Guide

Grok Bot for GTM: our templates, plus the official guide in brief

By Atlas AI Resources · Published October 20, 2026 · How we make these guides

An enterprise seller spends most of the day on calls, and the admin piles up around them: meeting prep, follow-up drafts, account research, CRM notes, and customer decks. This page summarizes how one enterprise GTM author hands that work to a Chief of Staff bot and a team of specialists, including one bot per strategic account and a slides bot that can update a deck before the call ends. Then it adds our own advice: where to start, three fill-in templates, and which steps should wait for your yes.

It is based on the official guide Grok Bot for GTM (August 16, 2026) by Krista Letz, who writes about using Grok Bot for Enterprise GTM at SpaceXSI. The guide began as an internal Notion doc the author shared with the team while the product was being tested inside the company. The workflows below are summarized in our words, with a few short quotes; read the guide for the author’s full account.

1. Connect what you use every day

The author’s first tip is to connect Grok Bot to the tools you rely on daily. The list includes Salesforce, Gmail, Calendar, Sheets, Drive, Slack, Notion, Granola for meeting notes, Figma for slides, X, LinkedIn, and your data warehouse. A few other points from the getting-started section:

• It has its own computer. The bot’s computer “runs 24/7,” so work continues while your laptop is shut.
• It remembers. Bots keep your preferences and learn your writing style; the author suggests asking the bot to scan your Gmail and Slack to see how you write.
• Phone and voice. One of the author’s favorite ways to use it is the mobile app with voice input, so the author can ask for slides or pull information from tools when away from the laptop.

2. The author’s bot team

The author runs a team of bots, each with its own job, coordinated by a Chief of Staff. In brief:

• Chief of Staff. Owns meeting prep, inbox and post-call drafts, and orchestrates the rest of the team. You can ask it to spin up other agents and organize your bots in sections. The author pins it because it is the bot the author uses most.
• Daily meeting prep (a routine). Ask your bot to create a routine that preps you for the day’s meetings from Salesforce, Gmail, Slack, Granola, Gong, or online research for a new meeting. The author prefers a short, skimmable brief to read on the phone during the commute. The bot can also have customized decks made ahead of the day’s calls by calling the author’s slides bot.
• Prospecting bot. Builds pipeline generation plans, pulls intent data, and does deep research: webinars, podcasts, blog posts, and relevant LinkedIn and X posts. It tailors messaging to company priorities and personal hooks, in your voice, and drafts it in Gmail. The author runs it overnight so it is ready in the morning.
• Customer expert. One agent per strategic account (it works for top prospects too). It watches the account’s Slack channels, digests calls, emails, and Slack threads, runs a weekly media rundown on the account, flags feature requests from calls, updates the author on support tickets, and finds new features relevant to the customer.
• Data analysis agent. Connected to the relevant tools to track usage data across the author’s book.
• Product expert. The author calls it “10x engineer.” It is connected to the codebase for questions deeper than the author’s own product knowledge. When a customer asks something technical on a live call, the author asks this agent and gets a customer-facing answer back without pulling in an engineer. Without repo access, the guide suggests an internal knowledge tool like Glean as the source instead.
• 1:1 agent. Preps the author’s weekly manager sync. The author records those meetings with Granola so tasks discussed in the 1:1 can be handed off and reminders set.
• Forecasting bot. Pulls from Granola, Gong, Slack, and email and automatically updates Salesforce opportunity notes in a format the author gives it, which the author says helps streamline forecast calls and keep notes current.
• Slides. Owns customer decks in Figma using the brand system and a master deck it maintains. It copies and customizes decks per customer and can translate them. The guide says to give the bot your design guidelines up front.
• Sales Coach. Watches your external calls on Gong and gives feedback on what went well and what to improve.

The author’s favorite skill is updating slides during a customer call from the Granola transcript. The tip: stop the Granola recording 5 or 10 minutes before the call ends and run the skill live, so you can close with a slide built from what the customer just said in Discovery, or use it as a follow-up or opener for the next call.

3. The weekly media rundown, in brief

The guide includes the full prompt the author uses for each account’s weekly media rundown. Our advice: its shape is worth copying even if your sources differ:

• Ask before the first run. Which industry and product category you sell into, which topics count as signals, and which roles at the account matter most. Save the answers and reuse them every run.
• New content only. Search for webinars, podcasts, talks, exec interviews, and X posts from or about the company, and keep a state file of what was already covered so only genuinely new items are reported.
• Actually consume it. Fetch the transcript or show notes, watch the recording, or transcribe audio, then report key takeaways, mentions of your category and competitors with sentiment, quotes with timestamps, and why it matters for the account. Flag anything that looks like a buying signal.
• Say so when there’s nothing, or when something is missing. One weekly roundup; if nothing is new, one line saying so. If a transcript truly can’t be obtained, say so and summarize from the show notes rather than skipping the item.

For a similar job in our prompt library, see the weekly competitor watch.

4. The guide’s tips, in brief

• Onboard it like a new teammate. The first time you do a task together, ask the bot to record while you take over its computer and demonstrate, then have it turn the recording into a skill. The author’s example is showing the bot how the author likes to research stakeholders on X.
• Feed it your writing. Have it scan your sent mail and build a style profile before it drafts anything, and give it a few examples of outreach you’re proud of.
• Turn feedback into skills and rules. The author keeps an anti-slop skill and has the bot add to it every time the author steers it.
• Parallelize. Tell your bot to spin up multiple cloud agents for parallel tasks, and make a skill for that.
• Ask the Chief of Staff to review the team. Have it look across your bots and tasks for anything to automate, parallelize, or organize better, and to clean up unnecessary routines.
• Let bots work together. Put bots in a group chat for a task, and let them call your other specialists; the author’s Chief of Staff often calls the slides bot to make slides.

