07 Selected work

Two products on
the shelf. Five
built to order.

The first two are finished products, running today and rebuilt for the U.S. — they deploy in weeks. The rest were built to order. Every one shows the same three things: the hours the work took before, how those hours disappear, and what's left after. You can check our math — that's why we show it.

Verified

A fact checked inside the running system.

Client-reported

The owner's own account of what the work cost them before.

Projected

Where it lands at full deployment, with the baseline and the mechanism stated so you can judge it yourself. It's a projection, not an audited result, and we label it every single time.

Own product · Mannacore Personal

On the shelf · U.S. ready

Mannacore Personal — finances that run inside a text thread

Turns a text, a voice note or a photo of a receipt into organized finances — no new app, no spreadsheet anyone has to maintain. Built and running; rebuilt for the U.S. market.

The mess

Money leaves in a hundred small ways and none of them get written down.

A receipt photographed in a chat
A statement PDF nobody opens
A card statement with dozens of lines
A spreadsheet abandoned in week two
By the time the bill arrives, the month is a surprise

What we built

Send what you spent — typed, spoken or photographed
It comes back sorted: amount, category, method
A whole statement lands organized in seconds
Lives in the messaging app they already have open — SMS in the U.S.
No new app, nothing to remember to do

Where the AI acts

Reads a receipt photo and pulls out the items
Extracts the transaction table from a bank or card PDF
Decides category, method, and whether it's a duplicate
Never moves money
Anything it isn't sure about waits for a person

The result

A whole card statement sorts itself instead of being typed line by line
One expense logged in the time it takes to say it out loud
The invisible leaks — subscriptions, delivery apps, duplicates — become visible

Before

~4 hrs a month

Typing a card statement line by line, plus the daily entries nobody keeps up with.

After

Under 15 minutes

Send the file, speak the expense. Review what the system flagged.

Where the hours actually go

VERIFIED four ways in — text, voice, photo, PDF.

PROJECTED baseline: an ~80-line card statement at ~15 s per manual entry ≈ 20 min, plus daily logging most people abandon by week two. The after figure assumes the file is sent monthly and daily entries go in by voice.

The card PDF is read whole — about 80 lines pulled out at once instead of typed.

Each line is categorized against this person's own history, not a generic list.

Recurring expenses stop asking to be sorted — once you've done it, they do it themselves.

Only what the system is unsure about comes back for a human to confirm.

Its own core · its own database · its own rules

What it knows

Each person’s own wording, their merchants, their recurring bills and the way they actually describe a purchase.

What it enforces

Their categories and payment methods — plus duplicate detection, so the same expense never lands twice.

What it handles at once

A whole statement of dozens of lines — and every user’s messages at the same time, day and night.

Rebuilt for the U.S. — not translated

Taken out

Installment plans, instant-transfer receipts and the Brazilian tax rules — none of it exists here and all of it confuses an American user.

Put in

Sales tax by state, quarterly estimated tax set aside as income lands, and the payment methods people actually use: ACH, Zelle, check, card.

What that buys you

A finished product, not a promise. Deployed in weeks instead of built in months.

Where else this fitsBookkeeping intake · expense capture for field crews · any household or office drowning in chat and paper.

Mannacore Personal is our own product. Named and shown with full permission.

Own product · Mannacore Business

On the shelf · U.S. ready

Mannacore Business — the platform an owner will actually open

Shows an owner what the company is really doing, in plain language, without an accountant's screen. Built and running; rebuilt for the U.S. market.

The mess

The numbers live everywhere and nowhere.

