2 ready-to-build workflows

AI agent workflows for Professional services & sales ops

Draft proposals and keep the CRM clean without the manual busywork.

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. RFP / proposal first draft from your own material

Every RFP is answered by hand — a consultant hunts through old proposals and case studies, copies boilerplate, and rewrites the same answers for the tenth time.

Trigger: Email / Webhook (RFP received)

How the workflow runs

  1. Trigger

    RFP arrives. An Email/Webhook trigger receives the RFP document.

  2. Knowledge

    Draw from your library. A Knowledge node holds past proposals, case studies and standard answers to ground the draft.

  3. Agent

    Draft tailored responses. An Agent node answers each requirement in your voice, reusing what fits and flagging what needs input.

  4. Approval

    Lead reviews before it goes out. An Approval node keeps the human as the author of record.

  5. Output

    Return the draft. An Output node delivers the draft to the owner to finish.

Channels & connectors

  • Email
  • Webhook
  • Knowledge base
  • Approval / HITL

Outcome

A tailored first draft in minutes, grounded in your real material — the consultant edits instead of starting cold.

Why it helps

Removes the copy-hunt-rewrite cycle that makes proposals slow, while keeping a human owning the final response.

Build spec

1 agent — 1 drafting agent; the lead's sign-off is an Approval node.

System prompt (paste-ready)

You are a proposal-drafting agent for {{company}}. For each requirement in the RFP, draft a tailored response in the company's voice, reusing relevant past proposals, case studies and standard answers from Knowledge. Ground claims (credentials, references, metrics) in the library material — never invent a client, certification, or capability the firm doesn't have. Where a requirement needs a subject-matter input you can't source, flag it clearly as [NEEDS INPUT] instead of guessing.

MCP connectors

  • RFP intake (email/upload) — via Email/Webhook trigger
  • Proposal library / DMS — via a Tool (MCP) node

Built-in tools

  • document parsing (RFP document)
  • knowledge_search (past proposals, case studies, standard answers)

Guardrails

  • Claims are grounded in the library — no invented clients, certifications or capabilities
  • Gaps are flagged [NEEDS INPUT] for a human, never bluffed
  • The proposal lead reviews and owns the final draft at the Approval node before anything is sent

Output & delivery

The agent produces a tailored first draft grounded in your library with gaps flagged; the lead reviews it at the Approval node, and an Output node returns the draft to the owner to finish.

2. Meeting notes → CRM, automatically

Reps skip updating the CRM after calls; pipeline data rots and follow-ups slip.

Trigger: Webhook (call/meeting ended)

How the workflow runs

  1. Trigger

    Meeting ends. A Webhook trigger fires with the transcript/notes reference.

  2. Agent

    Extract the CRM fields. An Agent node pulls next steps, stage, and action items.

  3. Tool (MCP)

    Update the opportunity. A Tool (MCP) node writes it straight to the CRM.

  4. Output

    Send the recap. An Output node emails the client a clean summary + next steps.

Channels & connectors

  • Webhook
  • Tool (MCP → CRM)
  • Email

Outcome

The CRM stays current and clients get a recap — with zero manual data entry from the rep.

Why it helps

Fixes the 'reps don't update the CRM' problem by removing the manual step entirely.

Build spec

1 agent — 1 notes agent; the CRM update and recap send are workflow nodes.

System prompt (paste-ready)

You are a sales-notes agent. From a meeting transcript, extract CRM updates {stage, next_steps[], action_items[{owner, task, due}], sentiment, key_points}. Capture only what was actually discussed — never fabricate a commitment, number, or next step that wasn't said. Then draft a clean, client-facing recap email summarizing decisions and next steps. Keep the recap professional and concise.

MCP connectors

  • CRM (Salesforce/HubSpot/etc.) — via a Tool (MCP) node
  • Meeting/transcript source — inbound via Webhook trigger
  • Email — recap channel

Built-in tools

  • http_request (fetch transcript by reference)

Guardrails

  • Only what was actually discussed is recorded — no fabricated commitments or figures
  • The CRM write is keyed to the opportunity/meeting id — idempotent, no duplicate updates
  • The client recap restates only agreed next steps — nothing promised that wasn't discussed

Output & delivery

The agent extracts the CRM fields and a Tool node writes them to the opportunity; an Output node emails the client a clean recap of decisions and next steps.

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.

Try this in the sandbox — no signupBrowse all industries
See how VegaDūta compares to n8n, Dify and BotpressWhy VegaDūta's architectureEstimate your WhatsApp API costs