By Dorna Moini, CEO of Gavel Exec, Former Attorney at Sidley Austin and Lecturer in Law at USC School of Law.
Writing motions and agreements was my day-to-day, long before I ever touched AI. As a near decade-long practicing lawyer at Sidley, I spent years on cases where precedent was the whole job, whether I was pulling prior examples from iManage or printing and highlighting paper copies of Westlaw cases (dating myself!) That’s why, now that I run Gavel, a software platform trusted by thousands of legal professionals across the globe, I pay close attention to both vertical and horizontal tech tools used by lawyers.
And after testing Microsoft Copilot myself, talking with dozens of lawyers across practice areas, and watching the online debates play out, one thing is clear to me: even as Copilot rolls out broadly through Microsoft 365, lawyers still aren’t using it for the real legal work.
But why? Copilot is a great general assistant. It writes emails, summarizes content, and makes all of Microsoft Office more useful. But the work lawyers do, like contract drafting, redlining, analysis, and negotiation, is a different challenge entirely. The expectations are higher, the context is critical, and the application layer (what it actually does beyond giving you text back) is more important.
1. Copilot Is a Generalist, But Legal Work Is Not.
Copilot sits inside Word, Outlook, Teams, Excel, and PowerPoint, helping users summarize, rewrite, and organize content. That’s very useful, but it’s not legal reasoning.
Attorney at Work, in its Copilot review for lawyers, put it plainly:
“Copilot is a general-use tool… not a legal-specific one.”
– Attorney at Work: Microsoft 365 Copilot for Lawyers
Similarly, Lawyerist said:
“Copilot does not search caselaw… Copilot’s sources are not legal-specific” and it will “not replace a purpose-built, AI-assisted contract drafting or document review tool.”
– Lawyerist: Microsoft Copilot Review for Lawyers
This is the heart of the issue. Copilot is a great tool, but it’s not built by lawyers for lawyers. Legal drafting, contract analysis and review, and negotiation require:
- structured logic
- jurisdictional nuance
- defined terms that must stay defined
- cross-references that can’t break
- positions that have to match prior deal history
Legal AI tools, like Gavel Exec, are trained and given access to data specifically for contracts, deal documents, risk positions, clause patterns, and market norms, making the two tools complementary.
2. Copilot Can’t Take Real Actions Legal Professionals Need in Word.
This is one of the biggest misunderstandings I see in the legal market: the idea that a generic AI that produces text is the same thing as a legal AI tool.
Copilot sounds like it can redline contracts because it can generate language. But generating paragraphs is the easiest part of legal drafting. Copilot can:
- write text
- summarize a document
- give suggestions in a sidebar
But beyond the surface-level language work, it cannot take actions that are specific to lawyers, like:
- apply redlines across the document
- insert comments in the document, using judgement to decide the tone and when a comment should he accompanied by a redline
- follow a rules-based playbook with fallback positions, escalation logic, and firm-approved standards
- analyze a clause against market terms or identify deviations
- compare language to your own trove of precedent documents, including your historical deal positions
- enforce consistency across teams, matters, or practice groups
- negotiate multiple documents at the same time, keeping consistency across a deal set
Those aren’t just AI tools; they require a software application layer on top of the best AI. The most thoughtful legal AI tools, like Gavel Exec, aren’t just AI systems that spit out text. They are applications built using AI models, and they can do all of the above.
It’s not just that the underlying model has access to better data. It’s that the software layer is designed for one job: helping lawyers draft, negotiate, and standardize legal work.
3. Copilot Doesn’t Allow for Playbooks, So It Can’t Use Your Negotiation Strategy or Rules.
Playbooks are one of the most overlooked (but frequently used) tools in modern legal work. Long before AI entered the picture, playbooks in Excel were how teams kept negotiations consistent. They distilled a firm’s or department’s entire institutional memory into a set of practical, repeatable rules: what we accept, what we reject, when we compromise, and how we phrase the language.
