Lawyers spend a large share of their billable workday in Outlook: drafting client emails, responding to opposing counsel, coordinating with co-counsel, fielding partner check-ins. Microsoft's April 15, 2026 Copilot release embedded a new lawyer-specific drafting capability inside Outlook: suggested responses with status updates and key links, grounded in the matter context Microsoft Graph already pulls from across Word documents, Teams discussions, and SharePoint repositories. The capability lands at $30/user/month as part of the standard Copilot for M365 enterprise add-on. For a 25-attorney firm where the partners and senior associates write 40-60 client emails per day, this is the highest-volume Copilot use case before contract review. Here's how the workflow runs, where it earns its keep, and where the privilege risks land that firms have to govern.
What the Outlook drafting capability actually does
Open an inbound client email in Outlook. Click the Copilot pane in the reading pane or compose window. Ask Copilot to draft a response: the prompt can be as specific as "draft a status update referencing last week's deposition prep" or as generic as "draft a reply." Copilot does three things in roughly 5-15 seconds:
1. Reads the matter context. Microsoft Graph pulls relevant context from across the firm's Microsoft 365 tenant: recent Word documents tagged to the matter, prior emails in the thread, Teams discussions referencing the client or matter, SharePoint files in the matter folder. The grounding happens at the tenant boundary, not by sending data to a third-party model. 2. Drafts a context-aware response. The output references prior matter activity ("as we discussed in Tuesday's call"), surfaces relevant attachments ("I've attached the revised draft we worked through last week"), and includes status update language matched to the matter's current phase. 3. Surfaces key links. When the response references a document, Copilot includes the SharePoint link inline. When it references a meeting, it links to the Teams recording or transcript. The links resolve based on the recipient's existing tenant access, clients on shared SharePoint libraries get working links, external recipients without access see resolved attachments instead.
The operational improvement over manual drafting is larger on standard responses (status updates, document confirmations, scheduling coordination) than on substantive responses (legal analysis, negotiation positioning, settlement discussions). The accuracy is high enough to ship without partner review on routine status updates, but every substantive client communication still requires the senior attorney's read before send.
Why Microsoft Graph integration is the differentiator vs ChatGPT Enterprise
ChatGPT Enterprise can also draft client emails. So can consumer Claude or Anthropic's Cowork legal plugin. The differentiation isn't the drafting quality, all three foundation models produce competent professional email drafts. The differentiation is the matter-context grounding.
When a partner asks ChatGPT Enterprise to draft a status update, ChatGPT has zero context about the matter unless the partner manually pastes it in. The partner spends 3-5 minutes loading prior emails, document references, and matter notes into the prompt before getting a useful draft. The drafting itself takes 30 seconds. AI drafting saves time versus manual drafting, but the amount depends on the email and how much you edit.
When the same partner asks Copilot, Microsoft Graph has already indexed the matter context, every email in the thread, every document tagged to the matter, every Teams meeting that referenced the client. Copilot grounds in that context automatically. The partner skips the 3-5 minute context-loading step. The total time savings versus manual drafting depend on the email type and how much you review. The differentiation is the install-base integration, not the model.
The second-order effect: every email Copilot drafts is also tagged with matter metadata in the firm's tenant. The audit trail of "which emails referenced which matters" is automatic. For ethics committee review, conflicts checks, and matter-file retention, that's a meaningful compliance improvement over the current state where matter-tagging happens manually or not at all. The third-order effect: when associates leave the firm, the matter context they accumulated in their Outlook stays in the firm tenant. Copilot grounds new associates in the same context the departed associate had access to. Knowledge transfer becomes mechanical, not manual. The Microsoft Graph firm knowledge management spoke covers the indexing architecture.
The privilege risk surface: what governance has to cover
Copilot's Outlook drafting is grounded in your firm's tenant, which means it's grounded in privileged matter context. Microsoft does not train its foundation models on tenant data under enterprise SKUs, that's contractually committed under the Microsoft Customer Agreement and the Online Services Data Protection Addendum. The privilege risk isn't training-data leakage. The privilege risk is misuse at the prompt layer.
