Long before Danial Jameel was selling artificial-intelligence agents to banks, he was losing sleep in a basement library. It was the summer of 2010 at the University of Toronto. Jameel and fellow students Alice Dinu and James Dang had settled into a cramped residence study room, working on a mobile app through days that frequently ended after four hours of sleep. They stayed until the building finally reclaimed the room from them. “We basically hijacked our residence library in the basement at UC until they kicked us out,” he later recalled.
The setting carries a little startup mythology, but the product began with a plain observation. University life had become a scavenger hunt. A schedule lived in one system, a club announcement in another, a campus event somewhere else. Students had smartphones, yet the institution around them still arrived as a pile of portals, calendars and notices. Jameel had seen the problem close up while working in student life, helping coordinate resources for roughly 15,000 incoming students.
OOHLALA pulled those fragments into a campus app. Its first version combined events, messages, calendars and practical student information. The founders entered the idea in a Rotman School business competition, won, and took the result as permission to keep going. By 2011 the app had launched. Within several years it was operating across campuses in multiple countries. In 2015, Forbes put Jameel and the founding team on its 30 Under 30 list for education.
This is the useful starting point for understanding Jameel's current company, Saris AI. His businesses appear to occupy different worlds. One served freshmen trying to find an event. The other serves loan officers and compliance teams working inside regulated institutions. Underneath, both begin with the same irritation: important work broken across systems that refuse to behave like one system.
The first map was a campus
OOHLALA did not remain the social campus app imagined in the basement. Customers pulled it toward a larger job. Universities needed branded software that could surface services, alerts, contacts and course information without asking students to understand the institution's organizational chart. Jameel described student life as “very complex and fragmented.” The company's work was to hide that complexity without pretending it did not exist.
“Pretty much your whole life revolves around school.”Danial Jameel, on the four-year campus experience
The team joined Y Combinator's Summer 2016 batch. In early 2017, OOHLALA announced a Series A while serving more than 200 institutions in eight countries. Its customers included large universities, but Jameel's rhetoric stayed with the individual at the other end of the software. Did a student find the service? Did the product change engagement or retention? “One of the biggest challenges in education is efficacy,” he said at the time. “Does your product actually tie into student outcomes?”
In 2018, OOHLALA and DubLabs combined under the Ready Education name. The platform expanded, and Jameel eventually exited. He also moved into institutional service, joining Tiffin University's Board of Trustees in 2021 and participating in the leadership council of the Brookings Institution's Center for Universal Education. The student founder had become a repeat operator and education advocate. Yet the old co-founding relationships survived the transition.
A recurring product instinct
The quiet machinery of money
Saris brought Jameel, Dinu and Dang back together. More than a decade after commandeering the library, the trio chose financial services as the new maze. Banks and credit unions move mortgages, small-business loans and agricultural financing through their communities, but much of the work behind those products is administrative and stubbornly manual. Documents must be collected. Data must be checked. Exceptions must be routed. Information must travel among loan-origination systems, core banking platforms, repositories and email.
A single lending workflow can exceed 50 steps. Some systems lack modern application interfaces. Others can display information but cannot complete a transaction elsewhere. The person in the middle becomes the integration layer: copying a value, checking an attachment, updating a record, then moving to the next screen. The work is repetitive without being simple, a particularly awkward fit for the previous generation of automation.
Saris calls its software agentic AI. The relevant promise is practical: an agent can move through a multi-turn workflow, use the institution's existing systems and write the result back into the system of record. Lending, compliance, collections and operations provide the initial territory. Human employees retain judgment and approval in regulated decisions. The agent absorbs the chasing, checking and re-keying around them.
Company-reported figures describe results across customer workflows; individual deployments vary.
The distinction matters because enterprise AI is crowded with impressive fragments. Extracting a number from a document can be useful. Drafting a reply can be useful. A bank, however, experiences the work as a completed loan file or a resolved exception. Jameel describes the economic change from traditional software in equally direct terms: “People often ask me what changed from SaaS to AI. For me, it's delivering on outcomes vs seats.”
That line contains a commercial challenge. Seat-based software gets paid when another employee receives access. Outcome-based AI has to prove that something finished faster, cheaper or more accurately. It also has to fit the controls of a regulated institution. Audit trails, human review, security and traceability are not decoration around the agent. They are part of the product.
The difference between 9:30 and 6:15
Jameel's most revealing Saris story is not a model benchmark. It is a clock. He visited a financial institution where a lending team was staying until about 9:30 each evening to clear its operational backlog. The staff was not slow, he said. They were doing document review, validation and reconciliation on top of the work that already filled the day. After Saris was deployed, he later said, the team was leaving around 6:15.
Efficiency percentages belong in procurement documents. An evening returned to a family is easier to picture. Jameel calls this “the dignity of work,” a phrase that prevents the automation discussion from floating away into abstraction. Saris is not promising that software should make people irrelevant. Its stated ambition is to move their time toward judgment, relationships and service.
“For us, this is not just about efficiency. It's about the dignity of work.”Danial Jameel, discussing community financial institutions
The choice of customers strengthens that frame. Credit unions and community banks rarely dominate technology headlines, but they finance homes, farms and local businesses. Their balance sheets may be smaller than a global bank's; their operational burden is not proportionally quaint. Jameel has described them as part of the heart of the economy. Building for them makes integration and implementation discipline more important, because few have spare teams waiting to supervise a science project.
A founder lesson worth stealing: translate operational ROI into a human unit. “Hours saved” becomes more vivid when the hour has a name - dinner, a child's bedtime, or a member conversation that no longer gets rushed.
Old collaborators, harder rooms
Repeat founders are often described as if their main advantage were a better investor list. Jameel's more interesting asset is a working relationship. He, Dinu and Dang have built together for more than a decade. Saris started with people who already knew how each other handled thin resources, unhappy customers and late product changes. Shared history cannot solve a 50-step lending process, but it can remove some of the uncertainty inside the team trying to solve it.
Jameel remembers week-long onsite sessions with early Saris customers and the late-night iterations that followed. That closeness has a strategic benefit. Financial workflows differ among institutions, but regulation standardizes much of their shape. Saris says roughly three quarters of a workflow definition can overlap from one institution to another. The patient work of mapping the first process can therefore become reusable infrastructure for the next.
Investors noticed. In May 2026, Saris announced a $28.8 million Series A led by 8VC, with participation from Audacious Ventures, Homebrew, BankTech Ventures and Service Credit Union. The company said customers could reach production in about 90 days. Soon after, Saris became a recommended partner of the Cooperative Credit Union Association, whose network covers nearly 200 institutions in the Northeast.
Funding is a marker, not the conclusion. Saris still has to show that its agents remain dependable as documents change, edge cases accumulate and examiners ask where an output came from. It has to turn bespoke observation into a repeatable product without smoothing away the details that make each institution distinct. The company is entering precisely the kind of market where trust is earned one finished workflow at a time.
Jameel's career offers a compact playbook for that work. Begin with the overlooked user. Sit beside the process long enough to see what the org chart hides. Connect what already exists before asking everyone to migrate. Then measure whether the person's day actually improved. In the basement library, the person was a student trying to make sense of campus. At Saris, it is an employee trying to get through a lending queue before the evening disappears.
The distance from a campus calendar to a bank's core system is enormous. The founder's question has barely changed: why should the person doing important work also have to carry the complexity of every disconnected tool beneath it?