A bank statement and a supplier's spreadsheet
Sales in a notebook, invoices in a chat, receipts in a drawer
Full ERP systems: dozens of screens, an accountant's vocabulary
So the owner opens it once and never again

What we built

Three tabs: did we make money, and where did it go
Sales, costs, receivables and clients arrive and organize themselves
Everything comes in through channels the business already uses
Simple on day one, a full system as the company grows

Where the AI acts

Reads every document and message that comes in
Classifies each one against the company's own history
Flags what looks wrong and drafts the summary
Nothing gets posted without a person seeing it

The result

The owner sees whether the business is profitable, in words they actually use
No more chasing paper — the data arrives and sorts itself
The tax side is watched, so it stops being a year-end surprise
One tool that grows with the company instead of being replaced

Before

8–12 hrs a week

Someone entering invoices, chasing receipts and rebuilding the month's picture by hand.

After

About 1 hr a week

The owner reviews what arrived and approves the exceptions.

Where the hours actually go

PROJECTED baseline: a small service company where one person handles bookkeeping admin — roughly 2 hrs a day of data entry plus a 1–2 day monthly close. The after figure assumes the company's main channels are connected; the remaining hour is review, not entry.

Sales, costs and receipts arrive through the channels the business already uses — no one uploads anything twice.

Every document is classified against this company's own supplier and client history.

The month's view assembles continuously, so closing it isn't an event.

The tax reserve is set aside as revenue lands, not discovered at the deadline.

Its own core · its own database · its own rules

What it knows

This company's suppliers, clients, price table and chart of accounts — not a generic industry template.

What it enforces

How this business classifies revenue and cost, and what its tax reserve should be, applied identically every month.

What it handles at once

Every document and message as it arrives, continuously, so the month is never assembled in a rush.

Rebuilt for the U.S. — not translated

Taken out

Installment plans, the Brazilian tax rules and their filings — every field that would make an American owner think the software was written for someone else.

Put in

Sales tax by state and jurisdiction, 1099 contractors alongside W-2 payroll, quarterly estimated taxes, and a QuickBooks export their accountant already knows how to read.

What that buys you

The tax reserve is the feature owners feel first: money set aside as revenue lands, instead of a bill discovered in April.

Where else this fitsRecurring services · retail and distributors · clinics and small practices · agencies.

Mannacore Business is our own product. Named and shown with full permission.

Independent financial advisory · our own practice

Live

A console that turns one busy advisor into a firm

The management layer behind an advisory practice: every client's portfolio, spending and review cycle in one place.

The mess

Each client's contract, portfolio and plan in separate spreadsheets
Reports built by hand, late at night, one at a time
Reviews slipped because nothing reminded anyone
No way to add clients without adding people

What we built

Clients text a photo or a message — no app, no portal
One console: spending, goals, portfolio, report on demand
Review cycles remind you on their own

Where the AI acts

Reads the receipt or message and sorts the transaction
Flags what looks off
Drafts the monthly narrative
The advisor reviews and signs — the advice stays entirely human, which is what regulators expect

The result

More clients served with the same hours
Every client gets a polished report, on time
Nothing slips — a renewal, a review, a follow-up
The practice looks like a firm, not one busy person

Before · a book of ~22 retainer clients

~18 hrs a month

Roughly 50 minutes per client, every month, rebuilding the same report by hand.

After

Under 4 hrs a month

The report is generated. The advisor edits the narrative and signs.

Where the hours actually go

PROJECTED baseline: ~22 clients on annual retainer, monthly report previously built by hand ≈ 50 min each. The after figure assumes about 10 min of advisor review and editing per client. The real gain isn't the 14 hours — it's that the same 18 hours could now carry a book three times the size.

Clients text in their spending as it happens — the advisor stops collecting.

Transactions arrive categorized, so the month is assembled before anyone opens it.

The report drafts itself against each client's goals; the advisor rewrites what matters.

Review dates come up on their own instead of living in someone's memory.

Its own core · its own database · its own rules

What it knows

Every client's goals, portfolio, history and review cadence, kept as records rather than as notes in a thread.

What it enforces

The practice's own methodology and report format — the same standard on every client, every month.

What it handles at once

The whole book at once, on schedule, without the advisor starting anything.

Where else this fitsWealth planners · insurance brokers · bookkeepers · tax practices · any book of recurring clients.