Three concrete scenarios firms have to govern:
- Cross-matter prompts. An associate working on Matter A asks Copilot a question that, in the response, surfaces context from Matter B because both matters share a tagged client. If Matter B's content is privileged for a different attorney's work, the cross-surface creates a privilege issue. Mitigation: configure matter-level access boundaries in SharePoint and Microsoft 365 sensitivity labels. Most firms ship this configuration as part of the Copilot deployment.
- External recipient drafting. Partner asks Copilot to draft a response to opposing counsel. Copilot's draft references internal strategy notes that shouldn't be visible externally. Mitigation: train associates and partners on prompt review before send. Don't rely on Copilot to know which context is internal-only.
- Conflict-check completeness. Copilot grounds in the matter context the firm has tagged. If the conflict-check process missed an adverse-party connection, Copilot may surface context that creates a conflict issue. Mitigation: integrate Copilot rollout with the conflict-check workflow review. The conflict-checks privileged information isolation spoke covers the configuration.
United States v. Heppner (SDNY, February 17, 2026) ruled that communications between a defendant and consumer Claude were not protected by attorney-client privilege, the defendant used the consumer product, and the court concluded Claude isn't an attorney. That ruling addressed consumer AI specifically. Enterprise Copilot under M365 has materially stronger protections, but the privilege analysis for any specific matter still depends on how the firm configured deployment. The attorney-client privilege analysis for Copilot covers the firm-policy implications.
Where Outlook drafting wins: and where it doesn't
Wins:
- Status updates to clients. "Here's where the matter stands, here's what we filed last week, here's what's coming next." This is the highest-volume use case in transactional and regulatory practices. The time reduction depends on the email type and on how much you edit.
- Document confirmation emails. "Confirming receipt of your draft, we'll have comments back by Tuesday." Copilot drafts these in 5 seconds with appropriate references to the document.
- Scheduling coordination. "Proposing 2pm Thursday or 10am Friday for the deposition prep call." Copilot integrates with Outlook calendar to suggest available windows.
- Internal updates to partners. Associate drafting a memo to the supervising partner gets a structured summary grounded in the matter activity.
Losses:
- Negotiating positions. Copilot drafts what's plausible, not what's strategically optimal. Negotiation language requires the senior attorney's judgment on what to concede vs hold.
- Difficult client communications. When the matter is going badly, the email drafting is the easy part, knowing how to frame the bad news is the work. Copilot doesn't replace that judgment.
- First-instance legal analysis. "Here's our position on the indemnification dispute." Copilot can draft the structure but the analysis depth requires the attorney's expertise.
- Privileged matter strategy emails. Anything that involves the firm's internal litigation strategy or matter-handling decisions should be drafted by the attorney, with Copilot used (if at all) only for structural improvements after the substance is set.
The operational rule: use Copilot for the high-volume, low-stakes, contextually-grounded email work. Don't use Copilot for the low-volume, high-stakes, judgment-driven email work. The distinction is the same one firms apply to delegating work to junior associates, except Copilot delegates faster and cheaper. The Copilot ROI vs Cowork vs Harvey comparison covers the per-use-case math.
The bottom line. My take: Outlook drafting is the second-highest-value Copilot capability for law firms, behind Word contract comparison and ahead of Teams deposition summaries. The Microsoft Graph matter-context grounding is what makes it work, without it, Copilot's drafting quality would be no better than ChatGPT Enterprise. With it, Copilot saves time on email drafting, more on routine work than on substantive work. The privilege governance is real but tractable: configure matter-level access boundaries, train partners on prompt review, integrate with conflict-check workflows. Most mid-market firms recover the $30/user/month Copilot cost on Outlook savings alone in the first 60 days of deployment.
AI-Assisted Research. This piece was researched and written with AI assistance, reviewed and edited by Manu Ayala. For deeper takes and the perspective behind the research, follow me on LinkedIn or email me directly.
Price sources, checked 22 October 2026: Microsoft 365 Copilot. Each figure above is a list price from the vendor's own pricing page, or arithmetic on one. Plans, seat minimums and promotions change; confirm on the vendor's page before you buy.