Synthesis is ontology software and an AI harness for consumer brands, sold as one plan. It connects your stores, ads, email, books and files, writes the business down once so every number means one thing, and checks each number against its source. You then ask questions inside Claude, ChatGPT or another AI app. It costs $999 a month for the whole team.
Meet Priya. She runs a made-up hand soap brand.
It is Monday, 8:10 a.m. Her store had its best month ever.
Amazon is up 9%. Meta says the ads paid back.
At 9:30 her accountant calls. “You made less than you did in March.”
Priya opens four tabs. Amazon, Shopify, Meta, the books.
Each one is right. None of them talks to the others.
So the one question that matters, “what did we actually make, and why?”, has no single place to go.
You have probably had a Monday like that.
Synthesis is ontology software and an AI harness for consumer brands, in one plan. It reads every system your brand runs as one brain, checks each number against its source, and answers inside the AI you already use.
That is the short version. The rest of this page covers what it does, who it is for, who it is not for, the five ways people use it, and what it costs. We also say plainly where it stops.
What Synthesis is, in plain words
The tag line on our homepage is “Your brand, with a brain.” The mission behind it is to put every owner in complete control of their business.
Two pieces of software do that work, and they come together in one plan.
- Ontology software. An ontology is a written-down model of a business. It lists the things your brand is made of, such as products, listings, orders, customers, campaigns and costs. It says how they connect. And it gives every number one definition. Stanford’s well-known ontology guide describes one as formal definitions of a field’s terms and how they relate (Stanford). Our post on AI-ready data explains why that matters.
- An AI harness. A harness is the layer around an AI model. It decides what the AI can reach, what it is told with each question, and what gets checked before you see the answer. Our post on what an AI harness is covers the five parts.
The ontology holds the meaning. The harness keeps the AI on it. You need both before you can trust an answer about money.
What it does, step by step
Here is what happens from the first call, as the homepage lays it out.
- We learn your brand. On the first call, one link reads your own website. It picks up who you are, how you sound, your colours and your typeface.
- We connect every source. Your stores, ad accounts, email and SMS, books and files are connected with you on that call. There are 49 integrations, including Amazon, Shopify, Meta, Google, TikTok Shop, Walmart, NetSuite and Klaviyo. The first full import runs that night.
- It builds your Brand Brain. It reads everything together and checks every number against its source. Amazon sales tie to settlements, the report of what Amazon actually paid you. Ad spend ties to each ad account.
- You start the next morning. You ask questions in your AI, or ask it to build something.
- It keeps going. It warns you first, writes the changes worth making, and makes them when you say yes.
Your account is live in about two hours. The full import finishes overnight.
The five ways in
Everything runs on the same brain, built from your own data. You can use one way in, or all five.
| Way in | What you get | Where it shows up |
|---|---|---|
| Brand Brain | A living map of every product, channel, customer and rival, and how each moves the others. Rebuilt every week. | A living map of your brand |
| Ask anything | Answers from your real numbers, with the sources | Claude, ChatGPT or any app that speaks MCP |
| The plugin | Pages, ads, decks and dashboards in your brand’s own look and voice | Claude Code |
| Monday reports | What moved, why, and what to do this week | Your inbox, with a live dashboard behind it |
| Automations | It warns you, writes the change, makes it when you say yes, and checks it landed | Meta, Amazon or Shopify |
Two things are worth saying clearly.
First, Synthesis has no chat window of its own. You ask inside Claude, ChatGPT or another AI app that speaks MCP. MCP is the open standard that lets AI apps like Claude or ChatGPT connect to outside systems (modelcontextprotocol.io). The reasoning stays in a tool you already own.
Both big AI apps support it. As of October 2026, Claude lets users on its Free, Pro, Max, Team and Enterprise plans add custom MCP connectors (Claude Help Center). OpenAI’s developer docs explain how a remote MCP server brings new data into ChatGPT (OpenAI docs).
Second, nothing changes in your accounts without a yes. The AI can read freely. Changes wait for a person. That matches the MCP rule book, which says users must clearly agree to each data access and action (MCP specification).
If you want the idea behind the map, see what a brand brain is.
Here is where it gets interesting.
What a finding looks like
Here is the kind of thing it is built to surface. Illustrative numbers, made-up brand.
Say a travel mug brand sells on Amazon and its own store. In September, the 2-pack sold 22% more units than in August. Every channel report looked good.
Then the brain lines up the pieces. Amazon’s fees per unit rose. A new ad set pushed the 2-pack at a higher cost per order. And a discount code ran on the store all month.
The finding: the 2-pack sold 22% more and made about $1,900 less. The extra sales went to fees, ads and the discount.
No one missed this. Amazon cannot see what the store discount cost. The ad account cannot see the Amazon fee change. The data hid it until the pieces sat in one place.
How it differs from a dashboard
A dashboard is a set of charts someone built in advance. It answers the questions its builder thought of.
That works until you ask a new one. “Which product family made money after ads last quarter?” If no chart was built for it, someone exports data and does the math by hand.
