The complete guide. Start with the map, then read about each module. Written so anyone on the team understands the whole machine.
Bank transactions, emails, rent payments, meeting notes and messages arrive from the real world
Files everything in one database, does the maths, matches payments and spots problems
Clear numbers, alerts when something needs attention, and one-click actions
This is the whole platform on one picture. Data flows in from the left, the app in the middle organises it, and you act on the right. Every coloured box is a module. Click one to jump to its full guide below.
Four stories explain most of what the platform does. Follow each chain from left to right.
Every bank movement is copied in, filed in the right place, and counted in three different views without anyone typing anything.
Invoices and meeting summaries never sit unread. Scripts read the labels and turn emails into records, tasks and Slack messages.
The platform notices on its own, raises the alarm, starts the chase, and stands down the moment rent restarts.
Every AI feature on the platform talks to Claude AI through a secure relay (a small cloud programme that holds the keys so the app never has to).
Each section below stands on its own. Read the promise line, follow the flow, then open the expanders for the step-by-step detail.
The whole business on one screen, refreshed every 15 minutes.
This is the first screen you see when you log in. The app loads the core data from Airtable: transactions (bank movements), tenancies (rental agreements), tenants, costs (regular bills) and businesses. Then it does the maths and shows the results as KPI cards. A KPI (Key Performance Indicator) is a number that tells you at a glance how the business is doing: money in, money out, profit, empty rooms, overdue bills.
Each business also gets its own strategic KPI cards, so you see which company is pulling its weight and which one needs attention. Click any card to expand the detail behind the number.
Core tables arrive from Airtable
KPIs are worked out fresh each time
Click a card for the detail behind it
The health pill confirms all is fresh
Smart refresh: the data reloads every 15 minutes, but politely. If you are clicking around, it waits until you pause so your work is never interrupted.
Shared data: the dashboard loads the core tables once and shares them with every other tab, so the rest of the app opens instantly.
Fast returns: a copy is kept in your browser (IndexedDB, a small local store), so the page appears at once when you come back, then quietly updates in the background.
Connects to: Accounts, Profit & Loss, Money Confidence, CFVs. They all read the data this page loads.
Decide where each business is going this quarter, with an AI coach checking your thinking.
This is where plans are made. Pick a business and a quarter (a three-month block of the year). Write the objective: what the business does, who it serves, and what must be true by the end of the quarter. Set your long-range targets too: 9-year, 3-year and 1-year.
An AI wizard acts as a boardroom mentor. It reads your plan, asks hard questions, and points out weak spots before reality does. When the plan is solid, you push the quarterly projects straight into Tasks & Projects, so the plan becomes real work with owners and dates.
Choose a business and a quarter
Objective, targets and up to 3 projects
The AI mentor stress-tests the plan
Projects land in Tasks & Projects
Saved in Airtable: objectives, strategies and projects each live in their own table, one row per business per quarter.
AI mentor: the wizard sends your plan plus context to Claude AI through the secure relay and returns pointed questions, not flattery.
Connects to: Tasks & Projects (receives the pushed projects), Plan Builder (for brand-new businesses), Leadership Dashboard (tracks the results).
Answer questions in a chat. Get a complete business launch plan.
Starting something new? The Plan Builder is an AI coach that interviews you. You type answers in a chat window and it drafts each section of a business launch plan for you: who the customer is, what you charge, how the work gets done, how people hear about you, and what the money looks like.
You stay in charge. The AI drafts, you correct, it redrafts. At the end you hold a full plan you understand line by line, because you built every part of it in conversation.
The AI asks about your idea
Each plan section is written for you
Correct anything; it redrafts
A finished plan, ready to act on
AI coaching: the chat runs on Claude AI via the secure relay, using the same boardroom mentor voice as the Strategy wizard.
Connects to: Objective & Strategy. Once the business is live, its quarterly planning happens there.
Everyone knows what to do today, and Slack tells them the moment it changes.
This is the engine room of daily work. The My Day view shows each person their own list for today. The full task list, the kanban board (a board with columns you drag cards across), and the capacity view (how much work each person is carrying) give you the wider picture. Recurring tasks recreate themselves after completion, so weekly and monthly routines never get forgotten.
