From call notes to a verified CRM update
See how an AI agent turns call notes into a checked CRM proposal, so a salesperson reviews every update before it is saved.
Update your CRM without spending hours on it.
After a sales call, important context often remains in the salesperson’s head. The CRM then contains only scattered notes and an open task, while colleagues need the agreements, objections and next steps for effective follow-up. Properly recorded CRM data is therefore essential to a good sales and business development process.
After a call, the AI agent retrieves the relevant customer and sales data from the CRM. It combines that context with the call notes and creates one proposal with agreements, objections and next steps. The salesperson checks the proposal and decides what is saved.
Less admin after a call
The salesperson checks a completed proposal instead of updating the CRM from scratch.
Better information for follow-up
Agreements, objections and next steps have a consistent place in the CRM.
A check before anything is saved
The salesperson checks the customer match and corrects the proposal where needed.
How information from the call reaches the CRM.
Every step has one purpose. The agent creates a proposal; a salesperson decides what is actually saved.
- 01SYSTEM
Call notes become available
The available call notes start one new processing run.
Why: This keeps each run linked to one call, with a clear source.
- 02AGENT
The agent retrieves the corresponding CRM data
The agent reads the relevant customer, contact and open sales opportunity.
Why: This context lets the agent link the information from the call to the right CRM data.
- 03AGENT
The agent proposes a CRM update
The agent puts agreements, objections, the person responsible and the next step into a proposal, including an indication of confidence.
Why: This turns information from the call into a proposal that fits the way you work in your CRM.
- 04HUMAN
A salesperson checks the proposal
The salesperson checks the selected customer, sales opportunity and proposed fields and corrects them where needed.
Why: The salesperson can take commercial context into account that is not fully captured in the notes or CRM.
- 05SYSTEM
The CRM update is saved
After approval, only the checked proposal is saved in the CRM.
Why: The approval step makes it clear which information is changed and by whom.
The steps stay the same. The tools can change.
Use the sources, AI solutions and integrations that already suit your team. Choose where the call notes come from, which AI processes them and how an approved update is saved in the CRM.
- Custom agent
CRM integration
Direct API integration
What you need for a first version.
Choose one clearly defined route that your team can follow, review and improve.
For this version, all you need is: Access to call notes, A read and write integration with the CRM, Agreement on the CRM fields the agent may use.
- 01
Define which CRM fields the agent may read and which fields it may propose changes to.
- 02
Give the agent access only to the customer and sales data needed for this workflow.
- 03
Have a salesperson approve every proposed update before it is saved.
Keep the first version's scope clear. One call leads to one proposed CRM update. New contacts and changes to sales stages are outside the scope of this first version.
The agent drafts a proposal; your team decides what to save. Test with anonymised call notes from recognisable real-world situations. Check that the agent selects the right customer, sales opportunity, agreements and next steps. When you test, include incomplete notes, ambiguous cases and an incorrect customer match.
Use a ready-made prompt.
Use this prompt in Claude, Claude Code, ChatGPT, Codex, Microsoft Copilot or Gemini. The agent starts with the questions below before building a clearly scoped first version.
Would you like en:twine to help you build it? Contact us.
- 01Which CRM do you use and which fields do you want to update after a sales call?
- 02Where do the call notes come from and how does the agent access them?
- 03Which customer, contact and sales opportunity must the agent be able to link the notes to?
- 04Who checks and approves the proposal before it is saved?
- 01Define which CRM data the agent may read and which fields it may propose changes to.
- 02Have the agent extract agreements, objections and next steps from the call notes.
- 03Link the proposal to the right customer, contact and sales opportunity.
- 04Test with example calls and save only approved proposals.
The full prompt includes the workflow URL, sources, AI solutions, integrations, questions, build steps and safety boundaries.
Does this work for one person? Make it reliable for your team.
en:twine helps with shared definitions, permissions, exceptions, monitoring and day-to-day adoption.