Every well-run legal team already uses playbooks, even if they’re scattered across Word docs, SharePoint folders, old emails, or the head of one senior lawyer who “just knows” how things should go. Without playbooks, negotiations depend on who happens to be reviewing the contract that day. With playbooks, the team behaves like a unified organization.
Playbooks encode things like:
- fallback positions
- escalation thresholds
- internal policy constraints
- firm-approved language
- standardized negotiation points
In-house teams rely on playbooks to protect the business from inconsistent risk-taking across dozens of internal stakeholders. Law firms rely on playbooks to maintain quality across partners, associates, contract reviewers, and new hires.
Legal AI tools like Gavel Exec let lawyers directly encode their playbook rules into the system:
- “If the liability cap exceeds X → replace with our standard.”
- “If the governing law isn’t Delaware → flag and add our clause.”
- “If the indemnity structure changes → insert our approved fallback.”
- “If a required clause is missing → add it with a comment explaining why.”
And because these rules live in software, they can be shared across the entire legal team, from junior lawyers to partners, from in-house counsel to procurement, sales, and HR, giving everyone the same starting point.
4. Copilot Doesn’t Know What’s “Market”
One thing I’ve learned working closely with transactional lawyers, in practice and now at Gavel, is that what’s “market” isn’t theoretical. It’s a lived experience that changes. Every firm builds its instincts about what’s standard by handling hundreds of deals, arguing over clauses with counterparties, and seeing what actually gets signed. When we built Gavel Exec, we didn’t try to guess what market was. We went straight to the source.
We’ve worked hand-in-hand with law firms, who we hired to build our playbooks, and also understand what lawyers consider typical or aggressive in a large variety of corporate and real estate transactional contract types. Firms shared anonymized precedent sets, walked us through their redline philosophy, and explained where they draw the line on certain positions. Partners and associates reviewed early outputs and told us where the tool was too soft, too harsh, or simply misreading the intent of a clause.
This kind of calibration only happens when you build the product with the lawyers who will use it. General-purpose tools like Copilot don’t have the benefit of that domain-specific feedback loop. Exec does, and that’s why its understanding of what’s market aligns more closely with how lawyers negotiate in the real world.
6. Copilot Doesn’t Know Your Precedent or Deal Data
One of the biggest gaps between Copilot and true legal AI comes down to precedent. Deal lawyers don’t draft in a vacuum. They rely on years of accumulated knowledge: old agreements, the context of the client, and seeing hundreds of transactions play out.
Copilot can search across your Microsoft 365 environment, but that’s where its understanding stops. It doesn’t grasp how clauses evolve from deal to deal, or how related documents connect to one another. It can’t tell when a change is consistent with your historical positions or when it deviates from the language your firm typically uses. It can’t turn past deals into drafting intelligence. At the end of the day, it’s just retrieving text, not learning from it.
With Gavel Exec, precedent is central to how the software works. Lawyers can load their deal documents and have Exec analyze those documents before giving any response. When you negotiate a clause, the tool recognizes patterns from your prior agreements and suggests language you’ve used before in similar situations. As more documents are added and as you interact with it, Exec becomes better at recommending positions that align with your practice, reinforcing consistency across the team instead of reinventing language each time.
Real-World Usage: Lawyers Use Both, Just Not for the Same Work
Real world usage is a good indicator of why purpose-built legal AI tools like Gavel Exec are being adopted. If you look at what large firms are actually doing, they roll out Copilot firm-wide for general productivity. Then, they adopt legal AI tools for contract drafting, redlining, analysis, and negotiation.
A survey shared on Artificial Lawyer showed that the most widely used AI tools in firms are still legal-specific, not general-purpose assistants.
That’s why lawyers rely on legal AI tools, like Gavel Exec, for the actual legal work. Copilot is excellent at what it does. It’s just solving a different problem.
