Fluid

01 — The car story

The fastest way to understand MIST.

Nobody went from a horse to a self-driving car in one step. Neither did commerce.

Stage 1

The horse.

Every task was manual. Reports by hand. Launches by hand. Slow — but familiar.

In commerce = spreadsheets & manual work

Stage 2

The engine.

Machines took the effort out. But parts from ten different factories don't make a car.

In commerce = disconnected software

Stage 3

The whole car.

Someone had to build every part — engine, brakes, wheels — engineered to work together.

In commerce = the Fluid platform

Stage 4

Self-driving.

Tesla recorded millions of real drivers. That's how a car learns to drive itself. Not magic — recorded experience.

In commerce = MIST

You can't bolt self-driving onto a horse. And you can't bolt real AI onto disconnected software.
We had to build the whole car. Now we get to teach it to drive.

02 — Chapter one

None of this existed.

Before Fluid, a commerce company was a pile of subscriptions.

A storefront from one vendor.

Checkout from another.

A back office from 1998.

An app that talks to none of it.

Analytics in a spreadsheet, three exports old.

None of it talks. Your team is the glue.

Every important action crossed three systems and two logins. That is not a software problem — that is an operating model problem. So we scoped the industry, protected what made it strong, and started with its most common weak point: the back office.

03 — Chapter two

We didn't want to build everything. We had to.

Every time the industry hit another hard limit, the generic answer failed. So we built the real infrastructure underneath it.

Fluid Connect. The back office, aligned.

Legacy back offices could not give every new experience clean, current business data. We had to standardize the plumbing for products, customers, reps, orders, ranks, and commissions — without fragile exports or scripts.

Rep App. The business in every rep's hand.

A separate mobile app would have created another data island. We had to build sharing, contacts, Smart Links, messaging, and Catchups on the same connected foundation — the next best action, wherever reps work.

E-commerce Website. Web experiences without the queue.

Shopify was built for ordinary retail, not MLM enrollment, subscriptions, rep context, and compensation-aware journeys. We did not want another website platform. We had to build one the industry could actually use.

FairShare. Credit where it is earned.

Generic ecommerce attribution stops at the last click. That is not fair to the rep who created the demand. We had to connect shares, engagement, follow-up, and orders so credit follows the person who earned it.

Commerce District. The complete business engine.

Generic commerce could not keep catalog, orders, subscriptions, checkout, offers, and rep context together. We had to build one connected district for the full transaction lifecycle.

Payments. Every transaction matters.

MLM merchants can be de-risked or shut down by a single processor with little warning. We had to build portable credentials, seven PSP relationships, underwriting, retries, routing, and local methods — a real bank beneath the marketplace.

Not band-aids. Underground plumbing. Full roads. Shared records. We built every layer because we had to.

04 — Chapter three

A pile of fixes would have failed.

So each new build used the same underground infrastructure: one customer, one order, one rep, one permission model. Every product reads from and writes to the same core.

A rep shares a page from the Rep App.

FairShare keeps their attribution attached.

Checkout converts the order.

Payments clears the transaction.

Connect updates the back office and commission engine.

Nobody exported anything. Nobody reconciled anything.

That is what "connected" actually means. Not an integration that syncs overnight. The same record, in the same place, seen by every product at once.

05 — Chapter four

Then we watched the best operators work.

For years, real people ran real companies on Fluid.

Opening new markets.

Launching products.

Recovering autoships before cutoff.

Building the dashboards the field kept asking for.

And we recorded those workflows. Every button. Every toggle. Every input.

Which settings a market launch touches. Which order the steps go in. What has to be checked before anything goes live. What good looks like when it's done. Thousands of real workflows, mapped — that's the driving data.

06 — Chapter five

Because we had to, now we get to.

AI can finally reason across a whole connected system. Because Fluid owns the city, the context, and the recorded routes, MIST can safely read across it and prepare writes back into it. The work we were forced to do became the advantage.

Step 1

You say it.

“Open a new market in India.” One sentence. Plain English. No manual.

Step 2

MIST carries it across the city.

The intelligent fiber brings shared context and recorded routes together — prepared, not published.

Step 3

You approve it.

Nothing ships without your yes. Not a page. Not a message. Not a market.

We had to build the city. Now MIST gets to operate it — with your approval.

07 — Now, you just ask

"Open a new market in India" is one sentence to you.

To MIST, it's a recorded workflow with hundreds of real inputs. Here's what actually happens — this is the part you're supposed to see.

