it is also triage.
Spellbook may determine which playbook applies, how a departure should be characterized, who owns the next step and whether the matter warrants legal attention at all. By the time a lawyer enters, the system has already shaped the decision before them.
That is the consequential part of Spellbook’s expansion beyond the contract-review product that established the company.
Spellbook launched its generative AI contract-review tool in 2022 and built its early position inside Microsoft Word, where transactional lawyers already worked. It now reports more than 4,500 legal teams across more than 80 countries. Autonomous Contract Management extends the company from the document in front of the lawyer to the work that determines when—and under what conditions—the document becomes legally valid.
The product is not yet broadly available. Spellbook says ACM is in early access and rolling out to select teams. Its stated design spans three stages: intake and first-pass review, an integrated workspace for active negotiations, and a repository that keeps signed agreements searchable and monitored for renewals and emerging risks.
The lawyer retains approval authority. Spellbook says the system handles repetitive work while lawyers remain responsible for legal judgment, approvals and final decisions.
But human approval at the end does not make the preceding automation neutral.
Before a lawyer sees the matter, the system may already have captured the agreement, selected the standard against which it will be reviewed, identified the deviations and decided whether the contract is ready for approval or requires closer attention.
The decision not to escalate can be as consequential as the advice given after escalation.
Stevenson’s answer is that those decisions are already being made throughout the business, only less consistently and with considerably less visibility.
Legal departments do not review every agreement their companies enter. At some organizations, he said, contracts below a substantial financial threshold never reach legal. That may be a rational response to volume, but it leaves the department with an incomplete view of the contractual commitments accumulating across the enterprise.
The contract does not become risk-free because no lawyer reviewed it. The risk simply remains somewhere else in the business.
Stevenson believes AI can extend legal’s standards across a broader field of activity without requiring a lawyer to inspect every document. The system can conduct the first pass, separate routine matters from exceptions and escalate the smaller number that appear to require judgment.
He compares the model to an organizational immune system: continuously operating in the background, covering more activity than a person could consciously monitor and calling attention to conditions that require intervention.
The analogy carries its own warning. An immune system is useful because it can distinguish the ordinary from the dangerous. When it draws that line poorly, it can miss a threat or attack what should have been left alone.
Spellbook’s value will depend on the same distinction.
The company’s argument begins with a familiar weakness in contract lifecycle management. Legal and legal operations teams have spent years configuring intake forms, approval paths, repositories and workflows intended to bring commercial contracting under control.
Those systems govern the matters that enter them.
They depend on an employee recognizing that legal work has begun, choosing the correct process and supplying the information the workflow requires. Meanwhile, agreements continue to originate in email, messaging platforms, sales systems and procurement channels that the legal department does not control.
“We come to where users already are,” Stevenson said.
Spellbook’s public launch materials describe ACM pulling contracts from email, Slack and Salesforce. Stevenson also cited Microsoft Teams in the interview. The system is intended to capture the agreement, organize the surrounding details, run an initial review against company standards and route the matter to the appropriate pipeline and reviewer.
Stevenson does not treat the persistence of informal contracting as a training failure. He treats it as a design constraint.
Traditional CLM generally required the business to bring the work into a controlled process. Spellbook wants the intelligence to reach the work first.
That does not place Spellbook outside the CLM market. Intake, review, workflow, storage and monitoring are established contract management functions. The distinction Stevenson is drawing lies in the company’s starting point.
Many CLM platforms began with process and are now embedding more substantive AI. Spellbook began with AI-assisted review and is extending outward into process.
Its strategic claim is that intelligence should help determine the appropriate workflow rather than merely operate inside one configured in advance.
A system that can interpret the agreement, the surrounding context and the company’s playbook may be able to determine whether a matter is routine, which exception matters, who has authority and where legal judgment belongs.
That is a credible theory of the market. It is not yet a proven operating advantage.
Spellbook must reliably recognize that contract work has begun, identify the correct document, preserve relevant communications, apply the appropriate standards, respect access restrictions and escalate consistently. Those requirements become more demanding as the product moves further ahead of direct legal supervision.
A lawyer can catch a poor suggestion in a contract already under review. A matter the system classifies as routine may never receive that scrutiny.
The more successfully Spellbook operates before legal becomes involved, the more the department must trust decisions it did not directly observe.
The value of setting the table depends on what the system leaves off it.
Stevenson argues that the comparison should be made against the current operating reality, not a theoretical process in which legal sees everything. An agreement that never enters the intake system is not controlled merely because legal designed a workflow for it.
That is the opening Spellbook is pursuing.
The company wants to move beyond helping lawyers produce faster redlines and into the interval before legal review, where commercial work is identified, organized and judged important enough to surface.
For in-house legal teams, the promise is greater reach: A system that reviews more agreements and flags exceptions can expand legal’s influence without making the department a bottleneck.
The risk is less visible. The same system that expands coverage will also make judgments about relevance, priority and escalation before a lawyer arrives.
Spellbook wants legal to begin with a prepared matter rather than a search for the information required to understand it.
The test is whether the system can expand what legal sees without making its most consequential decisions harder to see.
—
Scott Stevenson is co-founder and CEO of Spellbook. A software engineer who has spent more than 15 years building technology products, he started the company after frustration with the cost and turnaround time of legal work at an earlier venture. Spellbook launched its generative AI contract-review product in 2022 and serves law firms and in-house teams worldwide, with Stevenson leading its expansion into contract management.