Built for and run inside our own advisory practice. Named and shown with full permission.

Financial clarity engine · multi-client platform

Live · UK & Brazil

Whatever the client sends, turned into a number they can act on

A white-label reporting engine deployed across six segments in two countries. Same platform underneath, a different trade on top of each account.

The mess

Every client sends something different: PDF, export, spreadsheet, CSV
Bills and their payments double-count each other
Invoices inflate a month that hasn't been paid
The tax owed surfaces the day it's due — after the money's spent

What we built

One file a month, in whatever shape it comes
It reports what came in — and what didn’t
Every line reconciled to what actually moved through the bank
Tax separated out before anything else
The report carries the presenter's brand; ours stays invisible

Where the AI acts

Reads the file in whatever shape it arrives
Classifies every line against that account's own history
Reconciles bills against the payments that settled them
The consultant reviews the exceptions and presents the result

The result

The monthly close stops being a rebuild
Double-counted expenses surface instead of hiding
The report exports under the presenter’s own brand

Wellness & bodywork clinic · UK

A monthly export of hundreds of lines that mixed each supplier's bill with its payment, doubling the expense, with no view of real cash.

17 months of history consolidated automatically. The bill-vs-payment double-count found and removed.Verified

Wellness, mentoring & events hub · Brazil

A control spreadsheet that quietly disagreed with the bank, the owner's own capital mixed into revenue, and the monthly tax simply guessed.

Statement and spreadsheet reconciled to the cent. The owner learned the operation runs at a loss and only breaks even on her own capital — a truth the spreadsheet hid.Verified

Advisory firm serving SMEs · UK

Every client report rebuilt by hand in Excel — no brand, no standard, and no way to grow the book without growing the team.

A handmade report per client became a standardized production line under the firm's own brand. The whole book is run by two people.Verified

Industrial-equipment manufacturer · Brazil

A managerial P&L living in a giant spreadsheet, with no clear budget-vs-actual and no answer to the board's one question: does the operation stand on its own?

13 cost centers rolled up into one view, with contribution margin and budget-vs-actual updating each month.Verified

Before · per client, per month

4–8 hrs of manual close

Reading a ledger or statement line by line, hunting double-counts, rebuilding the same spreadsheet.

After

~20 minutes of review

The report arrives built. The consultant checks the exceptions and presents it.

Where the hours actually go

VERIFIED 17 months of history consolidated automatically · bill-vs-payment double-count found and removed · statement and spreadsheet reconciled to the cent · 13 cost centers rolled up into one view · a full book of clients run by 2 people · 6 segments live across 2 countries.

PROJECTED baseline: monthly close previously done by hand in Excel, 4–8 hrs per client depending on ledger size. The after figure assumes the account has run one full cycle, so the classification rules are already learned.

The client's file is read in whatever shape it arrives — nobody has to reformat anything.

Every line is reconciled to what actually moved through the bank, so bills and their payments stop double-counting.

Each correction the consultant makes becomes a rule — next month comes back pre-sorted, and the work shrinks each cycle.

The report exports under the presenter’s own brand, so nothing is rebuilt for the client meeting.

Its own core · its own database · its own rules

What it knows

Every supplier this account has ever classified, held in that account's own database — one client's history never informs another's.

What it enforces

Cash-basis rules and the tax reserve for the right jurisdiction, since the same platform serves two countries with different tax law.

What it handles at once

Hundreds of ledger lines per file, across every account, every month, all at once.

Where else this fitsAccountants, bookkeepers and outsourced accounting firms that want reporting under their own brand · clinics with many practitioners · owner-operators who put money in and take money out · any sizable operation with cost centers.

Every account on this platform requested confidentiality. They authorized us to present the process and the numbers as an example — nothing that identifies the client, the brand or their revenue.

Recurring field service · cleaning & property turnover

Live

Crew photos and texts that become the schedule and the invoice

An operation coordinating cleaning crews across short-let properties, where the whole business ran through a group text.

