The call arrives from an unknown number. Before the phone lights up with a name or a warning, a system has already made a small judgment: familiar, legitimate, suspicious. Chaitanya Sarda spent more than four years at Google working near that instant. His teams handled phone-number verification, spam protection, and caller ID - the machinery that tries to decide whether an unfamiliar entity deserves trust.
AiPrise, the company he co-founded with Rushabh Shah in 2022, moves the same question into a more bureaucratic setting. A fintech wants to open an account for a company in Mexico. A marketplace needs to understand the owners behind a supplier in Europe. A payments business spots a document that does not quite agree with a registry. Somewhere between a customer clicking “submit” and a compliance officer clicking “approve” sits a thicket of databases, vendors, rules, and judgment calls.
Sarda's bet is that this thicket can behave like one product. AiPrise connects local and international data sources, lets customers set decision rules, holds the evidence in a case-management layer, and uses AI agents to collect and organize routine review work. The ambition is operational rather than theatrical: reduce the amount of stitching a company must do every time it enters another country.
A friendship before a company
The co-founder story begins with Sarda's wife. She had known Shah since college and introduced the two men. They became close during the pandemic, when they belonged to the same social bubble. Both were engineers at large technology companies - Sarda at Google, Shah at Meta - and both wanted to leave. Shah later described the shared desire plainly: they wanted a more meaningful problem and a more direct effect on people's lives.
A Y Combinator application deadline supplied the forcing function. They met over a weekend, built something rough, and applied. They got in. The idea on the application was not AiPrise as it exists today. Before the program began, they were weighing identity verification against content moderation. Customer research pushed them toward identity. Sarda had watched trust systems operate at Google and had worked alongside teams including Google Pay. Shah spoke with people working on global payments. The same complaint kept surfacing: international compliance was a headache assembled country by country.
“We always wanted to build something fundamental.”Chaitanya Sarda, 2022 founder interview
Their backgrounds made the product legible, but not the company. Sarda has said that YC gave two engineers a framework for everything surrounding the code: customer discovery, sales, hiring, and the basic mechanics of building a business. The distinction mattered. At Google and Oracle, he had improved systems with years of history. AiPrise forced him to confront the blank page.
The pleasure of zero to one
When asked about his proudest early moment, Sarda did not reach for a funding number. He chose “going from zero to one.” He remembered logging into AWS, learning how deployments worked, putting together a website and a product, then receiving live customer feedback. A request might arrive and a production update could follow an hour later. The first customer was proof that the new object had crossed from a founder's head into somebody else's workflow.
That early product promised one connection to many identity-verification vendors. The founders looked beyond North America and Europe, where buyers could at least choose among a few established providers, toward Latin America, Africa, and other emerging markets. In those regions, a company crossing borders might need a national identifier in one country, a different database in another, and several imperfect providers to cover the gaps.
The hardest geography became the better product brief. AiPrise could act as an orchestrator for businesses that did not want to rebuild the integrations themselves. In the first year, Sarda talked about wanting the largest set of emerging-market providers. The later company widened that claim: by its October 2025 financing announcement, AiPrise said it united more than 80 local registries and vendors, served more than 150 customers, and supported onboarding across more than 150 countries.
The unglamorous middle
Orchestration sounds tidy. Operations are not. As AiPrise grew from roughly 20 to more than 100 customers in a matter of months, its own billing became a small systems problem. Contracts mixed platform fees, usage charges, minimums, and discounts. Sarda was running a large Python script each month just to generate invoices. The company eventually moved the workflow into specialized billing software.
It is a revealing founder anecdote because it mirrors the product itself. Growth creates fragmentation. Someone holds the process together manually. The script works until the exceptions become the job. Then the company has to decide whether to keep patching or make the workflow explicit. AiPrise sells that transition to compliance teams while living versions of it in its own back office.
The useful abstraction often appears one step after the homemade script. First, learn the exceptions. Then design the system that can hold them.
The product has also moved beyond the one-API pitch. A compliance officer is not an API consumer. That person needs to reconstruct what happened, compare signals, ask for missing information, and show a regulator how a decision was made. Sarda described case management as the durable layer early on: a record that developers, analysts, risk leaders, and eventually auditors could understand.
Where the machine stops
Sarda's recent writing draws a boundary between two kinds of compliance work. Level-one review follows a playbook: request proof of address, inspect a business website, retrieve the prescribed document, compare known fields. He argues that AI agents can absorb much of that gathering. Level two starts when the playbook runs out. An ownership chain looks strange. Evidence conflicts. A legitimate business resembles a risky pattern. Judgment becomes the work.
In his framing, automation promotes the compliance officer toward those harder cases. The claim is credible only if the software makes its reasoning inspectable and leaves the analyst in control. AiPrise describes its agents as reading registry and identity documents, extracting fields, collecting information into a structured case, and summarizing risk factors. The human still makes the consequential call.
“Global companies shouldn't have to rebuild compliance infrastructure every time they expand into a new country.”Chaitanya Sarda, Series A announcement
That view also explains Sarda's advice for buyers evaluating compliance software. Ask whether the vendor offers a system or a point solution. Ask what its AI actually does, how it is tested, and what happens when it is wrong. Ask whether it can adapt to a jurisdiction and risk profile. Then push the demo into edge cases. A clean happy path says little about a product whose value appears when reality becomes untidy.
Conviction, with an audit trail
In October 2025, AiPrise announced a $12.5 million Series A led by Headline, with Y Combinator, SixThirty Ventures, Correlation Ventures, and strategic angels participating. The company said it had a 40-person team and customers including Bridge, a Stripe company, and dLocal. Sarda wrote that the money was not the main validation. The reactions from compliance teams mattered more: surprise that the product had been built so quickly, frustration with existing contracts, and urgency about integrating it.
He has a compact phrase for the years before that validation: “Conviction is uncomfortable before it's validated.” Early listeners told the founders that a platform spanning global business verification was impossible. The registries did not connect. Useful data did not exist everywhere. The easier response would have been to narrow the map. AiPrise instead continued expanding it.
The map now creates a second challenge: making a global company feel like one company. Sarda has written that a global team can be harder to build than a global product. Distributed work forces leaders to state what an office can leave implicit - how decisions get made, what good work looks like, and which expectations are shared. Trust, the product's subject, becomes an internal design problem too.
Sarda's career has followed a consistent pattern through different scales. Oracle gave him developer tools and automation. Google put unknown callers behind split-second decisions. AiPrise places a business, its owners, and a web of evidence inside a slower judgment with larger consequences. The craft is not merely collecting more signals. It is arranging them so a person can act, explain the action, and revisit it when the world changes.
That is the quiet promise inside the phrase “compliance platform.” A new country should not require a new pile. A routine review should not consume the attention meant for an ambiguous one. And a decision should arrive with enough memory to be understood later. Sarda is building for the moment after the unknown number appears - when a system has to show why it deserves to be trusted.