Capacity. Planning is open. Company implementation is full — join the waitlist.
Join the waitlist
Simple Generative Foundation · Georgia and London
SGF sits with your team, talks through how you actually work, and finds the time-consuming processes AI should take — so you spend hours making money, not falling behind. Grok, Claude, and ChatGPT. Planning is open.
Agents and models
Data / system of record
Apps and portals
Knowledge
CRM
Workspace and chat
Mail, calendar, files
Commerce, payments, books
Support
Meetings and scheduling
Work and automation
Code and design
The offer
We come in and see what you do. We talk through how work actually moves. Then we find the slow, time-consuming jobs AI can take — follow-ups, filing, the same reply forty times — so the team is not left behind while everyone else is already using this. Planning is open. Company builds are waitlist-only.
What we actually install
Grok Bot is xAI’s always-on agents. One bot, one job. Three bots run three jobs at once. They share a cloud computer. They pass files and records. They message each other. You are not the middleman. You spend the hours making money.
Inbox. Ops. Follow-up. Not one chatbot pretending to be the company.
Three jobs, one computer. A lead does not wait for someone to copy it from mail to the base.
Files, records, a short message between bots. The human is not the swivel-chair.
Airtable, HubSpot, Slack, the inbox, meeting notes. We do not rip out what works. Outbound stays a draft until a named yes.
Typical first three
Inbox
Mail and forms become records. Qualify. Route. Draft the first reply.
Ops
Dispatch, exceptions, the book. Real capacity. Emergencies escalate.
Follow-up
The next text, the next RO, the declined estimate. Draft until a named person says send.
The models we suggest
Grok Bot is the agent layer. The reasoning seats we put people on today are Grok, Claude, and ChatGPT. Gemini if you already live in Google — it is not the default.
xAI
Live operations. Fast reasoning. The model when the work cannot wait on a research paper.
Anthropic
Long documents, careful writing, policy, and analysis — when getting it right matters more than getting it first.
OpenAI
The seat most teams already have. We connect its apps and hand the work into Grok Bot and the record.
Coming soon
A private in-home or on-site server that runs the company’s AI. Customer lists, quotes, and files stay on your hardware — not in someone else’s cloud. Join the waitlist.
Models run on a box you own. Quotes, customers, and files stay in the building.
Inbox, follow-ups, filing — the same work we set up on Grok, Claude, and ChatGPT, on your server when you need it that locked down.
Not for sale yet. Leave your information and we will tell you when a slot opens.
In plain language
We come in, see what you do, talk through your processes, and find the slow jobs AI should take — so the company does not fall behind in this new world. Here is what that looks like in a normal week.
S
If a person who does the real work cannot use it on a Tuesday, it is not done. No science-fair stack.
G
The AI types, files, follows up, and drafts. People decide. The bot does the swivel-chair work.
F
One place the business actually lives — so the company is not stuck in a notebook, an inbox, or one person's head.
How we set it up
01
Not a survey. We watch a real week: quotes, orders, inbox, meetings — how the company actually runs.
02
Who does what, where it stalls, what gets retyped. You already know the pain. We write it down so it can be fixed.
03
The jobs that eat hours and do not make money. Those get a bot. The work that needs a person stays with a person.
04
Grok, Claude, ChatGPT, and Grok Bot on the stack you already have. Then we show the team the handoff — before the rest of the market leaves you behind.
What gets easier
Now. The same reply, typed by hand, forty times a week.
After. The bot drafts it from the last quote. You tap send.
Now. A recording nobody opens. Action items live in someone’s memory.
After. Notes become a task on the right person, in the system you already use.
Now. A hunt through texts, email, and a spreadsheet with three versions.
After. One place to look. The order, the quote, the next step — already there.
Now. They live in one person’s head. If that person is out, they die.
After. The bot nudges. You close. Nothing waits on memory.
Now. Two weeks of shadowing to learn “how we do it here.”
