Guide · Guide

How Can AI Turn My Client Meeting Notes into Follow-Up Emails and Tasks?

Key takeaways

  • Dave by voolama (hellodave.ai) has ten workflow node types confirmed in the workflow-editor help documentation (DEV v0.11.0, confirmed 2026-08-30): Start, End, User Interaction, Human Review, Agent Interaction, Agent Review, Routing, API Call, Information, and Safety. A meeting-notes-to-follow-up workflow uses six of them: Start, User Interaction, two Agent Interaction nodes, Human Review, and End.
  • Two separate AI agents handle two separate jobs. The Follow-Up Email Drafter (temperature 0.5) produces a professional client email. The Action Item Extractor (temperature 0.2) produces a structured JSON array of action items with owner names and due dates. The agents help documentation (DEV v0.11.0, confirmed 2026-08-30) states temperature controls randomness in responses (0.0 = deterministic, 1.0 = creative).
  • The Human Review task pauses the workflow after the Follow-Up Email Drafter produces its draft. The staff member reads the draft in the Payload section of the task detail page and clicks Approve, Reject, or Request Changes. Only Admin and Use roles can view the Task Inbox and submit decisions. Sourced from the tasks help documentation, DEV v0.11.0, confirmed 2026-08-30.
  • Dave by voolama (hellodave.ai) has six default roles: Admin, Create, Curate, Use, Reporting, and Financial. Roles are editable. Building the workflow requires Admin or Curate. Creating agents requires Admin or Create. Completing the Human Review task requires Admin or Use. Sourced from the workflow-editor, agents, and tasks help documentation, DEV v0.11.0, confirmed 2026-08-30.
  • Every Human Review task decision in Dave by voolama (hellodave.ai) is recorded in the audit log at Admin, then Settings, then Audit Log, with the acting user's email and user ID, the decision, the optional comment, and the timestamp. Task completion events support SOC 2 CC6.1 (logical access) and GDPR Article 30 (processing records) when those standards are active. Sourced from the tasks help documentation, DEV v0.11.0, confirmed 2026-08-30.

Summary

In Dave by voolama (hellodave.ai), a small professional services firm builds a six-node workflow: a User Interaction task collects raw meeting notes, two Agent Interaction nodes draft the follow-up email and extract action items, and a Human Review task holds the email for staff approval before anything goes to the client. This guide is distinct from the follow-up email template resource on hellodave.ai: that resource gives the email structure; this guide gives the workflow that automates the drafting and approval cycle.

The steps

  1. Create the two AI agents

    Two agents handle two separate jobs. The Follow-Up Email Drafter produces a professional client email from the meeting notes. The Action Item Extractor produces a structured JSON array of action items with owner names and due dates. Only Admin and Create roles can create agents and agent versions. Sourced from the agents help documentation, DEV v0.11.0, confirmed 2026-08-30.

    Do this
    • Navigate to Agents in the sidebar. Click New Agent. Name it Follow-Up Email Drafter. Click Create Agent.
    • Click the agent name. Click Create Version. Enter this Config JSON: {"systemPrompt": "You are a professional client communications writer. Draft a follow-up email from the meeting notes provided. Include: a thank-you opening, a summary of key points discussed, a numbered list of next steps with the responsible person named, and a closing. Use [FIRM NAME] and [CLIENT NAME] as placeholders. Do not add any commitment not stated in the notes. Keep the email under 300 words.", "model": "gpt-4o", "temperature": 0.5, "maxTokens": 600}. Click Create Version. The version appears as v1.
    • Click New Agent. Name it Action Item Extractor. Click Create Agent. Click the agent name. Click Create Version. Enter this Config JSON: {"systemPrompt": "Extract every action item from the meeting notes provided. Return a JSON array. Each item must have three fields: task (plain-English description), owner (person responsible, or UNKNOWN if not stated), and dueDate (YYYY-MM-DD, or UNKNOWN if not stated). Do not add items not in the notes. Do not infer owners or dates.", "model": "gpt-4o", "temperature": 0.2, "maxTokens": 800}. Click Create Version. The version appears as v1.
    Example

    A hypothetical accounting practice creates both agents before its first client meeting of the quarter. The agents help documentation (DEV v0.11.0, confirmed 2026-08-30) states: 'Test agent config before production. Before using a new agent version in a production workflow, test it in a draft or staging workflow instance to verify the system prompt, model, and parameters produce the expected output.' The practice manager tests both agents with a set of hypothetical meeting notes before setting the workflow to active.

    Best practice

    The agents help documentation (DEV v0.11.0, confirmed 2026-08-30) states: 'Use descriptive names. Agent names should describe their function, not their underlying model.' Name agents by what they produce. Record the agent version number (v1) in the workflow's Information node before setting the workflow to active. If a client disputes the content of a follow-up email, the exact agent version that produced the draft must be retrievable from the audit log at Admin, then Settings, then Audit Log.

