Hummingbird, a San Francisco company that sells financial crime compliance software to banks and fintechs, has acquired Quantimon, a machine learning startup focused on anti-money-laundering (AML) risk detection. Hummingbird co-founder and chief executive Joe Robinson announced the deal in a company blog post on October 28, 2026.
Neither company disclosed the price or other terms. The Quantimon team is joining Hummingbird, where it will keep working on its technology as part of Hummingbird's transaction monitoring product, according to the announcement.
Quantimon's own website now carries a banner telling visitors that the company is joining Hummingbird. The companies did not say whether Quantimon will continue to sell its software on its own.
What Quantimon built
Quantimon makes machine learning models that help compliance teams spot money laundering. The models are trained on alert outcomes, meaning the decisions analysts make when they review a flagged transaction: whether to close it, escalate it or file a report on it. Pulse 2.0 reported that the models learn from each institution's own decisions.
The startup also built what it calls a federated learning network. Under that approach, models are trained inside each institution's own systems, and only privacy-protected model updates are shared across the network, so customer data does not leave the bank or fintech that holds it. Quantimon pitched this as a way to catch patterns of financial crime that a single institution would not see in its own data.
On its website, Quantimon claims its models use more than 150 features and cite a 19% reduction in false positives and 67% improved detection. The site does not explain how those figures were measured. It also says models are retrained regularly on analysts' decisions, with each version validated and tracked.
Hummingbird's post said Quantimon was founded by industry veterans from Apple, PayPal and Xoom. Ben Duranske, Quantimon's co-founder and chief executive, said the company was started after its founders saw compliance teams at those companies buried in alerts. The companies did not disclose when Quantimon was founded, how many people work there or whether it raised outside funding.
“At Apple, Xoom, and PayPal, we watched our compliance teams spend their days working on seemingly endless alert backlogs. We started Quantimon to fix that at the source. I’m thrilled Quantimon is joining Hummingbird, where our technology can be at the center of a loop that makes both detection and investigation smarter.”
Why Hummingbird is buying
Most AML transaction monitoring still runs on fixed rules, Robinson wrote. Rules fire on suspicious-looking activity, alerts pile up, and most turn out to be harmless. Machine learning has long been seen as a fix, but models need labeled examples of what was and was not suspicious, and that information has usually sat in free-text case notes spread across different systems.
Hummingbird argues that AI agents change that. Its platform already lets AI agents and human analysts work alerts in the same place, and it records their decisions in a structured form. Those records can now serve as training data for Quantimon's models, which would then decide which alerts deserve attention first. Robinson said the loop only works when detection and investigation run on the same platform.
"AI has changed how compliance teams investigate financial crime, but the systems generating the alerts have stayed largely the same," Robinson said in a statement carried by Pulse 2.0. "Today our customers can clear alerts faster than ever, but they shouldn't have to spend that capacity on noise."
Duranske described the deal as a way to put Quantimon's technology at the center of that cycle. "At Apple, Xoom, and PayPal, we watched our compliance teams spend their days working on seemingly endless alert backlogs. We started Quantimon to fix that at the source," he said.
What changes for customers
Hummingbird laid out three areas of work for the combined team. The first is better rule-based detection, with rules that are easier to build, test and manage for larger monitoring programs. Robinson wrote that machine learning works best on top of strong rules rather than as a replacement for them.
The second is models trained on each customer's own alert outcomes. These would rank alerts by risk and flag activity that looks like a customer's confirmed cases even when no rule has been triggered. The third, over time, is drawing on signals from outside a single institution, including third-party intelligence and patterns seen across the wider Hummingbird network.
Hummingbird's platform covers customer screening, transaction monitoring, investigations and regulatory reporting, including automated filing of suspicious activity reports. Customers named in the announcement include Raymond James, Symetra, Crypto.com, DraftKings and Etsy. Hummingbird did not give a timeline for when the new features would be available.
Hummingbird's backers
Hummingbird was started by fintech veterans, about half from Square and half from Circle, TechCrunch reported in December 2021. Robinson previously spent about two years running risk and data science at Circle.
At that time the company raised a $30 million Series B led by Battery Ventures, with Flourish, Homebrew, FinVC and Plaid co-founder William Hockey also taking part. That followed an $8.2 million Series A the year before and brought its total funding to $41.8 million. The company had 35 employees then, according to TechCrunch.
The 2021 article listed Stripe, Coinbase, Evolve Bank & Trust and Etsy among Hummingbird's customers. The Quantimon announcement did not say whether the purchase was paid in cash, stock or a mix.