After. The steps are in the system. They start on the work, not the folklore.
Now. Busy work until 7pm. The work that makes money waits.
After. The bots take the swivel-chair. You spend the hours on customers and cash.
How it protects the business
Quotes, customers, and “how we do it” live in the system — not in a departing employee’s inbox.
People stop pasting private data into a random chat. The bot works inside your tools, on the records you already own.
Sales does not get the whole books. A client does not see another client. Permissions are the product, not an afterthought.
Sick day, vacation, a person leaving. The next step is still in the record. The business does not pause.
What is open
Open now
We come in, see what you do, and write the plan — then sit with your team while they learn the handoff so busy work leaves their week.
Waitlist
Full implementation for company-scale work — mid-market through large organizations. Currently waitlisted. We still write the architecture so you are ready when a slot opens.
Open now
One person running a real company. We set up the models, notes, and stack — built to take more people later, not a toy.
Who it is for
Car dealerships (sales, BDC, service). Plumbers, HVAC, electrical, roofing, GCs. And the same pattern for any company whose slow work is intake, booking, and follow-up.
A founder who is the company, a mid-market operator, or a large organization with volume already in the stack. We take company work — not side projects.
We do not sell AI employees that replace staff. Role contracts, named-yes gates, a 90-day sequence, weekly coaching.
Where AI actually helps
We sell an operating system — role contracts, named-yes gates, a 90-day sequence, weekly coaching. Not AI employees that replace staff. We do not stand up a full AI OS on day one.
01
Answer the call, the form, the after-hours overflow.
02
Scripted questions. Live inventory or real capacity — not a guess.
03
Write the appointment or the request into the system you already run.
04
Negotiation, emergencies, complaints — a named person takes it.
Outbound SMS and email need a named human yes. Inbound answer and book stays inside approved scripts.
Sales · BDC · service · inventory
Internet leads, missed calls, and service status — written back to the CRM or DMS. People keep negotiation and complaints.
Qualify, then book a test drive or a service visit.
The call or form lands in the CRM or DMS. After hours and overflow do not die in voicemail.
SMS or email with live inventory and guardrailed pricing.
Negotiation and complaints escalate to a person. The bot does not make a deal.
Repair-order status on one customer record.
A missed service call becomes a write-back, not a second sticky note.
Declined service, recall, or a lapsed VIN — as a draft.
Outbound only after a named human yes. Nothing sends itself.
HVAC · plumbing · electrical · roofing · GCs
After-hours voice, dispatch, estimates, and inbound SMS — booked against real capacity. Emergencies escalate by keyword.
Book against real capacity, or open a request.
Emergency keywords go to a person. Everything else becomes a job or a request in the board.
Schedule an estimate or a job, escalate, or text back after a hang-up.
Hang-up texts stay drafts until someone approves them.
Real hours, real services, real availability.
It does not invent a slot. If the book is full, it waitlists or opens a request.
Drafts only.
Never auto-send. A named yes is the gate.
The bot may add a name and draft “a slot opened.”
The send waits for a named human approval.
The operating system
Each bot has a job, a script, and tools. It does not wander the company.
SMS and email campaigns stay drafts until a named human says yes. Inbound answer and book stays inside approved scripts.
One pilot. Then booking rules. Then optional outbound. We do not stand up a full AI operating system on day one.
Your people keep the work. We coach the handoff. This is not AI employees replacing staff.
How we roll it out
Dealer missed-call, or trades after-hours. Shadow only at the start. Outbound never ships without a named yes.
Days 0–30
Dealer missed-call rescue, or trades after-hours voice. No outbound. We watch it work next to the team.
Days 31–60
The bot can book against real hours and hold a waitlist. A person still approves what goes out. Weekly coaching.
Days 61–90
Declined-service or declined-estimate drafts may send after a named yes. Connect one CRM, DMS, or field system if the pilot is clean.
