Compliance & Legal

How to send an nda from a template with AI

Compliance & Legal3 AI tools7 steps6 friction points

Want an agent to handle this on your live data? See the Starch version →

Sending an NDA from a template sounds simple — find a standard form, fill in the names and dates, get signatures, file it somewhere. In practice, it touches three or four different tools, requires tracking who has signed and who hasn't, and tends to happen right before a meeting or a deal that can't wait. Most operators have a folder of NDA drafts that are slightly different from each other and a vague sense that this process could be tighter.

The workflow feels like an AI problem because most of the friction is language work: adjusting jurisdiction clauses, swapping out mutual versus one-way language, making sure the disclosing and receiving party definitions are accurate for this specific deal. That's exactly the kind of structured editing that LLMs handle well. The idea of describing what you need and getting back a ready-to-send document is genuinely appealing — and not entirely wrong.

ChatGPT, Claude, and Gemini can draft a clean NDA from a template, adapt clauses for specific contexts, and flag language that may need legal review. Claude in particular handles long document editing precisely. What these tools can't do is send the document, collect signatures, track completion, or remember what you sent to whom last quarter — every run starts fresh from whatever you paste in.

Skip the manual work → See the Starch version

Compliance & Legal3 AI tools7 steps6 friction points

AI walkthrough

How to do it with AI today

A practical walkthrough using ChatGPT, Claude, and other off-the-shelf LLMs — what they're good at, what you'll have to do by hand.

Tools that work for this

ChatGPTClaudeGemini

Step-by-step

1 Find or write a base NDA template — either one you already use or a standard mutual NDA you've verified with counsel. This is your starting document; the LLM will adapt it, not generate it from nothing.

2 Open Claude or ChatGPT and paste the full template text along with a description of the deal context: who the parties are, whether it's mutual or one-way, the governing jurisdiction, and the confidentiality period you need.

3 Prompt the model to fill in the party names and addresses, adjust the direction of disclosure (mutual vs. one-way), and flag any clauses that may need jurisdiction-specific changes for your state or country.

4 Review the output side by side with your template. LLMs occasionally reword standard protective clauses in ways that change meaning — check the definition of Confidential Information and the exclusions section specifically.

5 Copy the finalized document into Google Docs or Word and format it for signature. At this point you leave the AI entirely — the LLM has no way to send or track the document.

6 Use a separate e-signature tool (DocuSign, HelloSign, or similar) to send the NDA to the counterparty. Log the send and expected return date in whatever system you use to track open agreements.

7 When the signed copy comes back, save it to your document folder and update your tracker. If the counterparty doesn't return it, you'll need to follow up manually — there's no automated reminder in this workflow.

Prompts you can copy

Here is my standard mutual NDA template: [paste text]. Adapt it for a one-way NDA where [Company A] is the disclosing party and [Company B] is the receiving party. Governing law is New York. Confidentiality period is 3 years.

Review this NDA clause by clause and flag any language that is unusually broad, ambiguous, or likely to cause friction with a counterparty's legal team. Don't redraft — just annotate the issues.

Fill in the blanks in this NDA template: Disclosing Party: Acme Corp, a Delaware LLC. Receiving Party: Beta Partners Inc. Effective date: June 1, 2025. Purpose: evaluating a potential commercial partnership. Keep all other language exactly as written.

This NDA uses US standard language. The counterparty is based in the UK. List the specific clauses that may need adjustment for English law, and suggest replacement language for each.

Draft a short cover email to send alongside this NDA. Tone should be professional but direct. Include a request to return a signed copy within 5 business days.

Reality check

Where this gets hard

The walkthrough above works — until your numbers change, the LLM hallucinates, or you have to re-paste everything next month.

No memory of past NDAs — every conversation starts blank, so you're re-pasting your template and re-explaining the deal context each time you send one.

The LLM cannot send the document, trigger a signature workflow, or track whether the counterparty has signed — those steps happen entirely outside the AI, in separate tools.

Clause edits drift between sessions. The wording the model chose for the confidentiality definition last month may differ subtly from what it produces this month with the same prompt.

No connection to your contacts, CRM, or calendar — you're manually copying party names, addresses, and deal context rather than pulling from a live source where that data already lives.

There's no audit trail. If a counterparty later disputes what was agreed to, your record of what was sent is a downloaded Word doc and a sent email, not a timestamped signing log.

Following up on unsigned NDAs is entirely manual — you have to remember to check, no reminder surfaces in the AI tool, and nothing flags agreements that have been outstanding for two weeks.

Tired of the friction?

Starch runs the whole workflow on live data — no copy-paste, no hallucinated numbers, no re-prompting next month.

See the Starch version →

Starch alternative

The same workflow on Starch

Starch is an agentic operating system — an agent builds and runs the software for this workflow against your live business data, so sending an NDA from a template becomes a repeatable process rather than a one-off prompt session.

Contract Lifecycle Management (coming soon) will handle NDA creation, e-signature routing, and storage in one place — request beta access now to be notified at launch.

Connect Gmail or Outlook through Starch's integration catalog; the agent can draft and send the NDA cover email with counterparty details pulled from your existing contact data, without a separate copy-paste step.

Describe your NDA workflow in plain English — 'when I start a new deal, generate an NDA from my standard template, pre-fill the party details, and send it for signature with a 5-day follow-up reminder' — and Starch builds that automation.

Knowledge Management stores your canonical NDA template and clause library so every generated document pulls from the approved version, not whoever's copy was most recently saved to a shared drive.

Starch automations run on a schedule or triggered by events — an unsigned NDA outstanding for 7 days can trigger a follow-up email automatically, without you building a reminder into your personal calendar.

Because Starch persists the workflow as a real application rather than a prompt you re-run, a new team member can send an NDA using the same process without needing to know how you originally set it up.

Get closed-beta access →

Toolkit