# Introducing Synthia: an AI analyst for CPG founders

Published: 2026-11-10
Category: Product Updates

Synthia is an AI analyst for CPG founders. She reads this week's online shelf before she answers, and she beat a general AI chat on 41 of 43 founder questions. Here is why I built her and what an answer looks like.

CPG founders have enough to focus on, like perfecting the product, finding a co-packer, making content and landing their first retail buyer, so I built Synthia, an AI analyst for CPG founders that beats the standard chatbots used widely today. She's designed just for you, with data and tools not available anywhere else... Ask her what to charge, which pack to make or who you're really up against, and she checks historical prices & promos, social media buzz and competitors for nuanced and precise answers that beat the AI everyone uses today.

If you recently started a CPG company as a side hustle or have been running one for a while, Synthia is for you. If you are considering building one, it's easier than ever to get the site up and running with AI, but that's the easy part so why not onboard an agentic cofounder and analyst to save you a ton of time getting started, and managing your brand.

Synthia is invite only, request an invite today.

What the first year looks like

Picture the first year of a CPG brand. Before you even build a site, you're deciding what to sell, and that means knowing who your customer is and who your competitors are, because they already have the tools set up.

Then you're deciding what to charge, which depends on who you're selling to, who you're up against, and what they sell for. Then packaging, which has to stand out visually and grab attention on social media, and which sets your brand vibe: fun, playful, serious, or minimal. You're making each of these calls with very little to go on.

Around those sit the things that are hard but aren't what Synthia is for: the legal side, and whether you produce it at home, find a co-packer or find a manufacturer. They take a lot of time and effort if you do them on your own.

Then you start selling and the job changes. It becomes about your social media, your pricing, your competitors' pricing, what they're posting, and what creators are talking about.

These are all things a general AI agent is not fine-tuned for.

Where general AI agents fall short

They can do web research but it's surface level. Unless your agent has access to verified and historical data, or unless it has access to the content and insights across hundreds of thousands of relevant social media posts that have already been extracted with insights pulled out and cleaned, it's really just guessing. It gives you websites and citations that it's found but there's nothing that tethers those responses to anything objectively true when the data it needs is not connected. Synthia connects all sorts of data through the ontology we've built and the agent we fine-tuned so that when you ask a question about a product, anything that's relevant across pricing, strategy, legal, social media, online, and physical stores is all resolved into one clean and precise answer.

When a quarter of CPG products fail after the first year, it's even more important to get those first few months right, have the right data, and have the right tools that everyone else doesn't have and that can help you avoid mistakes.

Why I built Synthia

All of these questions are things that I used to answer for customers working in CPG analytics as a category manager and in advanced analytics across brands: from small startups to some of the largest brands in the world. I help them build research on their market, on competitors, pricing, promotions, and distribution, but the difference is I had access to a lot of data. I had a team of analysts working alongside me. CPG founders don't have this. You start with a major disadvantage. The odds are stacked against you because you don't have the datasets, because the datasets are very expensive, often tens of thousands of dollars a year, and you might be a solo founder or small team.

What we've done with Synthia is not only build an agent that has the expertise of a team and arm it with the data, but we've also built an agent made for founders, not analysts. The largest data companies in the world still rely on clunky dashboards and CSV dumps, something that I swear will never burden founders with.

What an answer looks like

Getting started with Synthia is just like any other chat app you've used. You ask questions in plain English and she answers, cites sources, shows data, creates charts, does research, accesses tools, finds retail buyer contacts and distributors, and drafts emails for your approval. For example if you were starting a coffee business and wanted to know whether a $25, 12 oz bag of whole bean or ground coffee is priced right, this is what the online shelf looked like on November 6, 2026:

The shelf (thousands of DTC stores online retailers) holds +1,647 whole bean and ground coffee listings, counting each bag size on its own. The middle price is $25.99, and nine in ten sell for between $12.99 and $123.43. The range is that wide because the shelf mixes small bags with 5 lb bulk bags and wholesale cases.

Compared like for like, the 321 single 12 oz bags have a middle price of $20.75, and nine in ten sell for between $16.00 and $27.46. A $25 bag sits above three in four of them.

Across the listings that record a size, the middle price is $1.67 an ounce and the top quarter starts at $2.00. A $25, 12 oz bag works out to $2.08 an ounce, so it sits in that top quarter.

More than one listing in four has no bag size on record, so check bag sizes before you trust a price match.

Same bag, same price, three different readings. Against the whole shelf, $25 is right around the middle. Against the 12 oz bags it's actually competing with, it sits above three in four. Per ounce, it's in the top quarter. If you stopped at the first number you'd think you were priced like everyone else. Which group you compare against changes the answer, so she shows you all three and tells you which listings are missing a size.