What you walk away with
Inbox, Ops, Follow-up. One shared computer. They pass the record. You are not the middleman. Hours go back to making money.
One bot, one job. Three jobs at once. Files and records pass between them. The human is not the swivel-chair.
Airtable, SmartSuite, or whatever you already run — structured so people and bots can trust it. The app layer your staff and clients actually click.
We stand it up and show them how. Then they own the handoff — not a demo that dies when we leave.
The stack
Select a name. Click or hover for what it does and when we use it. Data first — Airtable, SmartSuite, Coda, Monday, ClickUp, Sheets. Then the app layer, the CRM, meetings, and the rest.
Agents and models
Data / system of record
Apps and portals
Knowledge
CRM
Workspace and chat
Mail, calendar, files
Commerce, payments, books
Support
Meetings and scheduling
Work and automation
Code and design
Services
Inbox, Ops, Follow-up. One computer. The human is not the middleman.
Grok Bot is xAI’s always-on agents. One bot, one job. Three bots run three jobs at once on a shared cloud computer. They pass files and records. They message each other. Signed into Airtable, HubSpot, Slack, the inbox, meeting notes. We do not rip out what works. Then we show the team the handoff so they spend hours making money.
When one person is still the bottleneck — and the company will grow.
From a single operator to the first team. Models, meeting notes, and a source of truth built so it does not collapse when headcount and volume show up. Not a toy stack for a side project.
Agents that work in your systems, not next to them.
We map the jobs that should not be a human swivel-chair, then design agents that file, route, draft, update, and follow up. Evaluation, fallback, and a person who owns it — not a chatbot bolted onto the homepage.
The source of truth — Airtable, or what you already run.
Schema, relationships, interfaces, and automations. Airtable when it is the right base; SmartSuite, Coda, Monday, ClickUp, or Sheets when that is already the record. London specialists on Airtable. We do not start by ripping out a working system.
The software your team and clients actually use.
Noloco, Softr, Glide, or Stacker on top of the data. Roles, workflows, branding. Our London partners are among the strongest Noloco operators in the field — and we work in the app layer you already paid for.
Process your people and your agents can both follow.
Notion or Confluence. Structure, owners, and a place for the current way of working. If it is not in the wiki or the base, it is not the process.
The CRM, the store, the helpdesk, the glue — and the alternatives.
HubSpot or Salesforce or Pipedrive. Shopify or Woo. Zendesk or Gorgias. Zapier, Make, or n8n. Meeting notes from Granola, Circleback, Fireflies, Fathom, or Otter. Connected to whatever is the source of truth — Airtable or not. We do not rip out what already works.
A system your team can run without us.
We show the team how Grok Bot takes the busy work. Docs, office hours, and a cadence. Then they own the handoff.
Coming soon. Your own box. Your data stays home.
A private server in the home or on-site that runs the company’s AI so customer lists, quotes, and files never leave the building. Same jobs we set up on Grok, Claude, and ChatGPT — on hardware you own. Not for sale yet. Join the waitlist.
How a project runs
01
We sit with the people who do it. Quotes, orders, inbox, meetings. What you actually do — not what the org chart says.
02
How work moves, where it stalls, what gets retyped. We name the time-consuming jobs out loud so they can be handed off.
03
Follow-ups, filing, “where is that,” the same reply forty times. Those go to Grok Bot. People keep the work that makes money.
04
Grok, Claude, and ChatGPT on the stack you already run. Training so the team owns the handoff. The point is not a demo. The point is not getting left behind.
The firm
Simple Generative Foundation puts agents inside the company — on the stack you already run — from a single operator to a large organization. We work with a London practice of Airtable and Noloco specialists: some of the strongest operators on those platforms. Together we take company-scale jobs: agents, data architecture, and the app layer. Not a chatbot shop. Not a six-month strategy deck.
Questions
Start
Inbox, Ops, Follow-up. They pass the record. You are not the middleman. Planning is open. Implementation is waitlisted.