Building in Fintech in 2025: Where the Real Opportunities Are
Most fintech trend pieces are written for investors, not builders. Everything is "transformative," every market is "$X billion by 2027," and the practical question — should I build here, and if so, what specifically — goes unanswered.
This is a different kind of piece. It's for founders and product teams trying to figure out where the real building opportunities are, what's genuinely moving, and what's mostly hype that will be embarrassing to have cited in a pitch deck two years from now.
Embedded finance: the infrastructure shift is done, the application layer isn't
The interesting thing about embedded finance in 2025 isn't the concept — it's that the infrastructure matured faster than the use cases caught up.
Stripe, Plaid, Synapse, Unit, and a handful of BaaS providers have made it genuinely possible to add financial services to a non-financial product without becoming a bank. The plumbing exists. What's still underdeveloped is the application layer — products where the financial service is actually native to an existing user relationship rather than bolted on as a revenue line.
The distinction matters. Shopify Capital works because Shopify already has access to merchant revenue data — the loan decision is made from information Shopify already owns, and repayment comes automatically from the merchant's Shopify revenue. The financial product is native to the relationship.
Compare that to a loyalty app adding a debit card because embedded finance is a trend. The financial product has nothing to do with why users are in that app. It's monetization dressed up as product.
According to McKinsey's Global Payments Report, embedded finance revenues are expected to reach $183 billion by 2027. That growth will disproportionately accrue to products where the financial service genuinely belongs inside the user relationship, not products where it was grafted on.
The builder's question: what does your existing user base already trust you with, and is there a financial product that fits inside that trust?
AI in financial services: where it's actually changing things
There are parts of financial services where AI is genuinely rewriting how things work, and parts where "AI-powered" is a positioning claim applied to a decision tree.
Real shifts:
Alternative credit underwriting — Using cash flow data, transaction history, and behavioral signals alongside or instead of traditional credit scores. This is opening credit access to people and businesses that traditional models reject, particularly small businesses and gig workers. The Consumer Financial Protection Bureau has published guidance on AI in credit decisions, which is worth reading if you're building here — the regulatory environment is evolving fast.
Real-time fraud detection — ML-based transaction scoring that identifies novel fraud patterns faster than rules-based systems can be updated. This is now table stakes for payment processors, not a differentiator.
Document processing — Automating the extraction and verification of financial documents (tax returns, bank statements, pay stubs) for lending and compliance workflows. Significantly reduces manual underwriting time.
Still mostly marketing:
Robo-advisors that apply the same model portfolio to everyone, with an AI label. Chatbots that answer FAQ questions about account balances. "Personalized" recommendations that are just rule-based upsell triggers relabeled.
The test: does the AI change the actual decision or outcome, or does it just automate a process that was previously done manually? The former is a real advantage. The latter is an efficiency gain, not a product differentiator.
Super apps: the Western market question
WeChat Pay and Grab proved the model. The consolidated financial super app — one platform for banking, payments, investing, insurance, and credit — is mature in Southeast Asia and deeply embedded in daily life.
It hasn't happened at scale in Western markets. The structural reasons are real: regulatory fragmentation across US states and EU member countries, stronger incumbent banks with existing customer relationships, higher consumer trust in established financial institutions, and different cultural attitudes toward data sharing with tech companies.
What has worked in Western markets: deep vertical super apps. Not "all financial services for everyone" but "all financial services for this specific type of person."
| Approach | Example | What makes it work |
|---|---|---|
| Gig worker financial stack | Moves, Lance | Understands irregular income, instant payout needs |
| SMB all-in-one | Mercury, Relay | Designed around startup and small business workflows |
| Immigrant remittance | Remitly, Wise | Built for a specific cross-border financial need |
| Creator economy banking | Karat | Underwriting based on creator revenue, not salary |
The pattern: a specific user segment with a financial life that existing banks systematically fail to serve well. The super app serves that segment comprehensively rather than serving everyone superficially.
DeFi: past the 2021 hype, into the infrastructure phase
The 2021 DeFi cycle produced thousands of protocols, most of which don't have meaningful real-world usage. The projects that have survived and matter are doing something structurally useful: stablecoin infrastructure, cross-border settlement, tokenization of real-world assets.
Institutional adoption is accelerating. BIS research on DeFi and the emergence of regulated DeFi frameworks in the EU (under MiCA) and tentatively in the US suggest the next phase is compliance-compatible DeFi — products that work within regulatory structures rather than around them.
For builders: the regulatory environment in your target jurisdiction is the most important variable right now, more important than the technology. The products that will scale are the ones that engage with regulators early rather than treating compliance as an obstacle.
The filter worth applying to all of it
The fintech companies building durable businesses in 2025 have a specific answer to one question: why does this financial product belong inside our product, for our specific users?
The ones that don't have that answer are building features. Features get copied or displaced by the next infrastructure shift. Products built on a genuine user relationship compound over time.
Fintech product decisions — whether to build a financial feature, which infrastructure to use, how to handle compliance early — are the kind of thing that's worth getting right before you've committed to an architecture.
We work with founders on exactly this kind of early-stage product and go-to-market thinking.
agency.pizza →
