Key takeaways
- Give each Grok bot one core function. Inbox, meeting notes, bookkeeping, venue search, and negotiation are separate jobs, not one generalist.
- Put a chief of staff bot over the portfolio so specialists can jump into a group conversation and spin a project up together.
- Reverse-engineer the human org chart. For a pop-up, that means an event-planner bot on top and tasked coordinator bots underneath.
- Do not send a search bot after venues until timing, capacity, evening hours, food service, kitchen vs. catering, and vendor rules are written down.
- Have the event planner bot draft the production budget before anyone requests RFPs or negotiates. Budget is the baseline, not a later cleanup step.
- Do venue outreach and bid negotiation by email, inside parameters you set. Jenny uses the same idea to negotiate resale bids while she is doing other work.
- Grok bots can carry most of the planning load. Humans still need to handle registration and security. In California, permits are a first-class constraint, especially if alcohol is involved.
Who this Grok bot model is for (and who should skip it)
This is for a founder, operator, marketer, or ops lead who can already work in Grok bots and wants a staffing model, not a tool tour. Jenny’s stack assumes you will write role prompts, feed detailed criteria, and let bots draft budgets, search lists, and email RFPs. Her own setup also uses Wispr Flow on every meeting so a notes bot can intake the transcript and delegate. You do not need a large human team to start. She is running this while raising a baby and operating several companies at once.
Skip it if you expected a finished San Francisco venue shortlist, a priced budget, or a booked pop-up. None of that is completed here. Skip it if your plan depends on bots placing phone calls to venues; the recommendation is email. Skip it if you will not specify event details before searching. Vague criteria are treated as the reason venue hunts stall. San Francisco is described as one of the pricier cities for this work, and no dollar figures are given, so this is not a cheap-market playbook.
How the Grok bot chief of staff model works
Jenny leads a community of women building with AI, with a background in venture, technology, community, and a media business across platforms. She also runs seven businesses — real estate, a VC fund, media, an agency, and others — and a new agency, Gemini Media, that builds executive brands. The operating claim is direct: 22 Grok bot employees make it possible to keep building those companies with a newborn at home.
I have been able to double down on building all of my companies because I have Grok bot.
Shaan did not like the chief of staff model. Jenny kept one anyway. Her husband named it Master Chief. Master Chief oversees the whole of “Jenny Co.”: tenants and guests in rental properties, the agencies, inbound brand partnerships, and the rest of the portfolio. Specialists do not report to you one by one for every task. They report through that overlay.
One overlay bot, many single-task bots
The rule is individual tasks, not a blob of “AI help.” Boxy monitors the inbox. Scribe takes in every meeting captured with Wispr Flow, then delegates work to sub-agents. A CFO setup handles bookkeeping, finances, and receipts when she is out. Scout is the name she uses for a search bot that goes out against a tight brief. An event planner bot sits above coordinators for a live event. A separate bot resells clothes across Poshmark, Depop, and Mercari and negotiates bids inside a framework she set.
What “good” looks like: you can name the bot, name the job, and point to an input it always receives. What failure looks like: one bot asked to “run the business,” or a search bot launched before the brief is specific enough to return availability.
Spin the team up in a group conversation
Master Chief’s job is not only oversight. It spins specialists up so they join a project together, chat, and get the work moving. Treat that as the difference between a folder of chatbots and a staff. When a new workstream appears — a pop-up, a partnership, a venue hunt — you do not start by hiring a new human generalist. You assign the lead bot and bring the tasked bots into the same conversation.
Specialized Grok bots by business function
Use this as a staffing map, not a shopping list. You do not need 22 bots on day one. You do need one job per bot and an overlay that can call them in.
| Use case | Bot / tool | Input needed | Output | Owner |
|---|---|---|---|---|
| Inbox monitoring | Boxy | Incoming mail | Inbox watch on a single function | Ops |
| Meeting capture and delegation | Scribe, with Wispr Flow | Every meeting | Notes in; tasks delegated to sub-agents | Ops |
| Bookkeeping and receipts | CFO setup | Finances and receipts, including while out | Bookkeeping and tracking | Finance |
| Portfolio oversight | Master Chief | Rentals, agencies, inbound brand partnerships, other businesses | Coordination; group spin-up on a project | Founder |
| Venue search | Scout | Written venue and event criteria for San Francisco County | Venues that meet the brief, aimed at availability | Events |
| Production budget and RFPs | Event planner bot | Guest count, city, line items, event type | Budget template, then RFPs to venues that qualify | Events |
| Resale negotiation | Unnamed Grok bot | Listings and bids on Poshmark, Depop, Mercari | Back-and-forth negotiation inside a set framework | Founder |
Reverse-engineer an event producer into a bot team
The worked example is a San Francisco pop-up. The team had ideas, a list of places they could reach out to, and almost none of the logistics done. Jenny’s first move was not “have a bot find warehouses.” It was to treat event production as an org chart, then staff it.
