Guide

Grok Bot for post-sales: a chief of staff bot, account specialists, and routines that keep promises

By Atlas AI Resources · Published October 11, 2026 · How we make these guides

Post-sales is everything after the deal closes: kickoff through renewal. The hard part is not the customer relationship; it is the context across transcripts, Slack, email, calendar, and promises made in passing. This page shows how one chief of staff bot can run a specialist bench and quiet routines so more of your day goes to the customer, while you keep the final yes on anything that goes out with your name.

It is based on the official guide Grok Bot for Post-Sales (September 24, 2026) by Blake Schuller. The workflows below are the author’s own setup, summarized in our words. Read the guide for the full detail.

1. Why post-sales needs a bot team

In the author’s role, post-sales means owning onboarding, making sure teams see value, and keeping deployment as painless as possible. That work is energizing. What wears you down is reassembling context before every call: who said what, what you promised, which thread went quiet, and which account needs attention today.

Grok Bot gives you always-on agents with their own computers. They use the same tools you do, keep working while you are in back-to-back meetings, and surface the context you used to chase. They do not replace the relationship side of the job. They give you more time for it.

2. One chief of staff, not twenty chats

The author’s front door is Gus, a chief of staff bot. Gus manages the rest of the bench so the human can keep the parts of the job that matter: relationships, strategy on each account, face time, deciding what matters today, and the final yes on outbound.

Gus exists because the author first tried running twenty agents directly and found it overwhelming: twenty chats, twenty personalities, all wanting a reply. Now there is one front door. Specialists own their lanes; Gus surfaces what needs the human. Days go by without opening the specialist chats at all. They remember directions, report to Gus, and Gus decides what actually needs attention. That coordinator pattern is good practice when the specialists stay narrow and the human still owns relationships and send approval.

3. The specialist bench

Behind Gus sits a bench the author built up over a few weeks:

• Follow-ups holds every draft waiting on the human and remembers what happened to it.
• Account specialists, one per account, whose job is to know that customer thoroughly.
• Ink writes emails and Slack messages in the author’s voice.
• Questions is wired to the team's docs for anything a customer might ask about the product.
• Career gives weekly feedback and quietly keeps a list of wins.
• Scout reads internal channels so the human does not have to.

The guide also notes a dozen more specialists the author has not opened in weeks, which is the point: the bench works through Gus, and quiet specialists are a feature.

4. Computer and plugins for account work

Plugins cover systems with connectors: calendar, Slack, mail, docs, and call transcripts. The bot’s own computer covers everything else. The author logs the bots into the tools they use; the bots work the way a person would and keep going when the laptop is closed.

Beyond connectors, the author’s agents use the computer for:

• Checking LinkedIn before a first call
• Internal pages that no connector reaches (point the bot at the page instead of pasting it into chat)
• Deck review: open a PowerPoint and say what works visually and what does not
• Screenshots and short screen recordings so a customer how-to gets a picture instead of three paragraphs

5. Routines that stay quiet unless you need them

A routine has a trigger, scheduled or event-based, and runs whether or not you show up. Most runs of the author’s watches produce nothing, and that is by design. A routine that looks and decides you do not need to know is doing its job.

• Daily brief, 8:30 on weekdays. Calendar, Slack, and call transcripts folded into one short list of what to focus on.
• Before-every-call prep. Checks every fifteen minutes, but only writes a prep card when a call is fifteen to thirty-five minutes out. Who is on the invite and who accepted, the one thing to leave with, talking points with links, and a short do-not list.
• Unfinished promises, at 9, 1, and 4. Things said that still need doing, linked back to where they were said. Done items drop off on their own.
• Unanswered asks, every fifteen minutes. Looks at messages the human sent that got no reply, and only surfaces ones where the next move is still theirs. It never replies for the human. Most runs are silent.
• Visibility watch, weekday mornings. Finds one customer win worth telling the company about and stages a post to paste.
• Exhaust watch, weekday mornings. Scans live accounts for a usage cliff or a brewing escalation.
• Friday dashboard. Regenerates account heatmaps and trend charts into a Drive tracker so Friday opens as one view of where to spend time.
• Self-improvement scan, Wednesdays. Gus audits the routines: which fired and got ignored, what has been asked by hand three times, what should exist and does not. Gus proposes the fix; the human says yes or no.

Weekends stay quiet. Gus still watches for anything urgent. Success, in the author’s words, is measured by how much time goes to customers while the system runs in the background. “I didn’t automate my job. I stopped spending the first hour chasing context.”

6. When an account has you underwater

Some days an account gets away from you: too many threads, too many people, and you have lost track of what you owe whom. The author’s move is to open Gus and say exactly that: “I’m overwhelmed with x account.”

Gus pulls the picture together: what actually happened across transcripts, calendar, and Slack; what was promised and to whom; and what to do next. That is the chief of staff job in one sentence: specialists own the lanes, Gus assembles the state, and you keep the relationship and the decisions.

A typical day in the guide looks like this: the morning brief is waiting; prep cards land just before calls; while you are in customer meetings, transcripts land, recaps get drafted, research for questions they asked is done, and materials you promised are staged; you tweak drafts and press send; the delta between draft and sent goes back to Ink as feedback so the next message sounds more like you.

7. Keep outbound behind your approval

Anything that goes out with your name stays behind draft-then-human-send. The unanswered-asks routine never replies for you. Recaps and email responses sit in drafts until you tweak and send. Visibility posts are staged to paste, not posted alone.

Grok Bot’s guardrails are plain-language rules you write in Settings > General > Agent, checked by a review agent that you steer with allow and block lists. Spell out each bot’s stop points (“draft only,” “ask before sending”), connect only the tools each job needs, and keep sending, publishing, and customer contact behind approval. An approval controls the action it is attached to; it does not undo work already done, so keep consequential steps as drafts until you trust them.

General architecture advice, not a description of how Grok Bot works: if you also run your own scripts against mail, CRM, or calendar APIs, enforce hard limits such as send rates and write scopes with exact checks in that code. A confident draft is still a draft until a person sends it.

8. Where to start

• Record a voice memo of your day. The author talked for over ten minutes about how the day is spent, what the author loves doing, what isn't worth the time, and the parts the author refuses to automate away. Send that to your bot and ask it to build a system that makes the job easier.
• Or send the official guide to your first agent and ask it to turn the playbook into instructions it should follow.
• Start with one front door. A chief of staff that compiles and surfaces beats twenty parallel chats that each want a reply.
• Add specialists by lane: one account bot, a drafts holder, a docs bot. Keep each job narrow.
• Connect what you use most: calendar, Slack, mail, docs, and transcripts. Point the computer at pages that have no connector.
• Design routines to stay quiet unless a threshold is crossed. Silent no-ops are success.
• Keep outbound as drafts for the first few weeks. The gap between draft and sent is how Ink (or your writing bot) learns your voice.
• Test routines before enabling them. A test run does real work, so use safe inputs and keep write actions behind approval.
• Judge the system by finished work: time with customers, promises closed, prep cards that were useful—not how many times a routine fired.

Roll out gradually: one low-risk routine, watch what it drafts, then expand. Confidence is never permission; the human still owns the relationship and the final yes.

Useful public resources

Related: Grok Bot for SDRs · Grok Bot for support · Grok Bot for founder-led sales · Scheduling AI agent routines · Skills before routines · Complete Grok Bot setup guide · Small AI fleets: handoffs · Grok Bot chief of staff prompts · Grok Bot usage limits and cost · Keeping AI agents safe from prompt injection · Grok Bot for GTM