The unglamorous middle of banking - the forms, the checks, the follow-ups - handed to AI agents that keep a human in the loop.
There is a lot of money in banking, and a surprising amount of it is spent on people typing the same information from one screen into another. Saris AI looked at that and decided it was a business.
Here is a thing about banks that fintech founders usually try very hard to change, and that Saris AI has decided to leave completely alone: the core system. The thing that actually holds the money and the ledgers and the loan records. It is old, it is regulated, it is load-bearing, and every startup that has promised to replace it has learned that banks would rather do almost anything than rip it out. So Saris does not ask them to. It works around the edges instead - in the space where a loan officer opens a document, reads a number, checks a box, opens another system, and does it again. That space turns out to be enormous.
Saris AI, founded in 2024 and headquartered in both Montreal and San Francisco, builds what it calls agentic workflow agents. The pitch, stripped of the word "agentic," is this: give the software a repetitive, multi-step back-office task - document intake, data extraction, a compliance check, a lending step - and it will do the whole sequence, not just one click of it. The agents are trained on a given institution's own workflows and systems, and, importantly, a human supervises the output. The company says this can automate up to 70% of the tasks in consumer, mortgage, and commercial lending operations.
"Our vision is a future where humans and AI work side by side in financial services." - Danial Jameel, Co-Founder & CEO
The people building it are not first-timers. CEO Danial Jameel, along with co-founders Alice Dinu and James Dang, previously built Oohlala Mobile, a Y Combinator-backed campus-life app that eventually merged with DubLabs and became Ready Education. Going from student mobile apps to bank back offices is a genuinely strange career move, and the honest read is that the founders learned a portable lesson in edtech: software wins when it fits into how people already work, not when it demands they work differently. Saris is that lesson pointed at a much larger, much more regulated market.
In May 2026, the company raised a $28.8 million Series A led by 8VC - the firm co-founded by Palantir's Joe Lonsdale - with Audacious Ventures, Homebrew, Btech Consortium, and Service Ventures joining. The money is going toward the least surprising things imaginable: more deployments, deeper integrations, and a bigger team of people who train and tune the agents. That last item is worth noting. Saris is not a self-serve product you download. It is a white-glove operation, which tells you something about how hard it actually is to make software behave inside a bank.
Figures as reported by the company. Enterprise AI metrics tend to describe best-case deployments - read them as claims Saris is confident enough to publish, not audited averages.
A ~9.8x drop. The pitch to a bank isn't "AI is cool" - it's this ratio, multiplied by volume.
Banks run millions of these micro-tasks. At scale, a move from two dollars to twenty-one cents stops being a rounding error and starts being a line on the income statement. Saris frames a top-10 bank example at roughly $27.8M in combined annual ROI, and a single team at about $535,000. The framing is deliberate: it is not selling intelligence, it is selling the same work for less money, with a human still signing off.
The catch, and Saris knows it, is that regulated work cannot simply be handed to a black box. So the product is built around supervision, auditability, and security - the parts that make a compliance officer comfortable rather than nervous.
Leads Saris and is the public face of its "humans and AI side by side" thesis. Previously co-founded Oohlala Mobile (Y Combinator), now Ready Education.
Part of the founding trio that reunited from the Oohlala Mobile days to build Saris's agentic automation for financial institutions.
Rounds out the repeat-founder team, carrying operating experience from edtech into the far more regulated world of bank operations.
The platform delivers real, measurable results without disrupting the systems and teams institutions depend on.
Automate lending operations. Saris agents run multi-step consumer, mortgage, and commercial lending tasks end to end - from document intake through the compliance steps - rather than automating a single click at a time.
Extract and move data. The tedious job of reading a document and re-entering its contents into another system is exactly the kind of high-volume work the agents are built to absorb.
Run compliance checks with a human in the loop. The agent does the work; a person supervises and approves. That structure is what makes the approach usable in a regulated setting.
Keep your existing stack. Saris integrates with the software banks already run - Fiserv, MeridianLink, Encompass, Hyland/OnBase - instead of asking them to replace it.
Scale without hiring. The core promise is more throughput without more headcount, with the company claiming roughly 2.5 hours freed per employee per day.
Deploy with help. No engineering required on the customer's side - Saris trains and tunes the agents to each institution's processes as a service.
Danial Jameel, Alice Dinu, and James Dang launch Saris to bring agentic workflow automation to banks and credit unions.
Saris builds and deploys agents trained on institutions' own workflows, targeting back-office tasks across lending, underwriting, and operations.
Announced May 28, 2026. Capital earmarked for more deployments, deeper integrations, and a larger agent-training team.
Saris AI has demoed publicly (including at FinovateSpring). A dedicated, permanent YouTube interview or product-demo URL was not verifiable at publication - the most reliable place to find current demos and talks is the company's own site and LinkedIn, linked below. We've chosen to point you there rather than guess a link.