The author calls Grok Bot “the first AI tool that feels like a trusted colleague.” Because most of the author’s day is spent on calls, the guide says evenings used to go to hours of admin catch-up; now much of that work goes to the bots, and more time goes to customers. That is the author’s own account, not a measured result.

For ready-made prompts close to these tips, see write in my voice for a personal style guide and the anti-slop pass for cleaning up drafts.

5. Our advice: where to start

• Start with one read-only bot. Daily meeting prep is a good first job: it reads your calendar, CRM, mail, and notes and gives you a brief. It doesn’t write to anything, and you will know within a few mornings whether it helps.
• Drafts wait for a human yes. In the guide, the prospecting bot drafts in Gmail. Keep it that way: the bot writes, and you read and send.
• Preview CRM writes before you let them run on their own. The author’s forecasting bot automatically updates Salesforce opportunity notes. When you set up your own, have it show you the proposed note next to the current one for the first few weeks, and approve each change until its notes match what you would have written.
• Check live-call answers against the source. A product expert that answers from the codebase or an internal knowledge tool is fast, but a confident answer can still be wrong. Ask it to name the file or doc behind each answer, and say “let me confirm” when it can’t.
• Add specialists one at a time. Give each new bot one lane, and let the Chief of Staff route work between them once you have a few. Our guide on small AI fleets and handoffs covers giving each stage one owner.
• Save what works as a skill, then schedule it. The overnight prospecting run and the weekly rundown are routines, so prove the job by hand first, and use safe inputs for the test run, since a test run does real work. See skills before routines.
• Judge it by finished work: meetings you walked into prepared, first touches you actually sent, forecast notes you didn’t have to rewrite, and buying signals you caught early, not how many drafts or briefs the bots produced.

Atlas’s advice: a meeting prep brief you can read on your phone

The three templates in this section are Atlas’s own, not the guide’s. Adapt them to your tools.

• Sources: ____ (calendar, CRM, mail, Slack channels, meeting notes).
• Per meeting: who is attending and their roles, where the deal or account stands, what was promised last time, and one thing to ask.
• New meeting: a few lines of public research on the company and attendees, each with a link.
• Length: one screen per meeting.
• If a source couldn’t be read: say which one, rather than guessing.
• Stop points: no emails, invites, or CRM changes; brief only.

For ready-made versions, see the meeting prep card and our chief of staff prompts.

Atlas’s advice: a key-account bot brief

• Watched accounts: ____ (account names, plus their Slack channels, CRM records, and public sources such as ____).
• Signals to flag: ____ (leadership changes, funding or earnings news, product launches, competitor mentions, support escalations, feature requests from calls).
• Each item: one or two lines on what happened and why it matters for the account, with the source link and its date.
• If a source couldn’t be opened: mark the item “unverified” and say which source failed, rather than summarizing from memory.
• Nothing new: stay quiet, or send one line saying nothing new since ____.
• Stop points: read-only; no emails, posts, CRM changes, or messages in customer channels.

Atlas’s advice: a follow-up draft rule

• When to draft: within ____ hours after an external call that has notes or a transcript, or when a customer email has gone ____ days without a reply.
• Tone: ____ (for example, short and plain, in my voice from my sent mail; no filler or hype).
• What to include: a thank-you line, a recap of what was agreed, owners and dates for next steps, and any links or files promised on the call.
• Labeling: save it as a draft and mark it clearly as a draft (for example, “DRAFT for review” in the summary you send me), with the notes or thread it was based on.
• If something is unclear: leave a ____ blank or a question for me instead of guessing a commitment, price, or date.
• Stop points: nothing is sent, scheduled, or logged to the CRM without a human yes.

6. Keep sends and system changes behind your yes

GTM bots touch the systems your customers and managers see: outbound email, CRM records, and customer decks. Grok Bot’s guardrails are plain-language rules checked by a review agent, not code. The 101 guide puts them in Settings > General > Agent, with allow and block lists steering the reviewer. The current docs call this Settings → General → Auto-review: Ask first rules always stop matching actions for you, and Allow automatically rules let matching actions proceed only when the automated review does not find another reason to stop. Auto Review is model-based, so the docs say to use it alongside least privilege and explicit approval boundaries, not instead of them. The docs also suggest starting with read-only tasks and draft outputs, and keeping sending and publishing behind approval. An approval controls the action it is attached to; it does not undo work already done.

Our advice: write each bot’s stop points into its job (“draft only,” “show me the CRM change first,” “never share a deck outside the company without asking”). The 101 guide also notes that a login one bot uses is reachable by the others, so decide up front which bots should use which accounts.

Our advice: prospecting and account bots read a lot of text written by other people: webinars, podcasts, blog posts, X and LinkedIn posts, customer emails, and call transcripts. Any of it can contain instructions aimed at your bot, so treat it as data to read, not orders to follow. Our guide on keeping AI agents safe from prompt injection covers this in more depth.

General architecture advice, not a description of how Grok Bot works: if you also run your own scripts against your CRM or email, enforce hard limits such as read-only credentials, allowed fields for writes, and a send allowlist with exact checks in that code.

Read the author’s full account, including the complete weekly media rundown prompt, in Grok Bot for GTM.

Useful public resources

Related: Grok Bot for SDRs · Grok Bot for founder-led sales · Grok Bot for post-sales · Grok Bot for performance marketing · Grok Bot for Work · Decision forks for AI agents · Grok Bot for marketing