The AI that reads retail like a merchant.

It's Monday. The market already moved over the weekend.

A buyer opens her laptop. Competitors re-priced overnight, a heatwave just rewrote demand for half her assortment, and a vendor email is asking for a markdown decision by noon. The old answer was a morning lost to pivot tables. The new answer is a short, ranked briefing already waiting in her inbox - written by Profitmind, the company that decided retail's hardest decisions shouldn't require a spreadsheet marathon.

Profitmind is a decision-intelligence company. It builds agentic AI - software that does not just chart the past but recommends the next move - for the people who price, plan, and stock the things you buy. It is small, roughly sixteen people, and it is already steering merchandising calls for retailers that range from twenty-million-dollar specialty shops to hundred-billion-dollar giants. That is an unusual spread. Most software picks a lane.

Retail has more data than it can read, and less time than it needs.

Here is the uncomfortable truth the industry rarely prints on its glossy decks: most merchandising decisions are still made by exhausted humans squinting at exports. Pricing, assortment, inventory, promotions, marketing - each lives in its own report, its own tab, its own meeting. By the time a team finishes the analysis, the market has politely moved on without them.

The cruel part is that the answers usually exist. They are buried in the data. They are just not prioritized, not explained, and not delivered before the decision is due. Profitmind's founders looked at that gap and saw less a technology problem than a tempo problem. Retail did not need another dashboard. It needed something that would read everything and then say, plainly, do this next.

Field note: A dashboard tells you the building is on fire. An agent hands you the extinguisher and points at the exit.

A retailer who learned AI, an AI legend who learned retail.

Dr. Mark Chrystal spent more than twenty-five years inside retail - American Eagle Outfitters, rue21, David's Bridal, The Disney Store - doing, by hand, the very analysis his company now automates. He is not an outsider with a clever model. He is the customer, which is a rarer and more dangerous thing to compete against.

The other half of the bet is Andrew Ng, who needs little introduction in AI circles - Coursera, Google Brain, Landing AI - and who chairs Profitmind. The company was incubated inside his AI Fund, which means it was built where deep machine learning meets actual operating experience. Barney Govan, the co-founder and CTO, makes the architecture run. The wager was simple: pair someone who knows exactly what retailers need with people who know exactly what modern AI can do, and refuse to ship a generic tool dressed up for retail.

Caption: Two resumes that should not fit on the same business card, stapled together anyway.

Not one model. A staff of agents.

Profitmind's platform is built as a crew of specialized AI agents, each with a job a retail org would recognize. One loads and cleans the data. One holds the strategy. One watches competitors. Others own pricing, inventory, promotions, and assortment. They report to a conversational analyst you can simply ask a question - and to the agent everyone seems to love most, the one that produces the Monday-morning briefing.

The design choice that matters is the handoff. A single giant model that tries to do everything tends to be confident and vague. A team of narrow agents, each accountable for one slice of the business, can be checked, corrected, and trusted - and crucially, each can explain why it is recommending what it recommends. For a merchant about to bet a markdown budget on a machine's advice, that explainability is not a nicety. It is the entire reason she will click yes. Profitmind leans on a modern AI stack - retrieval, vector search, orchestration frameworks, and frontier models - but keeps the output stubbornly human: a short list, ranked, with reasons attached.

Why it amuses us: The flagship feature is named after the worst part of everyone's week. That is either marketing genius or a cry for help. Possibly both.

The numbers do the arguing.

Skeptics are right to ask whether any of this works. The company's own figures are bold but specific, which is the kind worth checking against your own data: across customers, Profitmind reports an average of 21% revenue growth and 14% higher gross margins, one client improvement north of 250 basis points of profit, and hundreds of hours of manual analysis erased each month. The most telling stat is quieter - 85% of retailers who run a demo become customers. Software rarely converts like that unless it solves something real.

Then there is the validation that money signals. Accenture - not a firm that invests in retail tech casually - took an equity stake and led the $9M Series A, joined by Thorndale Farm, Magarac Venture Partners, Andrew Ng's AI Fund, and Lightscape Partners. The named and associated customers read like a tour of real retail: Dunham's Sports, Kirkland's, Bare Necessities, Batteries Plus, Busy Beaver, Leonisa - apparel, footwear, auto parts, home goods, sporting goods.

An intelligence layer for every retailer.

Profitmind's stated mission is plain: empower fast, profitable decisions by surfacing quantified, prioritized opportunities in real time. The longer vision is more ambitious. The company believes every retailer will eventually run an AI intelligence layer inside its stack - a layer that sees the data, the market, and the business goals at once, and helps teams decide and execute faster than the competition.

There is a democratic streak to it. The platform is built so a first-week associate and a thirty-year merchant can both ask it a question in plain language and get a usable answer. Advanced analysis stops being the privilege of whoever is best at spreadsheets. That is a quietly radical idea in an industry where institutional knowledge often walks out the door at retirement.

The market won't slow down. The decisions can speed up.

Retail's tempo is only accelerating - more channels, more competitors re-pricing in real time, more weather and demand whiplash. Human teams cannot out-read that volume, and they should not have to. The bet Profitmind is making is that the winning retailers of the next decade will be the ones who decide fastest with the most confidence, and that the deciding will increasingly be done with an agent at the table rather than a spreadsheet on the screen.

So return to Monday. The buyer opens her laptop. The competitors still moved overnight, the heatwave still rewrote demand, the vendor still wants an answer by noon. The difference is the briefing already waiting - ranked, explained, profitable - so the morning that used to vanish into pivot tables is hers again. That is the whole point. Profitmind did not promise to replace the merchant. It promised to hand her the answer before the spreadsheet finished loading. So far, 85% of the people who watch it do exactly that decide to keep it.