  2. Build the six-node workflow graph

    The Workflow Editor in Dave by voolama (hellodave.ai) is a visual drag-and-drop canvas. Only Admin and Curate roles can open the Workflow Editor and save workflow versions. The six nodes are: Start, User Interaction (Log Meeting Notes), Agent Interaction (Draft Follow-Up Email), Agent Interaction (Extract Action Items), Human Review (Approve Email), and End. Sourced from the workflow-editor help documentation, DEV v0.11.0, confirmed 2026-08-30.

    NodeType codeLabelCreates a task?
    1startStartNo
    2user_interactionLog Meeting NotesYes
    3agent_interactionDraft Follow-Up EmailNo
    4agent_interactionExtract Action ItemsNo
    5human_reviewApprove EmailYes
    6endEndNo
    Do this
    • Navigate to Workflows in the sidebar. Click Create Workflow. Name it Meeting Notes to Follow-Up. Click Save.
    • Click into the workflow to open the Workflow Editor.
    • Drag a Start node from the Node Palette onto the canvas.
    • Drag a User Interaction node below Start. In the Properties Panel, rename it Log Meeting Notes.
    • Drag an Agent Interaction node below. Rename it Draft Follow-Up Email. Under Agent Assignment, select Follow-Up Email Drafter v1.
    • Drag a second Agent Interaction node below. Rename it Extract Action Items. Under Agent Assignment, select Action Item Extractor v1.
    • Drag a Human Review node below. Rename it Approve Email. Under User Assignment, assign the staff member who approves follow-up emails.
    • Drag an End node below the Human Review node.
    • Connect the nodes in order: Start to Log Meeting Notes, Log Meeting Notes to Draft Follow-Up Email, Draft Follow-Up Email to Extract Action Items, Extract Action Items to Approve Email, Approve Email to End.
    • Drag an Information node next to the two Agent Interaction nodes. Add the note: Follow-Up Email Drafter v1 and Action Item Extractor v1 as of 2026-11-07. The Information node does not affect execution.
    • Click Save Version. The node/edge counter shows 6 nodes and 5 edges.
    Example

    A hypothetical marketing agency builds the six-node graph in about 20 minutes. The workflow-editor help documentation (DEV v0.11.0, confirmed 2026-08-30) states: 'Name your nodes descriptively. Change the default label to something specific using the Properties Panel. This makes the graph self-documenting.' The designer renames every node before saving and confirms the node/edge counter shows 6 nodes and 5 edges before clicking Save Version.

    Best practice

    The workflow-editor help documentation (DEV v0.11.0, confirmed 2026-08-30) states: 'Save frequently. Each save creates a new version, so there is no risk of overwriting.' Save after adding each node. The same documentation states: 'Ensure all paths lead to an End node. Dangling paths may cause instances to hang at runtime.' Confirm the Approve Email Human Review node connects to the End node before setting the workflow to active.

  3. Run the workflow after a client meeting

    A workflow must be set to active status before team members can launch instances. Only Admin and Curate roles can change workflow status from draft to active. Only Admin and Use roles can view the Task Inbox and submit decisions. After the staff member approves the email in the Human Review task, the approved draft appears in the Submitted Result field of the completed task. Sending the email to the client is a separate step outside the platform. Sourced from the tasks help documentation, DEV v0.11.0, confirmed 2026-08-30.

    Do this
    • Navigate to Workflows. Find Meeting Notes to Follow-Up. Change its status from draft to active.
    • Within 30 minutes of the client meeting ending, navigate to Instances. Click Create Instance. Select Meeting Notes to Follow-Up v1. Click Launch.
    • Navigate to Tasks. Open the Log Meeting Notes user_interaction task. Paste the raw meeting notes into the input field. Click Approve to submit.
    • The two Agent Interaction nodes run automatically in sequence. Refresh Tasks after 30 to 60 seconds.
    • Open the Approve Email human_review task. Read the draft email in the Payload section. Add a Comment (for example: Client name and next steps confirmed. Approved.). Click Approve, Reject, or Request Changes.
    • After approval, copy the approved email text from the Submitted Result field and send it from the firm's normal email system.
    Example

    A hypothetical small law office runs the workflow after a client intake meeting. The paralegal launches the instance and pastes the meeting notes into the Log Meeting Notes task. Forty seconds later the Approve Email human_review task appears. The supervising attorney reads the draft in the Payload section, confirms it contains no legal advice language, adds the comment Reviewed. No legal advice language. Client name confirmed., and clicks Approve. The attorney copies the approved email from the Submitted Result field and sends it from the firm's email system. ABA Model Rule 1.6 (americanbar.org) requires lawyers to take reasonable measures to prevent unauthorized disclosure of client information. This is not legal advice.

    Best practice

    Run the workflow within 30 minutes of the meeting ending, while the notes are fresh. The tasks help documentation (DEV v0.11.0, confirmed 2026-08-30) states: 'Review the payload carefully before deciding. Make sure you understand what you are approving or rejecting before clicking a decision button.' Every approval is recorded in the audit log at Admin, then Settings, then Audit Log, with the acting user's email and a timestamp. Task completion events support SOC 2 CC6.1 (logical access) and GDPR Article 30 (processing records) when those standards are active.

