Grok Bot for Brevo: three practical guides

Pick one sent Brevo campaign whose results you already know. Ask Grok Bot to report its delivered messages and clicks, then compare the answer with Brevo. That small check tells you more than a polished summary of your entire account.

These three playbooks cover campaign reporting, email drafting and contact-data review. Each includes a prompt, an expected result and a check before you rely on it.

The connection recipe combines xAI’s documented MCP support with Brevo’s hosted MCP server. We have checked the documentation, not run this connection inside a Grok Bot account. Complete the read-only check below before using account data.

Connect only what the first task needs

Start with reporting. Brevo lists a campaign analytics server at https://mcp.brevo.com/v1/brevo_campaign_analytics/mcp. Other tasks can use its campaign-management or contacts servers later.

1. In Brevo, open Settings > SMTP & API > API Keys & MCP. Generate a key with Create MCP server API key enabled, following Brevo’s setup instructions.
2. In Grok Bot, ask to add a custom MCP connection using the reporting endpoint. If your client offers secure credential input, supply the MCP token there, with the Authorization header using the Bearer scheme. Keep the token out of ordinary chat and shared files.
3. Inspect the tools the connection exposes. Select the read operation for the campaign report; do not infer permissions from the server’s name.
4. Run the first prompt below with a real campaign ID you choose. Compare its result with the same campaign in Brevo.

If custom MCP setup or secure header entry is unavailable in your Grok Bot client, use a Brevo report export instead. Record its export time. Do not paste a token into a prompt to work around missing configuration.

Task	Brevo server path	First result to verify
Report on a campaign	/v1/brevo_campaign_analytics/mcp	The chosen campaign’s metrics
Prepare a campaign draft	/v1/brevo_email_campaign_management/mcp	Draft content and status
Review contact data	/v1/brevo_contacts/mcp	A bounded list or sample

These paths use https://mcp.brevo.com. A smaller server reduces the tools presented to the bot. It does not establish a read-only credential or a security boundary. Use the available tool controls and review proposed operations.

Our Brevo MCP guide covers the token and server choices in more detail. The Brevo API guide explains the separate REST API route.

Guide 1: get a campaign report you can check

Choose a campaign ID and an observation period. For a comparison, also specify which campaigns belong in the report and how long each has had to collect results.

Copy this prompt, then supply the campaign ID:

Read the Brevo campaign I identify. Make no changes. Return its ID, name, send time, retrieval time, delivered messages, unique clicks, unsubscribes and bounces where available. Label missing metrics as unavailable. Keep total and unique counts separate. For every calculated rate, show the numerator and denominator. Say whether the result is a lifetime campaign snapshot or a date-filtered report. Finish with one observation supported by the numbers and one question the data cannot answer.

Expected result: a small table with the campaign identity, source timestamps and metrics. Brevo’s campaign-report reference describes the underlying report; the connected MCP tool may expose only part of it.

Check: compare delivered messages and clicks with Brevo at approximately the same time. If the tool lacks a date filter, keep the snapshot label. Do not let the bot turn a lifetime total into “last week’s results.”

Clicks alone do not establish sales attribution. Ask for revenue only when a connected source provides it, and identify that source and its attribution method. The existing email metrics guide provides the measurement context.

Guide 2: turn an approved brief into an email draft

Give the bot an offer, its expiry, a destination URL and the intended audience. Include the product facts it may use. Drafting can happen in chat before any Brevo connection is needed.

Copy this prompt with your brief:

Use only the product facts, offer and audience I provide. Draft two subject lines, preview text, an email body and one call to action. Flag missing prices, dates, links or claims instead of inventing them. Include a fallback for a missing first name. Show the audience assumptions and the exact destination URL. Return the draft here. Do not create, schedule or send a campaign, add recipients, or change any contact.

Expected result: an editable email and a short list of unresolved details. The bot should make the missing expiry visible rather than decide one for you.

Check: open the destination URL, verify the offer and read the fallback greeting. Use the campaign planning guide if the audience or objective still needs work.

Once you approve the content, you can separately authorize creation of one unscheduled Brevo draft, if the connected tools support it. Specify the sender and recipients explicitly. Confirm its draft status in Brevo before approving any test email or send. Creating a draft, sending a test and sending to an audience are separate actions.

Guide 3: find contact-data problems without changing records

Choose one list or a small export containing data your organization permits Grok Bot to process. Minimal report output does not limit what a connected tool reads. Define which fields your process requires and where subscription evidence is stored. A blank first name is a data issue; a guessed first name is not a repair.

Copy this prompt with the list or export scope:

Inspect only the list or file I specify. Report missing first names, malformed email values and missing source or subscription evidence according to my field map. Separate confirmed findings from unknowns. Return counts and only the minimum record identifiers needed for review; do not reproduce the full contact list. State how many records you inspected and whether pagination or sampling limits the result. Do not create, merge, delete, unblock or resubscribe contacts, change attributes, add them to lists, or send messages.

Expected result: an issue table with counts, the checked scope and suggested next steps. Missing subscription evidence stays unknown until the authoritative source is checked.

Check: inspect a few flagged records yourself. A sample cannot establish the condition of the entire list. If business email is your signup policy, evaluate it against an explicit rule; a company domain does not prove consent.

Keep new subscriptions in the existing double opt-in process. An AI classification must not override confirmation or a later unsubscribe. For form data, the Webflow-to-Brevo guide shows how to preserve that boundary.

Repeat a proven report

Grok Bot supports routines. After the report passes its manual check, define the cadence, timezone, campaign scope and private destination before enabling a recurring run.

Have it state the last successful retrieval time. An authentication failure should produce a failed-run notice, not a recycled report presented as fresh. Keep campaign sends and contact changes outside the reporting routine.

Where Tajo fits

This Brevo MCP connection does not require Tajo. Tajo’s own agents and integrations use a separate connection and permission system, described in MCP & Agents.

If the report exposes missing store or form data, repair that source connection before asking the bot for more analysis. Our Shopify-to-Brevo guide covers the store setup. Start with one campaign whose numbers you can verify.

More engagement guides

Choose the customer event you need to handle. These guides connect source data, an eligible audience, a Brevo action and a result you can check:

• Onboarding: welcome confirmed subscribers and help eligible trial users take the first action.
• Lead nurturing: use verified interests and follow up on actual sales-call commitments.
• Ecommerce engagement: recover carts, explain purchased products and recommend relevant additions.
• Retention and renewals: distinguish inactivity, service problems and approaching contract dates.
• Product feedback: tell the right customer when a requested feature is available to them.
• Campaign experiments: compare fixed variants with a defined audience and outcome.