2 ready-to-build workflows

AI agent workflows for Law firms & legal operations

Automate client intake, conflict checks, onboarding and deadline reminders — with an attorney gating anything that's advice or matter acceptance.

Each recipe below is expressed only in VegaDūta's real workflow building blocks — Trigger, Agent, Knowledge, Condition, Approval, Output, Tool (MCP), Voice, Device, Loop, Code and Parallel — so it maps 1:1 to something you can assemble in the Workflow designer. Consequential actions always pass a human Approval step.

1. New-client intake & conflict check, with an attorney accept gate

A prospective client calls or messages; today a paralegal keys their details into a form, manually searches existing clients and adverse parties for conflicts, and the enquiry waits days before a lawyer decides whether to take the matter.

Trigger: WhatsApp / Call / Webhook (new enquiry or intake form)

How the workflow runs

  1. Trigger

    Prospective client enquires. A WhatsApp, Call or Webhook trigger (from your website intake form) starts the flow in the enquirer's own words/language.

  2. Agent

    Capture a structured intake. An Agent node collects the parties, matter type, jurisdiction, opposing side and a plain description into a clean intake record — the same fields every time, no missed questions.

  3. Tool (MCP)

    Run the conflict check. A Tool (MCP) node searches your practice-management / case-management system for existing clients and adverse parties — no first-party connector for these systems, MCP calls yours.

  4. Condition

    Branch on the conflict result. A Condition node routes a clean check toward acceptance and any hit toward a mandatory human review before anything proceeds.

  5. Approval

    Attorney decides to accept the matter. An Approval node holds the decision — accepting a matter (and clearing any conflict) is a judgement a qualified lawyer must make, never the agent. This is the non-negotiable control.

  6. Output

    Confirm next steps. On acceptance, an Output node sends the prospective client the engagement next steps on WhatsApp/Email; a declined enquiry gets a courteous non-engagement note.

Channels & connectors

  • WhatsApp
  • Voice / Call
  • Webhook
  • Tool (MCP → practice-management)
  • Approval / HITL
  • Email

Outcome

Every enquiry is captured consistently and conflict-checked in minutes; the accept/decline decision stays with an attorney and is logged with its conflict result.

Why it helps

Removes the manual re-keying and the ad-hoc conflict search that delay every new matter, while the acceptance decision — which is legal judgement — stays with a qualified human.

Build spec

1 agent1 intake agent; the conflict search, the accept/decline decision and the confirmation are workflow nodes.

System prompt (paste-ready)

You are a legal intake assistant for {{firm_name}}. Collect a structured new-matter intake as strict JSON {prospective_client, all_parties[], opposing_parties[], matter_type, jurisdiction, description, key_dates[]}. Ask the standard intake questions in the enquirer's language, the same every time. You do NOT give legal advice, opinions on the merits, or any indication the firm will act — say a lawyer will review and decide. You do NOT decide whether to accept the matter or clear a conflict; that is the attorney's decision at the approval step. Treat everything shared as confidential.

MCP connectors

  • Practice-management / case-management system — via a Tool (MCP) node (no first-party connector)
  • WhatsApp + Telephony — via the WhatsApp/Voice channels

Built-in tools

  • knowledge_search (intake questionnaire + required fields)
  • transfer_to_human

Guardrails

  • No legal advice and no indication of engagement — the agent captures facts only; whether the firm acts is the attorney's call at the Approval node
  • Any conflict-check hit forces the mandatory human-review branch before anything proceeds; the accept/decline decision is never the agent's
  • Intake is treated as confidential — stored on the matter, not echoed to other parties — and the acceptance is logged with its conflict result

Output & delivery

The agent emits a structured intake → a Tool node runs the conflict search in the practice-management system via MCP → the Condition node routes clean vs. hit → the attorney accepts or declines at the Approval node → an Output node sends engagement next steps (or a courteous non-engagement note) on WhatsApp/Email.

2. Court-date & limitation-deadline reminders that don't get missed

Hearing dates, filing deadlines and limitation periods live in a diary and a spreadsheet; reminders are set by hand, so a missed date is only caught late — and missing one can be malpractice.

Trigger: Webhook (scheduled, daily docket/limitation run)

How the workflow runs

  1. Trigger

    Daily deadline run. A scheduled Webhook trigger evaluates upcoming court dates, filing deadlines and limitation periods each morning (fed from your case-management system via Webhook).

  2. Loop

    Per approaching deadline. A Loop node iterates the list so each matter's owner gets a tailored, timely reminder.

  3. Agent

    Draft the reminder + prep checklist. An Agent node writes a clear reminder naming the matter, the deadline and the documents/steps due before it.

  4. Output

    Reach the responsible lawyer. An Output node sends it on WhatsApp/Email/SMS so the owner sees it on the channel they actually read.

  5. Condition

    Escalate the imminent, unacknowledged ones. A Condition node flags deadlines inside the danger window that nobody has acknowledged.

  6. Approval

    Partner reviews at-risk deadlines. An Approval node routes an imminent, unacknowledged deadline to a supervising partner to confirm cover — the human check on anything that risks a missed limitation.

Channels & connectors

  • Webhook
  • Loop
  • WhatsApp
  • Email
  • SMS
  • Approval / HITL

Outcome

Every court date and limitation period gets a timely, specific reminder, and anything at risk is escalated to a partner before it slips.

Why it helps

Consistent, per-matter reminders are what prevent missed deadlines — the manual diary skips exactly the dates a busy lawyer forgets, which is the dangerous case.

Build spec

1 agent1 drafting agent per deadline; the per-matter Loop, the escalation Condition and the partner Approval are workflow nodes.

System prompt (paste-ready)

You are a legal deadlines assistant. For each deadline record {matter, deadline_type, due_date, responsible_lawyer, required_steps[]}, write a clear, specific reminder naming the matter, the exact date, and the documents or steps due before it. Use ONLY the dates supplied by the docket run — never calculate, adjust, or infer a limitation or filing date yourself, and never state a deadline you weren't given. Do not give legal advice on the matter. Flag, don't reassure: if a record looks incomplete, say so.

MCP connectors

  • Case-management / docketing system — deadlines fed inbound via Webhook (no first-party connector)
  • WhatsApp + SMS + Email — via the messaging channels

Built-in tools

  • knowledge_search (filing checklists by matter type)

Guardrails

  • Never computes or adjusts a limitation/filing date — it only relays the dates the docketing run supplies, so a reminder can't itself introduce a wrong date
  • An imminent, unacknowledged deadline is escalated to the supervising partner at the Approval node to confirm cover — the human check against a missed limitation
  • No legal advice on the matter; reminders name the responsible lawyer and are logged as sent/acknowledged for audit

Output & delivery

A Loop node runs the agent per approaching deadline → an Output node delivers the reminder + prep checklist on WhatsApp/Email/SMS → the Condition node flags imminent, unacknowledged ones → the partner confirms cover at the Approval node.

Related industries

Build one of these in minutes

The sandbox gives you a live agent workspace — no account, no card. Or head back to the full catalogue and compare patterns across every industry.

See how VegaDūta compares to n8n, Dify and BotpressWhy VegaDūta's architectureEstimate your WhatsApp API costs