Contractor jobs live here too, and so do meetings. When a meeting summary email is labelled in Gmail, a script logs the meeting, creates the agreed tasks with owners and dates, and sends each owner a Slack DM (direct message). Nobody leaves a meeting wondering who is doing what.
Typed in, pushed from Strategy, or born from a meeting email
Stored in the Tasks table in Airtable
The owner is told straight away
Board updates; recurring tasks respawn
slack-notify worker: a small programme in the cloud watches for task assignments, completions and comments, and sends the right person a Slack DM each time.
Meetings script: label an email "meeting summary" in Gmail and a script reads it, logs the meeting, creates the tasks it mentions, and DMs the owners.
contractor-bot: when a contractor replies in Slack, an AI reads the message, understands it, and updates the matching job.
Connects to: Objective & Strategy (projects arrive from there), Team (owners come from the directory), CFVs (chase tasks are created here automatically).
Every property, room, tenant and safety certificate in one place.
CRM stands for Customer Relationship Management, which in plain words means "the book of everyone and everything you look after". For a property business that book holds four layers that link together: properties (the buildings), rental units (the individual rooms and flats inside them), tenancies (the rental agreements), and tenants (the people).
Compliance is built in. Gas and electric safety certificates are tracked with their expiry dates, colour-coded so you see danger before it arrives: green means current, amber means expiring soon, red means expired. Staying legal stops being a memory game.
Open a building
See its rooms, tenancies and tenants
Check gas and electric expiry colours
Fix anything amber or red
Linked records: properties, units, tenancies and tenants are separate Airtable tables joined together, so a change in one place shows everywhere.
Expiry alerts: certificate colours are computed from the expiry dates every time the page loads. Nothing to maintain by hand.
Connects to: Accounts (rent roll comes from tenancies), CFVs (missed rent is detected per tenancy), Tasks & Projects (maintenance and certificate jobs).
An inbox where the oldest message shouts loudest, so nothing slips.
Email is where things get missed. Inbound Comms shows your Gmail sorted by label (Inbox, To Pay, Compliance, Archive and more) and orders every list by age, oldest first. The message most at risk of being forgotten is always at the top, staring at you.
Invoices handle themselves. Label an invoice email "3. to pay" and a script picks it up on its next pass, reads the sender and details, and creates an invoice record in Airtable. It then appears in Accounts under AP Variable, ready to approve and pay. The email becomes a record without any copying and pasting.
Lands in Gmail as normal
Inbox, To Pay, Compliance and so on
Nothing hides at the bottom
"3. to pay" emails become records
Gmail invoice script: a Google Apps Script (a small programme that lives inside your Google account) runs on a 15-minute timer, reads the "3. to pay" label, and writes invoice records into Airtable.
Live connection: the page talks to Gmail directly through Google sign-in, so what you see is your real inbox, not a copy.
Connects to: Accounts (AP Variable receives the invoices), Tasks & Projects (the meetings script uses the same email-to-record trick).
Record how a job is done once. The platform turns it into a process, a manual, and then an AI agent.
This module is the heart of the whole product. Most business work repeats: the same steps, week after week, held in somebody's head. Systemisation gets those steps out of heads and into the platform. You map the customer journey in stages, break each stage into processes, and each process into numbered steps. Every step is marked AI, Staff, or Both, so you always know what a human still has to do.
Then comes the special part. Record yourself doing the job once (a short screen video is enough). The pipeline extracts the steps, writes the SOP (Standard Operating Procedure, a step-by-step instruction manual), and turns the process into an AI agent: a programme that does the job for you from then on. That is the promise the platform is built around: up to 90% of repeatable work done by AI, not people.
Do the job once on video
The pipeline extracts each step
A manual anyone follows
The job runs itself from now on
Process to agent: one recorded video in, one working AI agent out. The script of the recording is the raw material; the SOP is the checkpoint where a human confirms the steps; the agent is the finished product.
Business blueprint: as processes stack up, they form a full picture of how the business runs, like a build manual for the whole company. That document is an asset in its own right.
