The three engineers who started Nosto in Helsinki in 2013 had a small, stubborn complaint. A good physical shop remembers you - the assistant who knows your size, points you at the thing you almost bought last time, notices when you are lost. Online, that memory vanished the moment you closed the tab. Juha Valvanne, Jani Luostarinen and Antti Pöyhönen decided to rebuild it in software, and the company they founded has spent more than a decade turning that one complaint into a platform.
Today Nosto describes itself as a commerce experience platform. In plain terms, it is the layer that decides what an individual shopper sees when a product page, a search box or a category grid loads - and it does that decision-making for roughly 2,500 retail and consumer brands. The company reports having delivered on the order of 25 billion personalized shopping experiences. Most shoppers have never heard the name, which is rather the point of good infrastructure.
01 / What it doesPersonalization, unbundled from the guesswork
Nosto started where most personalization companies do: product recommendations. Its recommendation engine now ships with more than 20 self-learning algorithms - 1:1 personalized suggestions, cross-sells, visually similar items, and replenishment nudges for things a shopper is due to run out of. The idea is not a single clever trick but a set of small, constant adjustments that add up to a store that behaves as if it is paying attention.
Over time the product widened. Nosto added AI-powered onsite search that tries to read intent rather than match keywords, category-page merchandising governed by rules a merchant can actually control, post-purchase upsells, dynamic bundles and personalized email. Underneath sits what the company calls experience.AI - a data engine that unifies customer, product and content signals so the recommendation, the search result and the pop-up are all reading from the same picture of the shopper.
02 / Who uses itMid-market brands without a data-science team
Nosto's sweet spot is the mid-market: brands large enough to care about conversion rate and average order value, but without an in-house machine-learning group to build personalization themselves. Its published customers skew toward fashion, beauty, homeware and specialty retail - names like Dermalogica, Perry Ellis, Robert Dyas, Industry West and outdoor brand PELAGIC. The platform is installed through apps and APIs on the commerce systems those brands already run: Shopify, BigCommerce, Adobe Commerce and Shopware among them.
A good physical shop remembers you. Online, that memory vanished the moment you closed the tab. Nosto's entire business is rebuilding it.
The case studies read in the language retailers respond to - percentage points of conversion and order value rather than model architecture. That framing is deliberate. For a merchandiser choosing between vendors, the promise is not smarter AI in the abstract; it is a search box and a set of recommendations that a small team can turn on and measure quickly.
03 / The strategyBuy the missing pieces, sell one platform
What separates Nosto from a pile of point tools is a roll-up thesis. Rather than build every capability from scratch, it acquired them. Search and category merchandising came in part through SearchNode and Findologic. In 2021 it bought Stackla, a visual user-generated-content platform, which also gave Nosto a foothold in Australia and Southeast Asia and turned social content into a merchandising surface. Each acquisition was folded into a single stack rather than sold as a separate product.
Product Recommendations
20+ self-learning algorithms: 1:1 suggestions, cross-sells, visually similar and replenishment.
Search & Merchandising
Intent-aware site search and rule-based category merchandising, added via SearchNode and Findologic.
Visual UGC (Stackla)
Sourcing, rights-managing and merchandising shoppable social content inside the store.
Huginn
Agentic AI on Nosto's Large Intent Model, built to automate merchandising and campaign work.
The counter-argument is that bundling everything can mean being best at nothing. In each category Nosto faces a specialist - Algolia in search, Bloomreach and Mastercard-owned Dynamic Yield in personalization, a shifting field of merchandising tools; in late 2025 several of them, Searchspring and Klevu among them, merged into a single competitor. Nosto's answer is integration: for a mid-market team, one vendor that covers search, recs, merchandising and content is worth more than four best-of-breed tools that never share a shopper profile, each with its own contract and its own idea of who the shopper is.
That distinction matters more than it used to. Third-party cookies are being wound down across browsers, which erodes the tracking that older personalization tools leaned on. A platform built around first-party behavioral signals - what a shopper does inside the store, in real time - is on firmer ground. Nosto has spent years collecting exactly those signals, which is one reason its recommendations and search sit on the same shopper profile rather than a rented one.
04 / The numbersWhat the AI search bet is producing
The clearest recent signal is search. Nosto reported that queries on sites running its AI search grew 323% in 2024, and that search-driven revenue for brands rose 1,024% between 2023 and 2024. Those are the company's own figures and worth treating as directional rather than audited, but they explain where the roadmap is pointed: the search box, long the most neglected part of an online store, is becoming the place where intent is easiest to read and monetize.
05 / The businessSubscription software, sold by the module
Nosto is a B2B SaaS company. Retailers pay a subscription for platform access, generally scaled by traffic, catalog size and the modules they switch on, with add-ons for search, merchandising, UGC and content personalization. It sells directly and through a network of implementation agencies and technology partners. Third-party estimates put annual revenue in the low tens of millions of dollars; the company has raised across several rounds - including a reported 16 million dollar raise in early 2023 - from investors including OpenOcean, Sanoma Ventures, Tesi, the European Investment Bank and Kreos Capital.
The team - about 140 or so people the company fondly calls "Nostonians," reported larger on some professional networks - works remote-hybrid across offices in Helsinki, London, Paris, New York, Stockholm, Salzburg, Berlin and Sydney. The engineering stack leans on Scala, TypeScript, Elasticsearch and stream processing with Apache Flink and Spark on AWS, which is roughly what you would expect from a company whose product is real-time decisions at page-load speed. The performance constraint is real: personalization that slows a page down defeats itself, so much of the hard work is invisible plumbing rather than the features a merchant sees.
That engineering depth is also the company's moat. Anyone can bolt a recommendation widget onto a store; the difficult part is doing it for thousands of catalogs at once, reacting to a shopper's last three clicks before the next page paints, without dragging down load times. Recognition in Gartner's 2025 Magic Quadrant coverage points to the same thing - Nosto has moved from a single-feature startup to a platform enterprises are willing to shortlist.
06 / What's nextAn AI raven named Huginn
In 2025 Nosto launched Huginn, an agentic AI for commerce built on what the company calls its Large Intent Model. The name is a nod to Norse mythology - Huginn is one of Odin's two ravens, dispatched to fly the world and report back what it sees. The ambition is the same: a team of agents that reduce the manual merchandising and campaign work that fills a retail team's week, executing inside the Nosto platform and, the company says, alongside third-party LLM applications.
The next fight in ecommerce is not who ranks the results. It is who runs the store.
That framing marks the shift Nosto is making. Personalization was about deciding what the shopper sees. Agentic commerce is about deciding what the merchant does - which product to promote, which rule to change, which campaign to run - and doing some of it automatically. It is early, and the claims are ahead of the evidence, as they are across the whole agentic-AI wave. But it fits a company that has spent twelve years trying to close the gap between how a store feels in person and how it behaves online.
The mission Nosto states has not really changed since three Finns wrote the first version of it: making every impression relevant. What has changed is how much of the store now falls inside that sentence.