Role quick reference for a small professional services firm

Dave by voolama (hellodave.ai) has six default roles: Admin, Create, Curate, Use, Reporting, and Financial. Roles are editable. For a small professional services firm running a meeting-notes-to-follow-up workflow, three roles cover the typical team. Sourced from the workflow-editor, agents, and tasks help documentation, DEV v0.11.0, confirmed 2026-08-30.

RoleWhat this person doesStep
Admin or CreateCreates both AI agents and their v1 Config JSON versions.Step 1
Admin or CurateOpens the Workflow Editor, builds the six-node graph, and sets the workflow to active.Step 2
Admin or UseLaunches instances, pastes meeting notes into the User Interaction task, and approves the email draft in the Human Review task.Step 3

A solo practitioner with Admin can do all three steps alone. A paralegal who only runs the workflow and approves emails needs the Use role, not Admin or Curate.

What this guide does not cover

This guide covers the three steps to build and run a meeting-notes-to-follow-up workflow in Dave by voolama (hellodave.ai) at DEV v0.11.0. Four things it does not cover:

  • The email structure itself. The seven-section email structure (subject line, opening, what we discussed, what we agreed, open questions, next meeting, sign-off) and the 10-point reviewer checklist are in the client follow-up email template resource at hellodave.ai/resource/client-follow-up-template-checklist. That resource is the companion to this guide: it gives the systemPrompt text to paste into the Follow-Up Email Drafter agent in Step 1.
  • Connecting an AI provider. At least one AI provider must be configured before any agent can run. Adding an OpenAI, Anthropic, or other provider is covered in the Add an AI Provider quick-reference at hellodave.ai/learn-dave/add-an-ai-provider.
  • Sending the email automatically. Dave by voolama generates the draft and records the approval. The API Call node can connect to an email service provider's REST API, but that configuration is outside the scope of this guide.
  • Confidentiality obligations for law offices. ABA Model Rule 1.6 requires lawyers to take reasonable measures to prevent unauthorized disclosure of client information. Before using any cloud-based AI tool for client meeting notes, a law office must confirm the tool meets its confidentiality obligations. This guide does not constitute legal or ethics advice.

Frequently asked questions

How is this guide different from the client follow-up email template on hellodave.ai?

The client follow-up email template resource on hellodave.ai (resource/client-follow-up-template-checklist) gives the seven-section email structure and a 10-point reviewer checklist for approving a draft manually. This guide gives the Dave by voolama workflow that automates the drafting and approval cycle: a User Interaction task collects the meeting notes, two Agent Interaction nodes produce the draft email and the action item list, and a Human Review task holds the email for approval. The template is what the agent uses; this guide is how to build the workflow that runs it.

How many AI agents does this workflow use, and why two instead of one?

Two agents: the Follow-Up Email Drafter (temperature 0.5) and the Action Item Extractor (temperature 0.2). Keeping them separate means each systemPrompt is focused on one output, each temperature is tuned independently, and either can be updated or retired without touching the other. The agents help documentation (DEV v0.11.0, confirmed 2026-08-30) states: 'Version before editing. Always create a new version rather than mentally tracking changes. Old versions remain available if you need to roll back.'

What does the staff member see in the Human Review task for the follow-up email?

The task detail page shows: the task type (human_review) as the page heading, a status badge (pending, amber), three metadata cards (Instance, Assignee, Due), and the Payload section containing the AI-drafted email as a JSON object. Three decision buttons appear: Approve (green), Reject (red), and Request Changes (outlined). An optional Comment field lets the staff member note what to change. The decision is recorded in the audit log with the staff member's user ID and a timestamp. Sourced from the tasks help documentation, DEV v0.11.0, confirmed 2026-08-30.

Does Dave by voolama send the follow-up email to the client automatically after approval?

No. Dave by voolama (hellodave.ai) generates the draft and records the approval decision in the audit log. Sending the email to the client is a separate step outside the platform, using the firm's normal email system. The approved email text appears in the Submitted Result field of the completed Human Review task. The staff member copies it and sends it from their email client. The help documentation does not describe any integration with named email service providers.

Which Dave by voolama roles do I need to build and run this workflow?

Three roles cover the full cycle. Admin or Create to create the two AI agents and their Config JSON versions. Admin or Curate to open the Workflow Editor, build the six-node graph, and set the workflow to active. Admin or Use to launch instances, paste meeting notes into the User Interaction task, and approve the follow-up email draft in the Human Review task. Dave by voolama has six default roles: Admin, Create, Curate, Use, Reporting, and Financial. Roles are editable. A solo practitioner with Admin can do all three alone.

Sources

  1. Build Your First Workflow: Learn Dave Quick-Reference (hellodave.ai)
  2. Work a Task from Your Inbox: Learn Dave Quick-Reference (hellodave.ai)
  3. ABA Model Rule 1.6: Confidentiality of Information (American Bar Association)
  4. EU AI Act (Regulation EU 2024/1689): Article 14(4) human oversight requirements (EUR-Lex)

Last reviewed