Connects to: Skills Library (finished agents show up as runnable skills), Site Map & Guides (SOPs are versioned there), Tasks & Projects (staff steps become tasks).
A shelf of ready-made AI helpers. Press Run and the job gets done.
A skill is a packaged AI job: a set of instructions the AI follows to complete one task from start to finish. The library holds them all in one searchable place. Onboarding a new tenant, reconciling transactions, generating a tenancy document: each is a skill you run with one click instead of a job you do by hand.
Mark your favourites as active so they sit at the top. When you press Run, the skill-runner worker (a small cloud programme that runs the skill safely, away from your browser) does the work and files the result in the right table.
Search or filter the shelf
One click starts the skill
The skill-runner does the job in the cloud
The result lands in the right table
skill-runner worker: skills execute in the cloud with their own credentials, so the browser never holds secrets and long jobs keep running after you close the tab.
Growing shelf: every process that finishes the Systemisation pipeline arrives here as a new runnable skill.
Connects to: Systemisation (the source of new skills), Reconciliation and Operations CRM (common skill targets), Tasks & Projects (skills log what they did).
The business remembers everything, even the things you forgot you knew.
Knowledge leaks. A decision made on a Zoom call, a note typed on a phone, a document filed in Drive: three months later nobody remembers where the answer lives. The AI Brain fixes this by gathering it all in one memory. Every night a publisher (an automatic collector job) reads your Google Drive files, Apple Notes, and Zoom meeting summaries and files what it learned into a brain feed table in Airtable.
The Today view shows you each morning what the brain took in overnight: new documents, fresh decisions, meeting outcomes. Sensitive items are shown as link-only pointers, so private content is referenced but never copied.
Files, notes and meetings pile up all day
The publisher reads Drive, Notes and Zoom
New knowledge lands in the feed table
Today shows what the brain learned
Nightly publisher: runs once a night, on its own, and writes to a private feed table in Airtable. No buttons to press.
Privacy rule: sensitive items are stored as a link back to the original, never as a copy.
Connects to: Tasks & Projects (meeting outcomes become tasks via the meetings script), the AI features platform-wide (the brain is context the AI draws on).
One number that tells you how much you are safe to spend today.
Bank balances lie. The account shows £20,000 but wages are due Friday, two bills land Monday, and half of the balance is already spoken for. Money Confidence does the honest sum for you. It starts with cash in the bank, then subtracts the wages float (money set aside for wages), a timing cushion (a buffer for bills that land at odd moments), and any fixed costs not yet covered this month. What is left is the "safe to act today" figure.
A traffic light makes the answer instant: green means spend with confidence, amber means be careful, red means stop. A priority spending order tells you what to pay first when money is tight.
Live balances from the bank feed
Wages float, cushion, unpaid fixed costs
Safe-to-act figure with a traffic light
Spend, hold, or follow the priority order
Always current: the figure is recomputed from the live bank feed and the fixed-cost list each time the tab loads.
Connects to: Accounts (bank balances and fixed costs), Leadership Dashboard (the same core data), Wealth (the long-term view of the same money).
Watch your net worth move, month by month, with property values kept fresh by AI.
Net worth is one sum: everything you own minus everything you owe. Wealth takes that snapshot every month and lines the snapshots up so you see the direction of travel. Assets are grouped into six classes (types of thing you own, such as property, cash and investments), so you see not only the total but the shape of it.
Property values go stale fast, so an AI job refreshes them monthly. It searches the web for current market evidence for each property and files its suggested values as "pending review". Nothing changes until you approve or reject each one. Income buckets track where your monthly income comes from and how it is allocated.
A net worth snapshot is taken
Web search suggests fresh property values
Accept or reject each pending value
Net worth updates across six asset classes
Monthly valuation job: a scheduled job runs on the 1st of each month, does AI web research per property, and writes pending-review valuation records. Human approval is the gate; the AI never overwrites a value on its own.
Connects to: Operations CRM (the property list), Money Confidence (short-term money view of the same picture), Leadership Dashboard.
All money in and out, in six tidy views.
Accounts is one module with six sub-tabs, each answering one question. AR means Accounts Receivable (money owed to you). AP means Accounts Payable (money you owe). Fixed means it repeats every month; Variable means it comes and goes.
