A rice-shop owner in Battambang needs working capital. The traditional route runs three days: a branch visit, a stack of forms, a loan officer, a wait. Boost Capital's route runs about ten minutes, and it happens inside a conversation she is already having on Facebook Messenger. No app to download. No branch to find. She answers a few questions, snaps photos of her documents, does a short video call, taps to sign, and the money moves.
That gap - three days versus ten minutes - is the entire business. Boost Capital, a Singapore-headquartered fintech founded in 2018, decided the friction in financial onboarding was not a customer problem but a software problem. Fix the software, and a whole population that banks had written off as too small, too remote, or too much paperwork suddenly becomes reachable. Two million applicants and merchants later, that thesis has held up.
01 / What it isOnboarding, rebuilt inside the chat thread
Boost does not describe itself as a lender anymore, and that is the interesting part. Its product is a white-labelled onboarding platform that financial institutions plug into their existing systems. When a bank wants to sign up a customer, Boost's technology runs the whole gauntlet - identity checks, business verification, document capture, anti-money-laundering screening, credit-consent, contract generation, e-signature - through the chat apps people already use: Messenger, Telegram, WhatsApp. The bank keeps its core banking system. Boost brings the customer to the door and walks them through it.
Agentic AI to deepen your customer profiles and broaden your customer base. Boost Capital, company positioning, 2026
Under the hood sits a stack of modules: e-KYC for identity parsing and liveness, e-KYB for business verification, document validation the company puts at more than 99% OCR accuracy, e-contracting, AML checks, a data dashboard, and omnichannel deployment via SDK, API, or hosted portals. The newer framing is "agentic AI" - software that builds a customer profile progressively instead of demanding everything up front. The older framing was simpler: a chatbot that acts like a very patient loan officer.
02 / The customersThe unbanked don't have apps. They have chat.
Boost's original users were micro and small business owners in Cambodia - the kind of entrepreneurs who run on cash, keep records in their heads, and have never held a formal credit line. The founding observation was almost embarrassingly obvious: this population does not download banking apps, but it lives on chat platforms all day. So the product met them there.
In the Cambodia pilot alone, Boost's platform chatted with more than 200,000 customers and registered roughly 96,000 MSMEs. About 68,000 of them - more than two thirds - were led by women, a cohort traditional lenders routinely underserve. Backed in part by the UN Capital Development Fund, the platform disbursed US$1.7 million in loans in a single year and put 21,000 entrepreneurs through digital and financial-literacy training on the way.
Today the customer has shifted. The people filling in the chat flows are still small business owners, but the people paying Boost are the institutions serving them - banks, microfinance lenders, and fintechs across Southeast Asia and India.
03 / The problemWhy a loan cost 25% more than it needed to
The problem Boost attacks is not credit risk. It is the cost of acquiring and verifying a customer. Traditional microlending is expensive because it is physical - branches, field officers, paper, travel. Those costs get passed to the borrower and baked into the interest rate. Boost's claim is that when you move onboarding into a chat window, the process runs roughly 100 times faster, costs borrowers about 25% less, and - crucially - keeps the same repayment rates as in-person lending.
That last number - matching repayment rates - is the one that unlocked everything. It let Boost walk into a risk-averse bank and say the digital channel is not a downgrade in credit quality. It is the same borrower, reached more cheaply.
04 / The turnFrom lender to the software lenders buy
Here is the move worth studying. Boost set out to lend money to Cambodian entrepreneurs. To do it well, the team had to build onboarding tooling that worked over chat, parsed documents, ran KYC, and generated contracts. Somewhere along the way, the tooling turned out to be more valuable than the loan book. So Boost stopped trying to be everyone's lender and started being everyone's onboarding layer.
It is a familiar pattern - the product a company invents to survive its own hardest problem becomes the thing it sells - but Boost executed the version that plays well with incumbents. The pitch to a bank is disarmingly small: keep your core system, we will bring you customers. No rip-and-replace. Integrations with core-banking platforms like Musoni and Instafin mean a microfinance institution can bolt on chat-based onboarding without tearing anything out. RAFI Microfinance did exactly that in 2024.
05 / The peopleAn investor who went and built the thing
Boost was founded by three co-founders: Gordon Peters, chief executive, who came out of Mekong Strategic Partners and years of corporate finance across the Greater Mekong region; Lucinda Revell, co-founder, later named a Changemaker by the Mastercard Center for Inclusive Growth; and Sergey Lesninsky, who leads technology. It is a telling mix - a regional financier, an inclusion-focused operator, and an engineer - for a company whose whole trick is making finance, impact, and software agree with each other.
UNCDF commended Boost Capital's success in working with multiple financial institutions to disburse US$1.7 million in loans to MSME owners in a single year. UN Capital Development Fund
06 / The moneyA seed round, and the receipts to raise the next one
Boost raised a US$2.5 million seed round announced in 2023, with a roster that reads like an emerging-markets who's-who: Village Global, Iterative Ventures, Hustle Fund, Epic Angels, Xcel Next, Insitor, and Orbit Startups. The company has since reported a Series A milestone. Along the way it collected the kind of validation that helps a B2B fintech get into bank procurement queues - winning Women's World Banking's 2021 Fintech Innovation Challenge and landing on the Center for Financial Inclusion's list of top inclusive fintechs.
07 / The marketWhere it sits, and who it sits against
In the broad map of fintech, Boost lives in the onboarding-and-identity layer - the same neighborhood as global verification vendors like Jumio, Onfido, Sumsub, and the regional player ADVANCE.AI. What sets Boost apart is less a single feature than a posture. Most verification tools assume the customer arrives at a bank's app or website. Boost assumes the opposite - that the highest-value customers in emerging markets never show up there at all, and have to be met inside a chat thread, in their language, with no download. That focus on the underbanked-via-chat corner is the moat: it is a harder distribution problem than most identity vendors want to solve, and Boost has 2 million applicants' worth of practice at it.
The business model follows from that positioning. Boost is B2B SaaS - it licenses the platform to institutions and earns on usage tied to onboarding volume, rather than carrying loans on its own balance sheet. It grows by adding banks, not branches, which is why a company of roughly 54 people can operate across four countries.
Underbanked usually means under-served by your process - not uncreditworthy. Boost's whole business is the difference between the two. YesPress reading of the model
The unresolved question is the one facing every "agentic AI" pitch in 2026: how much of the profile-building genuinely runs on autonomous models versus well-orchestrated automation, and whether that distinction matters to a bank that just wants more approved customers at lower cost. Boost's answer, so far, has been to let the numbers do the arguing - same repayment rates, lower cost, more reach - and let the label sort itself out.