› "Open a new market in India." Work order · prepared by MIST

Connect District — market and back-office syncProducts, customers, reps, currency, and commission-engine mappings preparedINR · products · reps · comp data

Droplets — market extension packApproved extensions, regional tools, and reusable launch capabilities installedinstalled · marketplace · workshop

Commerce District — India transaction engineCatalog, subscriptions, checkout, address formats, and offer logic configuredcatalog · enrollment · autoship · offers

Payments District — local clearingUPI, cards, COD rules, retry routing, and settlement preparedUPI · cards · COD · retries

Rep Experience District — field rolloutRep notifications, launch kit, and leader briefing ready to sendnotifications · launch kit · briefing

Fair Share — attribution rulesShare links and market-credit rules prepared so commissions land rightlinks · market credit · Order Journey

≈ 214 inputs · 9 workflows · 6 districts · one sentence from you Nothing ships until you approve

MIST doesn't guess at those inputs. It fills them from your context — your products, your prices, your comp plan, your markets — using the same workflows our best operators ran by hand for years. Every button. Every toggle. Every input. Recorded, mapped, and taught to MIST.

08 — And you approve everything

Three things MIST will never do.

✕Spend your money without asking.

✕Send anything you didn't approve.

✕Touch anything you've locked.

Guardrails aren't a feature we bolted on later. They're the reason executives let MIST near real work.

09 — The whole story, in three lines

This only works because of what's underneath.

Reason 1

We own the whole product stack.

MIST isn't guessing at someone else's software. Fluid built Commerce, Droplets, Rep Experience, Payments, Connect, and Fair Share as one system. One company. One city.

Reason 2

It learned from the best.

We recorded years of the best operators running real companies on Fluid — opening markets, launching products, saving autoships. That recorded experience is MIST's training.

Reason 3

You hold the keys.

MIST prepares. You approve. It can start read-only and private, and nothing important ships without your yes. Self-driving, with your hands near the wheel.

10 — Why this is hard to copy

Why others can't just add AI and catch up.

MIST is not a chatbot sitting above software. It is the final layer of an operating system Fluid spent years building, connecting, and teaching.

The homegrown ceiling

Internal tools solve a task. They don't create a platform.

A script can move an order. A spreadsheet can patch a report. An admin screen can fix one workflow. But each one encodes a local workaround — usually owned by one person, tied to one database, and invisible to the rest of the business.

Knowledge stays trapped in people and one-off code.
Every new market or product creates another exception.
There is no shared record, permission model, or reusable workflow layer.

Homegrown tools automate fragments. MIST operates the system.

The AI-wrapper ceiling

AI can't safely operate tools that were never designed to work together.

Put AI above Shopify, a separate mobile app, a legacy back office, payment processors, and spreadsheets, and it inherits every gap between them. Each tool has different data, permissions, APIs, and definitions of the customer, rep, order, and outcome.

It cannot know which system is authoritative.
It cannot preserve attribution and approvals across handoffs.
It can suggest steps, but it cannot reliably execute the entire workflow.

A smarter interface does not turn a fragmented stack into an operating system.

What had to exist first.

Intelligence became useful only after the infrastructure, context, and operating knowledge were in place.

01

Build the infrastructure.

Commerce, payments, attribution, field tools, extensibility, and the website — owned as one platform.

02

Connect the core.

One customer, order, rep, permission model, and shared business context.

03

Watch the best.

Real operators opened markets, launched products, recovered autoships, and handled exceptions.

04

Record the work.

Every decision, field, dependency, approval, and definition of done became a governed workflow.

05

Build MIST.

Only then could AI understand the context, prepare the work, and run it safely — with people approving the outcome.

Others are starting with the intelligence layer. The industry forced us to build what intelligence needs underneath.

Questions everyone asks

Ask us the hard ones.

Do we have to replace everything at once?+

No. Start with one command. A read-only one, if you like. MIST can observe first, then recommend, then draft. You grow it at your speed — no bet-the-farm moment.

What if MIST gets something wrong?+

Then you don't approve it. Wrong drafts get thrown away — not published. That's the whole point of prepare-then-approve: mistakes cost you a glance, not a cleanup.

Is our data safe?+

Your data stays in your Fluid account, under your permissions. MIST reads it to do work for you — it doesn't sell it, share it, or train public models on it.

Do our people need training?+

If they can ask a question, they can use MIST. That's the test we build against. The manual is the part MIST memorized so your team doesn't have to.

Start here

Now bring us a workflow of yours.

You've seen the city. You've seen what's behind a command. The next step is watching MIST run something you actually do every week.

Book a demo

This browser can't show the 3D map.

Welcome toCité Fluid.