Start with one pop-up, then decide what follows. Put an event planner bot over the top — the senior producer — and split coordinators into tasked bots underneath. That is the same chief of staff pattern, scoped to one event. Jenny’s credibility claim here is practical: she has built a large pitch competition, produced big conferences, and knows venue selection. Use that as context for why she starts with logistics, not marketing copy.
What you are trying to accomplish: a bot team that can budget, scout, request RFPs, and negotiate without you scanning the internet for hours. What to prepare: a written brief that an event producer would actually need, not a vibe. What “good” looks like: the planner can build a budget, Scout can search against that budget and those constraints, and outreach has parameters. What failure looks like: a long list of neighborhoods and no criteria, which is where the team was when she joined.
Lock venue and event criteria before anyone searches
Venue searching used to take event producers hours of scanning. Jenny’s point is that a search bot only beats that if the brief is specific. San Francisco is 49 square miles, a hot market, and heavy on legality. More detail in the brief means a faster path to availability.
Write these down before Scout runs:
- Guest count. The working range discussed was about 100, maybe more, up to 200. The budget prompt they started feeding the bot used 150 as a number you can change later.
- Space. Warehouse-style, open, with room for a gallery. The leading priority was a great space and a fun environment, then working around the constraints that space imposes.
- City and county. San Francisco; Scout searches venues in the county that meet the brief.
- Date window. Sometime in the next month, longer if permits force it.
- Time of day. Evening. Not every venue is open past 9-to-5, so this has to be in the search, not assumed.
- Alcohol. No, in this run, to keep the event simpler. California, she says, loves permits; her last conference needed more than a dozen types. Alcohol is the fork that adds a lot of them.
- Food. Served hot food, not grab-and-go.
- Kitchen vs. catering. On-site prep adds a logistics layer. A catering company that brings food in is cleaner, but then the venue must accept that caterer. Many San Francisco venues have preferred catering vendors.
- Vendor posture. If the space is the priority, lock into the venue’s constraints rather than hunting for fully bespoke catering. A Grok bot can research those preferences once the space is the lead criterion.
- Event type in the prompt. Say “pop-up event,” not only “pop-up.” A pop-up could be outdoors, which changes venue search.
The more specific we can get in the venue search the faster we can actually get to a result of the availability.
Do not have Scout go look until this list exists. The speaker’s sequence is criteria first, search second.
Build the production budget before you request RFPs
They had not budgeted the event. Jenny would not let negotiation start there. Establish a baseline for venue rates, food rates, and how those line items tie together. SF is one of the pricier cities for this. Kick off price-line research with the same event planner bot that will own the budget.
Ask that bot, as a senior event planner in San Francisco, to create the first production budget as a template. Line items she named:
- Venue
- Food and beverage (F&B)
- Marketing
- Staffing and other production costs
Staffing has a split you must encode. Some venues include staff; some do not. Grok bots can carry planning most of the way, but you still need people on registration and security. Put those humans in the budget instead of pretending the bot team is the floor team.
What “good” looks like: a template the planner can reuse, with F&B, venue, marketing, and staffing visible before anyone talks to a building. What failure looks like: requesting availability and then discovering the venue’s preferred caterer or included staff blows up a number you never wrote down. After the budget exists, the same bot can request RFPs from venues that already meet the qualifications.
Scout, email RFPs, and negotiate inside budget parameters
Scout is the search bot: which venues in San Francisco County meet the brief. It should work in conjunction with the rest of the event details, not as a one-off chat. Once venues qualify, outreach is email. Jenny was asked whether bots should call. She would definitely integrate by email: ask for RFPs and availability, then go back and forth.
Negotiation is a taught skill, not a personality trait you hope the model has. She set a framework for how to negotiate on the clothing-resale bot, including parameters around budget, and that bot was negotiating bids on Poshmark, Depop, and Mercari while she talked. Steal the structure, not a secret script she did not give: establish budget first, give the bot the number and the rules, then let it request and counter inside that box.