The mess

The coordinator's entire day was the phone.

Job details, check-ins and proof photos all arrive as texts
Some in English, some in Spanish
Booking platforms disagree with each other about dates
Someone rekeys all of it into a schedule, then into invoices
A missed message meant a missed clean or a wrong bill

What we built

Each message and photo becomes a scheduled job and a crew assignment
Conflicting dates across platforms reconciled in one place
The proof photo the cleaner sends becomes the evidence the client gets

Where the AI acts

Reads the request and pulls out address and time
Checks the crew's photo against what was asked for
Drafts the schedule and the client update
A dispatcher approves in one tap — nothing reaches a client unseen

The result

Crew notes and photos become the schedule and the invoice on their own
Fewer missed jobs and fewer wrong bills
The office runs on the days the key person is out
It scales to more crews without adding office staff

Before · two partners, combined

60–90 hrs a week

Two people living inside the group text, seven days a week.

After

~15 minutes a week

They open the dashboard and answer only what the system couldn't settle.

Where the hours actually go — this is not magic, it's four bottlenecks removed

What's left for a human is the queue of things the system couldn't settle: the crew that didn't check in, the photo that doesn't match, the client who changed the date. That queue is the 15 minutes.

CLIENT-REPORTED the 60–90 hrs is the partners' own account of their week before the build — coordination, chasing confirmations, assembling reports and invoicing.

PROJECTED the 15 minutes is the steady state at full deployment, once every booking channel is connected and the crews are checking in through the system. It is only what the system could not settle, not zero involvement — the partners still run the business, they just stop transcribing it.

Calendars stop disagreeing. Every booking platform syncs into one schedule, and conflicting dates are reconciled automatically instead of being caught by eye.

Jobs assign themselves. Each clean is routed to a crew by location and availability — no messages sent, no one asked to confirm.

Proof arrives as proof. The cleaner checks in on site and photographs the finished job. The system reads the photo, checks it against what was requested, and files it.

The report and the invoice write themselves. The client receives the finished-job report without anyone typing it, and billing follows from the same record.

Its own core · its own database · its own rules

What it knows

Every property, every crew, and what “done” means at each site — including the client who always wants one thing done differently.

What it enforces

The checklist for that specific property, so a photo either matches what was promised or gets escalated.

What it handles at once

Every crew photo the moment it lands, all routes at once, without a dispatcher watching the thread.

Where else this fitsLandscaping · pest control · pool and aquarium service · HVAC · facility maintenance · any recurring work with a field crew.

This client requested confidentiality. They authorized us to present the process and the numbers as an example — nothing that identifies the company, its owners or its market.

Agriculture · sustainability certification

In development

Scattered field audits, assembled into one compliance file

A producer group working to restore a sustainability certification after the paperwork, not the farming, fell behind.

The mess

Field audits, supplier spreadsheets and land-use data
Technical opinions arriving as voice notes and loose documents
Farms with poor connectivity
Evidence for each requirement sat in a different place
Nobody could say what was missing until an auditor said it

What we built

Reads the spreadsheets and documents and computes the indices
Attaches evidence to each requirement
One dossier, with the gaps flagged before the audit — not during it
Works out in the field, where there's no signal

Where the AI acts

Extracts data from documents never designed to be machine-read
Assembles the file against the audit's requirements
Every judgment call belongs to a certified specialist
The system prepares the evidence — it doesn't grade it

The result

The gaps are known before the auditor arrives, not after
Evidence is attached to the requirement it answers, permanently
Field data is captured on the farm, offline, and syncs later

Before · per audit cycle

3–4 weeks of a specialist

Assembling evidence across farms, then reworking whatever the auditor found missing.

After

~3 days of review

The dossier is already assembled. The specialist judges the gaps it flagged.

Where the weeks actually go

The system prepares the evidence. It does not grade it — every judgment belongs to a certified specialist, and the scoring method stays with its owner.