AR Fixed is the rent roll: every tenancy and the rent expected each month. AR Variable is the invoices you send out and whether they have been paid. AP Fixed is your recurring bills, matched against the bank feed so overdue and odd-amount payments get flagged. AP Variable is the invoices you receive, arriving automatically from Gmail, waiting for your approve-and-pay. Transactions is every bank movement, searchable and filterable. Bank Accounts watches the feeds themselves and warns you if any account stops syncing.
Rent, bills, invoices, payments
Fintable syncs the bank; Gmail sends invoices
Each movement lands in the right view
Overdue and unexpected items stand out
Fintable sync: a service that copies bank transactions into Airtable, roughly hourly. A separate scheduled job mirrors those transactions to Supabase (the future database) every hour, ready for the platform's next chapter.
Gmail invoice script: fills AP Variable automatically from labelled emails, every 15 minutes.
Connects to: Reconciliation (files every transaction), Profit & Loss (counts them), Money Confidence (uses balances and fixed costs), CFVs (rent detection reads the same feed).
Are we making money? Twelve months, one table, and an AI to explain it.
A P&L (Profit and Loss statement) answers the only question that decides survival: is more money coming in than going out? This one is built live from your categorised bank transactions, twelve months side by side. Money in, minus money out, equals profit, split by category and by business, so you see which company earns and which one leaks.
Charts show the trend at a glance. Click any number to see the exact transactions behind it, so no figure is ever a mystery. An AI financial analysis reads the whole statement and writes a plain-English commentary: what improved, what worsened, and what deserves your attention.
Reconciliation has filed each transaction
Totals by month, category and business
Money in minus money out
Plain-English commentary on the numbers
Quality upstream: the P&L is only as accurate as reconciliation. That is why the Reconciliation engine tracks its own accuracy score with a 90%+ target.
Connects to: Reconciliation (its categorised output is the raw material), Accounts (Transactions view is the drill-down), Leadership Dashboard (headline profit KPI).
Every bank transaction filed in the right place, and it learns as it goes.
Reconciliation means matching each bank transaction to what it was for: this payment is October's rent from tenant A, that one is the insurance bill, that one belongs to business B. Done by hand it is hours of tedium. Here, the engine does it. It reads each unfiled transaction and suggests the match, using a knowledge base of rules built from every match you have ever confirmed.
You stay the judge. Confirm a suggestion and the engine gets a point; correct one and it learns the new rule. Accuracy is tracked openly with a target above 90%. Splits handle one payment covering several things, and undo reverses any mistake safely.
Unfiled transactions from the bank feed
The engine proposes a match for each
Confirm or correct in one click
Each answer sharpens the next batch
Knowledge base: rules built from vendor names, amounts, descriptions and past confirmations, applied to every new batch.
Accuracy log: every suggestion is scored right or wrong, so the accuracy figure is real, not a guess.
Field sync: after a match, linked cost records are updated in the background with the latest payment date, amount and variance.
Connects to: Accounts (its input), Profit & Loss (its output), CFVs (rent matches prove a tenant paid), Money Confidence.
Know the same day rent stops, and stop chasing the moment it restarts.
A CFV (Cash Flow Void) is a tenancy that has stopped paying: the room is occupied but the money is not arriving. The engine checks every active tenancy against the bank feed. No payment when one was due? The tenancy is flagged, a badge appears in the sidebar, and the chase workflow starts with the amount owed, days overdue, and contact details ready to use.
Every tenancy also gets a rent statement: a day-by-day ledger of rent charged versus rent paid, with a running balance. It shows exactly who owes what and flags when arrears reach the level where legal action becomes possible. For tenants on Universal Credit (a state benefit that pays rent in some cases), verification tasks are created automatically so payments get checked at the right time. And when a flagged tenant pays again, the system spots it and clears the flag itself, with an audit note.
A due date passes with no payment
CFV raised; sidebar badge appears
Workflow steps and tasks guide the follow-up
Rent restarts; the flag removes itself
Auto-detect and auto-return: detection runs when the tab loads; auto-return clears flags when payment reappears in the bank feed, with an audit comment for the record.