What “good” looks like: RFPs going only to venues that match capacity, evening hours, food, and catering rules, with a bot that already knows the ceiling. What failure looks like: negotiating before a baseline, or picking a venue that will not accept your caterer after you have already fallen in love with the room.
Prompts, bot instructions, and workflows
Only one prompt was actually drafted in the conversation. The rest are reconstructions from how Jenny described the jobs. Do not treat the reconstructions as tested system prompts; they are the brief she gave, cleaned up so you can paste and then fill brackets.
Event planner bot — production budget
Reconstructed from the speaker’s description and the prompt the team started feeding a Grok bot.
You are a senior event planner in the city of San Francisco creating a budget for a pop-up event that will have 150 guests.
We need a production budget template that includes:
- Venue costs
- Food and beverage (F&B)
- Marketing costs
- Staffing and production costs
Notes:
- Specify pop-up event, not only pop-up, because a pop-up could be outdoors.
- Guest count is a working number and can be changed later (range discussed: 100 to 200).
- Some venues provide staff; some do not. Show that split.
- Even with Grok bots doing planning, include human beings for registration and security.
- San Francisco is a pricey market. Establish a baseline for venue rates and food rates and how those line items tie together.
After the budget, you can request RFPs from venues that meet the qualifications in the brief.Scout — venue search
Reconstructed from the speaker’s description. No full prompt was typed.
You are Scout, a venue search bot for San Francisco County.
Do not start a wide internet scan. Identify venues that meet this brief:
- Location: County of San Francisco
- Event type: pop-up event (indoor space unless the brief says otherwise)
- Capacity: [guest count]
- Space: [e.g. warehouse, open, gallery]
- Timing: [date window]; evening hours required (exclude venues that are 9-to-5 only)
- Alcohol: [yes/no]
- Food: served hot food
- Kitchen: [on-site prep vs. catering company bringing food in]
- Catering rules: flag preferred-vendor requirements and whether outside catering is accepted
- Priority: a strong space and environment; work around the venue’s constraints
Return venues that match, with anything you can find on availability. Specificity is the point; do not dilute the criteria to get a longer list.RFP and negotiation — email, inside a budget box
Reconstructed from the speaker’s description. She did not dictate a full negotiation script.
You handle venue outreach by email, not by phone.
1. Send RFPs only to venues that already meet the Scout brief.
2. Ask for availability and pricing against the production budget.
3. Negotiate back and forth using this framework:
- Budget ceiling: [amount or range]
- Must-haves: [capacity, evening, food service, catering rules]
- Tradeoffs allowed: [e.g. lock into preferred vendors if the space is right]
4. Do not agree to terms outside the parameters. Surface the constraint and wait.Wire this the way Jenny describes the rest of the staff: event planner on top, Scout and outreach as tasked bots, Master Chief (or your equivalent) to put them in one group conversation when the project needs to move.
Measurement and business impact
No time-saved percentages, dollar amounts, lead counts, or support-load numbers were given. The checkpoints Jenny actually used are qualitative. She is running seven businesses and a new agency with 22 Grok bot employees, including while raising a baby. A clothes-resale bot was negotiating bids in the background during the conversation. Venue search is framed as work that used to take event producers hours and hours of scanning, which is the job Scout is supposed to compress — after the brief is specific. The on-site checkpoint is the opposite: planning can be mostly bots; registration and security still show up as human line items in the budget.
Pitfalls and guardrails
The failure mode she kept blocking is searching too early. A bot that “finds venues in SF” without evening hours, catering rules, guest count, and food service will not get you to availability faster. The second failure is negotiating before a baseline. Teach the bot to negotiate only after venue and F&B rates have a box around them.
Brand and customer-facing risk on the event itself is mostly operational here, not voice-tone. Preferred-vendor lock-in can kill a catering plan. On-site kitchens add a logistics layer the team had not priced. Outdoor vs. indoor is easy to get wrong if the prompt only says “pop-up.” California permits are not a footnote; she cited more than a dozen types on a prior conference, and alcohol is the detail that explodes that list. They cut alcohol to keep the first pop-up simple.
Data and privacy beyond ordinary business use were not discussed. Keep a human in the loop where she said you must: registration, security, and any live floor work the bots cannot do. Keep a human in the loop where the method is still thin: final venue choice when the space is the product, and any negotiation that would break the budget parameters you wrote.
