The Building Blocks of AI-Driven Legal Workflows
The nine building blocks of every legal workflow, what AI adds to each, and a worked marketing review example.
Last updated November 2026
Every legal workflow, from an NDA to a marketing review to a privacy request, is built from the same parts: something starts it, rules decide where it goes, people do tasks and approve things, documents get produced, everyone hears what happened, and a record is kept. Learn the parts once and you can design, read or fix any workflow.
The nine building blocks
1. Trigger. What starts the workflow: an email to legal, a Slack message, a form, a contract reaching renewal. AI adds: recognizing a request in free text, so nobody has to find the right form.
2. Intake. The details the workflow needs: counterparty, value, deadline, documents. AI adds: pulling details from the message and asking only for what's missing.
3. Conditions. Rules that decide the path: over a value threshold, add finance; personal data involved, add privacy. AI adds: classifying the request so the right conditions fire.
4. Tasks. Work someone does, each with an owner: review a draft, gather information, negotiate. AI adds: first-pass review against a playbook, and summaries of long documents.
5. Approvals. Sign-offs from finance, security, a senior lawyer or an executive. AI adds: a drafted approval summary, so approvers can decide without reading everything.
6. Documents. What the workflow produces: an NDA from a template, a redline, a memo. AI adds: generation from intake answers and drafting from your clauses.
7. Notifications. Who hears what, when. AI adds: answering status questions on demand in Slack or email.
8. Deadlines and escalations. Target times per step and what happens when one is missed. AI adds: early warnings and flags on stuck work.
9. Record. The request, decisions, documents and approvals, saved together. AI adds: finding precedent when a similar request arrives.
A worked example: the marketing review
Marketing is a good test case because the stakes are visible. In Streamline's 2026 Legal Friction benchmark, 69% of marketing respondents said a legal delay had cost their company a deal or caused one to slip, the highest of any department, and 77% said they bypass formal legal channels at least occasionally. A marketing manager summed up the experience: "The process is very slow, and it takes too much time to get simple contract approvals."
Here is a campaign review built from the nine blocks:
- Trigger: a marketer posts in the legal Slack channel asking for a review.
- Intake: AI asks for the asset, where it runs, the launch date, and whether it includes claims, comparisons or a contest.
- Conditions: a contest adds a sweepstakes rules check; a comparison claim adds substantiation review.
- Tasks: the assigned lawyer reviews, with AI flagging claims that need support.
- Approvals: for contests, the head of marketing confirms the prize budget.
- Documents: official rules generated from a template, if needed.
- Notifications: the marketer sees each status change in the same Slack thread.
- Deadlines: review due three business days before launch, with an alert at two.
- Record: the approved asset, comments and rules saved for the next campaign.
To build a first workflow step by step, see the practical guide to legal automation. To find where an existing one loses time, see optimizing legal workflows.
Result from a team that built theirs
Lighthouse had its forms and workflows running within a month, and within two months cut time spent tracking down information by 40 to 50%.
Which tools cover which blocks
Streamline AI runs all nine blocks for business requests to legal: intake and classification, your conditions and approvals, template documents, notifications and the full record, with dashboards on time at each step. For contract-specific work such as obligation tracking across a large portfolio, teams usually add a CLM. Book a demo.
Evisort covers contract workflows: AI term extraction, obligation tracking, approvals and renewals. GC AI helps with the task block, as an assistant for in-house lawyers that quotes source text exactly. Vincent AI handles research tasks. Eudia builds customized AI programs for large departments; ask which blocks it covers out of the box.
Where to start
Draw your three most common workflows using the nine blocks. Mark where work waits and where AI could take a step off someone's plate.
.avif)