PROJECTED baseline: preparing a certification dossier by hand across multiple properties, plus a rework cycle after the auditor’s findings. The after figure assumes field capture is in use through a full season. The largest saving is not the three weeks — it is not losing the certification cycle to a gap nobody saw in time.

Evidence is captured on the farm, offline, at the moment of the visit — not written up from memory a week later.

Each document is attached to the requirement it answers, permanently, instead of living in a folder someone has to remember.

Indices are computed from the source spreadsheets rather than re-entered.

Gaps surface continuously, so they're closed before the auditor arrives instead of during the audit.

Its own core · its own database · its own rules

What it knows

Each property, each requirement, and exactly which evidence is already on file against it.

What it enforces

Which document satisfies which requirement — and refuses to mark a requirement met without it.

What it handles at once

A whole season of documents from multiple properties, assembled continuously instead of in an audit-week scramble.

Where else this fitsFood supply chains · ESG and sustainability reporting · any regulated operation where evidence must be assembled for an outside auditor.

This client requested confidentiality. They authorized us to present the process and the numbers as an example — nothing that identifies the group, its properties or its certifier. The scoring methodology belongs to a partner and is not described here.

Professional services · business consulting

Live · pending partner sign-off

Every engagement used to start from a blank page

A consultancy serving small businesses, where the diagnostic was rebuilt from scratch for every new client.

The mess

Each engagement's data in its own spreadsheets and folders
Two weeks of slides and analysis before the first real conversation
The quality depended on which consultant happened to run it

What we built

Takes the intake — financials, interviews, scores
Assembles a standardized, branded assessment
Hands the consultant a first draft to react to instead of a blank page

Where the AI acts

Reads the intake and organizes it against the firm's own framework
Drafts the deliverable
The consultant edits and signs — the thinking, and the reputation riding on it, stay theirs

The result

A first draft in hours instead of days
The same rigor on every engagement, whoever runs it
More engagements handled by the same people

Before · per engagement

~2 weeks to first draft

Slides and spreadsheets built from scratch before the first real conversation.

After

~2 days

The consultant opens a draft to react to instead of a blank page.

Where the weeks actually go

CLIENT-REPORTED the two-week baseline is the firm's own account of its pre-engagement work.

PROJECTED the after figure assumes the firm's framework is encoded and the intake is collected through the system. The second benefit is harder to put in hours: the same rigor on every engagement regardless of which consultant runs it.

The intake — financials, interviews, scores — is structured on arrival rather than transcribed later.

It's organized against the firm's own framework, so the analysis starts from the method instead of rebuilding it.

The deliverable drafts in the firm's format and branding, every time, whoever is running the engagement.

The consultant edits and signs — the thinking, and the reputation attached to it, stay theirs.

Its own core · its own database · its own rules

What it knows

The firm’s framework and the shape of every engagement it has run before.

What it enforces

The same structure and rigor on every deliverable, whichever consultant is running it.

What it handles at once

Every open engagement at the same time, so no client waits behind another.

Where else this fitsAccounting firms · fractional CFOs · legal · recruiting · franchise support · any practice with a repeatable method and a blank page problem.

This client requested confidentiality. They authorized us to present the process and the numbers as an example — nothing that identifies the firm or its partners. Shown subject to their sign-off.

08 How to start

Start small. Then grow.

We don't try to fix everything in the first contract. We solve the one thing that already pays for itself, ship it in weeks, and grow from there.

STEP 1 · 30 MINUTES

Mapping call

No slides. You walk us through one job, from first contact to money in the bank. We find the spot where information comes in sideways.

STEP 2 · 3–6 WEEKS

Your first core

Built and deployed on the tools you already use. Your part is showing us how the work really flows. That's it — we adapt to it.

STEP 3 · ONGOING

Expand

Each new piece reuses what we already own, so the second win costs less and lands faster than the first.

Book a 30-minute mapping call Email us instead

Engagements start with a one-time setup fee, then a monthly fee.
Priced after the mapping call, never before.