Even though the Grok bots can get us most of the way, you are still going to need human beings doing the registration and the security.
7-day implementation plan
This follows the order she taught, not a generic sprint. The team on the call was behind on outreach; you should not spend the week collecting venue names with no brief.
Day 1. Map the businesses or workstreams you actually run. Name the chief of staff bot. List the individual jobs worth a specialist (inbox, meetings, money, search, planning). Do not stand up 22 bots. Write the one-job rule down so you do not merge them later.
Days 2–3. Stand up two or three core-function bots only. If you run meetings, give Scribe a consistent intake (Jenny uses Wispr Flow on every meeting, then delegates to sub-agents). If money is messy, give the CFO setup receipts and bookkeeping as its only job. If email is the bottleneck, give Boxy the inbox and nothing else. Practice spinning them into one group conversation on a live task so the overlay is real.
Days 4–5. If the first project is an event, lock the criteria list: guest count, space type, city, date window, evening vs. daytime, alcohol, food service, kitchen vs. catering, vendor posture. Feed the event planner bot the production-budget prompt. Tell it to establish venue and F&B baselines for your city and to include human registration and security. Do not send RFPs yet.
Days 6–7. Run Scout against the brief for your county. Email RFPs only to venues that match. Load negotiation parameters from the budget. Decide which constraints you will accept if the space is right (preferred caterers, included vs. outside staff). Write down the human roles for registration and security so they are in the plan, not an afterthought the week of the event. Stay at one pop-up until this loop works.
Get the first bot on a single job
The business outcome is not a cleverer chatbot. It is a staff you can run while you are in another company, another meeting, or at home: specialists on rails, a chief of staff that can assemble them, and an event workflow that refuses to search or negotiate until the brief and the budget exist. Watch the video for the walkthrough of the budget prompt and how those bots get assigned.
FAQ
Do I need 22 Grok bots to use this model?
No. Twenty-two is how many Jenny said she is running across her companies and Gemini Media, not a minimum viable staff. The method is one core function per bot, plus a chief of staff that can spin the right ones up together. Start with the jobs that already eat your week — inbox, meetings, bookkeeping, or a single project lead — and add specialists when a new function shows up.
Chief of staff bot vs. only specialized Grok bots?
Jenny kept both. Specialists do the individual tasks; Master Chief oversees the portfolio and pulls them into a group conversation on a project. Shaan did not like the chief of staff model; she used it anyway because she is running multiple businesses as one “Jenny Co.” If you only have specialists and no overlay, you still have to be the person who assigns every task by hand.
Should venue bots email or call?
Email. She was asked about giving bots the ability to call places and said she would definitely integrate outreach by email: RFPs, availability, then negotiation back and forth. Her live example of negotiation is a bot dealing on Poshmark, Depop, and Mercari inside a framework she set, not a phone script.
Do I search venues before I have a budget?
No. She would not start Scout as a wide search, and she would not teach negotiation until a baseline existed for venue and food rates. The event planner bot writes the production budget first, including F&B, marketing, and staffing. RFPs then go to venues that already meet those qualifications.
Event planner bot vs. a human event producer?
The bot is a reverse-engineered producer: senior planner on top, coordinator bots underneath. It is meant to take the hours of scanning, budgeting, and RFP traffic. It is not a replacement for people on the floor. She was explicit that registration and security still need human beings, and that some venues include staff while others do not.
On-site kitchen vs. a catering company?
On-site prep is another logistical layer. A catering company that brings food in is simpler, but the venue has to accept that company, and many San Francisco venues have preferred vendors. If the space is the product, she would choose the room and work around those constraints, then have a bot research the vendor rules — not the other way around.
Do I need permits in the brief even for a simple pop-up?
Yes, as a planning constraint, especially in California. She said the state loves permits and that her last conference had more than a dozen types. Alcohol is the detail that makes that worse; they chose no alcohol to streamline. If permits push the date, the brief should allow a window longer than “next month.”
What if I care more about vibe than logistics?
Write the vibe as a priority, then still fill in logistics. The team wanted a fun, open, gallery-like warehouse and was willing to live with the venue’s vendor constraints. Jenny still made them decide evening vs. daytime, hot food vs. grab-and-go, kitchen vs. catering, and guest count before any bot went looking. Specificity is what makes the search fast; the environment is what you rank results by.