Rent statement engine: builds each ledger on demand from tenancies and transactions, and flags Section 8 eligibility (two or more months of arrears, the legal threshold).
Connects to: Operations CRM (tenancies and tenants), Accounts (the bank feed proves payment), Tasks & Projects (chase and UC tasks), Leadership Dashboard (badges).
Who is on the team, what they know, and how they are doing.
The Team module is the people layer of the platform. The directory lists everyone with their role and department, and the org chart shows who reports to whom. Onboarding gives each new starter a checklist so nothing gets skipped in week one. Training coverage is tracked with progress bars, so you see who is ready for which work.
Reviews keep feedback regular rather than annual, and the handbook holds the rules of how you work, in one place everyone reads from.
New person added to the directory
Checklist covers week one
Progress tracked per skill
Regular reviews and achievements
Connects to: Tasks & Projects (assignees come from the directory, and Slack DMs reach the right person), Systemisation (SOPs are the training material for staff steps).
Raw footage in, finished content out.
Marketing content dies in the gap between "we filmed something" and "we published something". The Content Machine closes the gap with a pipeline: upload the raw footage, move it through editing, and publish. Each piece of content is a card that travels the pipeline, so you always know what is stuck and where.
AI does the heavy lifting on the words: captions, post text and descriptions are generated for you to approve rather than write from scratch. The module runs as its own app, shown inside the platform in an iframe (a window to another page), so it updates on its own schedule without touching the rest of the system.
Raw footage enters the pipeline
Captions and post text generated
Review, trim and approve
Finished content goes out
Separate app: the Content Machine lives in its own code repository and is embedded in the platform, so content work and business-data work never trip over each other.
Connects to: the platform shell (navigation and design tokens), and your publishing channels downstream.
A map of every page, and proof its manual is up to date.
The platform keeps a register of every page it contains, each with a version number. Every page also has an SOP (Standard Operating Procedure, its instruction manual), and the SOP has a version number too. The Site Map compares the two. If a page has changed but its manual has not, the mismatch is flagged, so the guides never quietly rot.
It is also the quickest way to get anywhere: every page, guide and link in one directory.
New code is pushed live
The page number rises automatically
Page version checked against its guide
The guide gets updated, then the flag clears
Auto version bump: a scheduled job watches every code push and raises the version number of any page whose file changed. No human bookkeeping.
GitHub check: the page compares what is deployed against the latest code, so you know the live site is current.
Connects to: every page on the platform, and Systemisation (SOPs are written and maintained through the same discipline).
These run on their own. Each card says what starts it, what it does in one sentence, and where the result lands.
Reloads the core business data on a timer, but waits politely while you are clicking around so your work is never interrupted.
Runs a set of checks each time a tab loads (is the data fresh, do the numbers add up) and shows the result as an "N/N checks" pill, with a pending state while checks are still running.
Reads emails labelled "3. to pay" and turns each one into an invoice record, so bills file themselves.
Reads the meeting summary, logs the meeting, creates the agreed tasks with owners, and sends each owner a Slack DM.
Sends a Slack DM to the right person the moment their task changes, so nobody has to check the app to find out.
Copies new bank transactions from every connected account into the platform's data store, so the money picture stays current without any typing.
Copies the transactions across to Supabase (the platform's future database) every hour, keeping the next-generation data store warm and ready.
Researches current market values for each property using AI web search and files the suggestions for you to approve or reject; nothing changes without your say-so.
Gathers new Google Drive files, Apple Notes and Zoom meeting summaries and files what it learned into the business's memory, with sensitive items stored as links only.
Raises the version number of any page whose file changed, so the Site Map always knows which guides have fallen behind.
Flags any tenancy whose rent has stopped, and clears the flag on its own (with an audit note) the moment payment reappears in the bank feed.
Creates a verification task automatically when a tenant's Universal Credit payment (a state benefit that pays rent in some cases) is due to be confirmed.
Reads the contractor's message with AI, works out which job it is about, and updates the task with the news, so a text like "boiler fixed, invoice to follow" files itself.
The health checks above show their results as coloured dots next to every page in the sidebar. Here is the code.