Use CasesLegal
Counsel that keeps pace.
A mid-Atlantic commercial law firm that pairs partner-level judgement with an operations layer most firms its size cannot afford.
A firm this size cannot put a partner on every intake. Dave reads the contract first, rates it against the firm's own playbook, and only a high rating reaches the managing partner.
demo-legalIt can do the first read and the rating. In this demonstration tenant the Contract Risk Analyst extracts the parties, term, termination, liability caps, indemnification, IP ownership and governing law, rates each position aligned, negotiable or off-market, and writes an overall risk level onto the run. It is instructed to produce findings only, with no legal advice to end clients. A person still signs: the rating decides which person.
With a rule, rather than an inbox. A Routing node reads the risk level the AI analysis wrote onto the run: high risk goes to Partner Review, everything else goes to Associate Review. Both paths carry the same authority to end the run, so the rating changes who reads a contract, not whether anyone does.
Not a summary of them. These are the graphs as they are seeded: every step in order, its type, the agent on it, who it is assigned to, and where a rejection sends the run back to.
Guide: Build your first workflowEvery contract that arrives is read against the firm's playbook positions and rated. The rating, not a person's inbox, decides whether it goes to a partner or an associate.
7 nodes
Start
User Interaction
Matter name, client, counterparty and contract type. The type is a fixed list: MSA, NDA, SaaS agreement, employment, lease or other.
Agent Interaction
Extracts parties, term, termination, liability caps, indemnification, IP ownership, governing law and anything unusual, rating each aligned, negotiable or off-market, then writes an overall risk level.
Routing
High-risk contracts require partner review; everything else goes to associate review.
Human Review
The high-risk path. Approving ends the run.
Reject returns to Contract Intake
Human Review
The standard path. Same authority to end the run, a different pair of eyes.
Reject returns to Contract Intake
End
A research memo drafted by an agent, then checked by an agent for authorities that need verifying, then finalized by an associate. The cite-check is a workflow step with a vote, not a habit someone is trusted to remember.
6 nodes
Start
User Interaction
The matter, and the question to answer.
Agent Interaction
Question presented, short answer, analysis with authorities, open issues.
Agent Review
Every authority that a human must cite-check is flagged. Uncertain authority may never be presented as settled.
Reject returns to AI Research Draft
Human Review
Reject returns to AI Research Draft
End
The letter is drafted from the matter details, with every assumed detail marked for confirmation, and the responsible attorney is the one who issues it.
5 nodes
Start
User Interaction
Matter name and client.
Agent Interaction
Scope, fees, staffing, conflicts confirmation, termination and jurisdiction disclosures, emitted as an HTML artifact.
Human Review
Reject returns to AI Draft Letter
End
An agent in Dave is a configured worker, not a chat box: a written prompt, a pinned provider and model, and its own context. The line that matters most in each of these prompts is the constraint.
Guide: Create an agentExtracts key clauses from contracts and rates risk against firm playbook positions.
GuardrailFindings only. No legal advice to end clients.
Produces structured research memos with citations for associate review.
GuardrailFlag every authority that must be cite-checked by a human. Never present uncertain authority as settled.
Drafts client engagement letters from matter details using firm templates.
GuardrailMark any assumed detail with a confirm marker rather than guessing it.
Validates expense submissions against company policy: limits, categories, receipts and justification quality.
GuardrailAssess against policy. It does not approve or reject; the routing rule and the manager do that.
Five seats, and the Dave role each one carries. Role decides what a person can reach; the workflow decides what lands in their inbox.
The runs seeded into this tenant, and for each one still moving, the step it is waiting on.
Guide: Watch a run and read its timelineAn excerpt from a real run in this tenant. Read what it does and does not do: it grades every finding and hands the decision to the next node.
CONTRACT RISK ANALYSIS: Corville Health / Brightline SaaS Agreement Liability cap: 3 months fees INCLUDING data breach liability. [OFF-MARKET, critical for a health system]Data terms: PHI processing addendum missing; vendor claims HIPAA compliance without BAA attached. [OFF-MARKET]Termination: vendor may suspend for "suspected misuse" without notice. [OFF-MARKET]IP: usage data license to vendor is perpetual and sublicensable. [NEGOTIABLE]Governing law: vendor home state, unilateral. [NEGOTIABLE] OVERALL RISK LEVEL: HIGH. Routed to partner review. BAA must be executed before signature.
The same graph drawn above, as it is actually built in the Dave workflow designer.
Both lists as this workspace holds them: every workflow active, and every agent pinned to a named model.
The same team listed above, as it is configured in Tenant Settings.
A partner filters the run list, opens a matter, fills in the step the workflow asked for, and completes it. Nothing here is sped up or staged: it is the seeded Hartwell tenant being used.
This is the smallest of the three tenants and the clearest illustration of leverage. One routing node is the difference between a managing partner reading every intake and a managing partner reading only the ones rated high.
Every fact on this page is read from this tenant's seed data (packages/backend/prisma/demo-packs/legal.ts, with templates/expense-report.ts, demo-packs as of 2026-08-29) or from the firm's own brand pack. Nothing here is a customer story.
No. Hartwell & Associates LLP is a fictional firm we invented as a demonstration tenant. It is not a customer of HelloDave.ai, it has no website, and this page is not a case study. Every workflow, node, routing rule, agent guardrail and run on it is read out of the seeded demo-legal Dave tenant.
Not on their own, which is exactly why the cite-check is a workflow step here rather than a habit someone is trusted to remember. The research agent must flag every authority a human has to verify, and may never present uncertain authority as settled. An Agent Review node votes on the draft before an associate finalizes it.
It marks it instead of guessing. The Engagement Letter Drafter is required to put a confirm marker on any assumed detail, and the responsible attorney reviews the letter before it is issued.
Same canvas, same node types, your process. Put the first version on it and watch one run end to end.