Battle tested against a general AI chat

I didn't want to just say she's better, so I tested it. Both got the same 43 questions founders ask about price, packaging, wholesale and price history to Synthia and to a general AI chat. A judge that couldn't see which answer was which picked the one a founder would rather act on, then did it again with the order swapped. The judge was an AI model. Synthia's answer was picked 41 times, two were a tie, and the chat's was never picked.

Here is what that looks like in her answers:

She makes the call. Asked what to quote a convenience chain for a 2.5 oz bag of jerky, she opens with "Open at $4.59 a bag, and don't go below about $4.00." She tells you the 42.5% retailer margin behind it is her own assumption, then asks for your real cost.

She compares like for like. A $34.99 box of 12 protein bars is $2.92 a bar, about 9% above the typical bar. Not box against box.

She compares you with the right products. For a jar of salsa she checks retail shelves for jars your size and tier, so you're measured against what a buyer would measure you against.

Every figure traces back to the data, your files, your question, or an assumption she tells you about. See the benchmark.

Congrats, you sold out, now what?

You sold out of your first batch in the first couple of weeks and proved that there is demand for your product. So now what? It's time to re-order, and you want to upgrade the packaging. You you check the social media and hold two mockups up to your phone, a pouch in one hand and a tin in the other. "Help me pick," you say. "Comments are open." By morning a few hundred people have replied, and the pouch is ahead. The tooling deposit goes out that afternoon.

That's a real decision with real money behind it, and asking your audience is a fine way to settle it. But look at where the choice came from. Why a pouch and a tin, and not single-serve sticks or pods? Those two options came from your own head, a shelf you happened to walk past and a competitor's post. Nobody laid out the whole field: which formats are on the shelf, what each one sells for per ounce, how many brands sell it, which are on promo this week, and which look thin. That step comes before the poll, and it's the step I built Synthia for. She turns "pouch or tin?" into a short list of options worth putting to your audience, each one with its numbers.

We saw some version of this scene over and over in the hundreds of founder and small-brand accounts we read: which flavor ships first, which cities to open in, whether the label works. Real decisions, put to an audience of a few hundred because nothing else was within reach.

How we got the data

It took a few steps to get here. First we built Crush, an app that rewards people for sharing their receipts and helps us collect pricing information in physical stores, which will soon be available in Synthia. Then we built e-commerce datasets, so we could see the same products across many channels, and opened them up to AI agents through Syntalic. I also opened a pay-per-query API for agents who can pay just cents for pieces of data through the X402 and machine payment protocol by Stripe. So, if you're comfortable using AI and your terminal and you want to test out the data, definitely give that a go on the developers page, but if you want a smooth user interface with dashboards, rich data, and an agent similar to the app you're used to, then heySynthia is the tool that lets you put all of it to work.

Why now

Small food brands are having a moment, and the money is following. PepsiCo closed its purchase of poppi for $1.95 billion in May 2025 and agreed to buy Siete Foods for $1.2 billion in October 2024. More people are also trying to get in. The Bureau of Labor Statistics counts 43,277 food manufacturing establishments in the first quarter of 2026, up from 34,350 in 2019, and Whole Foods' emerging-brand accelerator drew more than 2,000 applications in 2026, up from more than 1,600 for the 2024 cohort. Drinks look different: Census applications coded to beverage and tobacco manufacturing fell from 2,910 in 2019 to 1,950 in 2025. Nobody counts brands, so we won't claim there are more of them, only that more people are making food and applying for the same shelf space. AI may be part of the reason. One Bloomberg columnist recently wrote that AI could bring back the autonomy and flexibility that once defined work, and a business of your own is the clearest version of that.

More small businesses are turning to AI chats. A 2025 survey of 3,870 US small businesses found 58% using generative AI, up from 40% in 2024, though the Census Bureau's official count is lower, 17% to 20% of firms as of May 2026.

Getting access today

This is a soft launch, and requesting an invite is free. We'll email your invite when your place opens, with a live read of your category. An invite starts you on a 30-day Team trial, no card required. The data tools are already open to developers through Syntalic, paid per call.

Next on the roadmap: Crush members running missions and tasks in physical stores for you, so the question you can't answer from a screen has a person on the other end.

If you're setting a price or heading into a buyer meeting soon, request an invite, or email hey@heysynthia.ai with your category and the price you're considering.

Talk soon,

Azad William Neenan

Founder

More posts: https://heysynthia.ai/blog