Synthesis starts from a model of the business, not from charts. So a new question is answered on the same definitions as the last one. You can still get a dashboard. You ask for one, and the plugin builds it from your live numbers.
Now for the shortcut most owners already try.
How it differs from pasting exports into a chat
Many owners already paste a spreadsheet into Claude or ChatGPT. It feels like it works. The answer comes back fast and sounds sure.
The risk is not that the AI makes things up. The risk is that it gets the wrong inputs and answers them confidently. A sales export does not say whether it counts orders placed or money paid out. It does not know your product cost. It does not know September is not over yet.
So two chats, a week apart, can give two different “revenue” figures for the same month. Both sound right. Our post on why AI gets your ecommerce numbers wrong walks through ten of these traps.
| Dashboard | AI chat on exports | Synthesis | |
|---|---|---|---|
| A question nobody planned | Needs a new chart | Answered, on whatever was pasted | Answered, on shared definitions |
| Same answer next week | Yes, for built charts | Often not | Yes |
| Joins Amazon, store, ads and books | Only if someone built it | Left to the AI each time | Yes, before you ask |
| Numbers checked against the source | Rarely | No | Yes |
| Labels a month still in progress | Depends | Rarely | Yes, with basis, period and caveats |
| Makes changes | No | No | Yes, after your yes |
Who it is for
Synthesis is built for consumer brands that sell in more than one place. A typical fit sells on Amazon and its own store, runs paid ads, and keeps its costs in a set of books.
It also fits agencies that run several brands. Each client sees only its own brand, and the agency can work across its roster.
The common thread is that the important questions cross systems. “Is our Meta spend paying back?” needs ad data and sales data. “What does each product really make?” needs sales, fees, ads and product cost.
So who should skip it? More brands than you might think.
Who it is not for
It is the wrong tool for some brands, and it is better to say so.
- One channel, simple questions. If you sell only on your own store and mostly ask about that store, the platform’s own AI connector may be all you need. Our guide to MCP servers compares the options.
- Very little history. The brain reads what your business has done. A brand a few weeks old has little for it to read yet.
- Changes with no approval. Synthesis waits for a person before it changes anything. If you want an AI to act on its own, this is not it.
- A tight budget for software. $999 a month is real money for a small brand. Free connectors exist, and they are a fine place to start.
It also answers what has been modeled. A rare question about something nobody wrote down may need new work. And it does not replace your AI subscription. You still pay for Claude or ChatGPT.
What it costs
The price is public and there is one plan.
| Item | Price |
|---|---|
| Synthesis plan, whole team, up to three brands | $999 a month |
| First month | Free |
| Seat fees | None |
| Annual contract | None. Month to month, cancel any time |
| Paid yearly | $899 a month |
| Omni-channel weekly brief | Free on every plan |
| Specialist weekly reports | Optional, $100 a month each, per brand |
| Meta to Amazon add-on | Optional, $1,000 a month per brand |
Every connector and unlimited questions come in the plan. There is no cheaper tier with the checks switched off.
A quick way to judge it
Whatever tool you look at, run the same test.
- Write down three numbers you already know, such as last month’s net revenue by channel.
- Ask each one in a fresh chat.
- Check that each answer names its basis and its period without being asked.
- Ask the same question a week later and compare.
- Ask it to change something, and see whether it waits for your yes.
The short version
Synthesis is one model of your brand, checked against its sources, with the AI you already use kept on it. You can reach it five ways: a map, questions in your AI, a plugin, a Monday report, and changes you approve.
It costs $999 a month for the whole team, and the first month is free. It is a poor fit for a one-channel brand with simple questions, and a good fit when the questions that matter cross systems.
Questions people ask
What is Synthesis Intelligence?
Synthesis is software for consumer brands. It connects every system a brand runs, from Amazon and Shopify to ads and the books. It writes the business down once so each number means one thing. Then it keeps the AI you already use on those numbers.
How much does Synthesis cost?
$999 a month for your brand and your whole team, with no seat fees. The first month is free and there is no annual contract. Some specialist weekly reports and one add-on cost extra, and none of them is required.
Does Synthesis have its own chat app?
No. You ask questions inside Claude, ChatGPT or another AI app that speaks MCP, the standard way AI apps connect to other software. Your AI subscription stays with that provider.
Is Synthesis a dashboard?
No. A dashboard shows the charts someone built in advance. Synthesis holds one written-down model of the business, so you can ask a question nobody planned for and get an answer on the same definitions as every other answer.
Can Synthesis make changes in my ad accounts or store?
Yes, with your approval. It warns you first, writes the change, makes it in Meta, Amazon or Shopify when you say yes, and checks that the change landed. Nothing changes without a yes.
How long does Synthesis take to set up?
About two hours from the first call to the first answer. Sources are connected with you on that call, and the first full import runs that night.
Who is Synthesis not a good fit for?
A brand on one channel with simple questions may be fine with that platform’s own AI connector. A brand with very little sales history has less for it to read. And anyone who wants changes made with no approval step should look elsewhere.