An AI chatbot that automated 80% of loan applications
A community-lending platform needed to process loan applications faster without adding reviewers. We built the chatbot that took the manual step out.
The problem
The platform runs peer-to-peer lending pools for members who have limited access to traditional financial services. Every loan request went through manual review in two languages, which capped how many applications the team could clear in a day and slowed the exact members the platform exists to serve.
What we built
- A bilingual (Japanese/English) NLP chatbot that answers loan inquiries, runs automated eligibility assessment, and moves applications through processing with minimal manual review.
- Automated risk assessment for the peer-to-peer lending pools, with flexible repayment scheduling built around real community lending patterns.
- Compliance checks and audit trails built into the transaction flow itself, to meet the host country’s financial regulatory requirements from day one.
Stack
Scale & result
The chatbot now handles the first pass on eligibility and paperwork for most applications, so the review team spends its time on the applications that actually need a human judgment call instead of re-keying the ones that do not.
More AI work we have shipped
Every claim above is a shipped, production system.
AI internal-linking engine
A semantic SEO engine built on OpenAI embeddings and vector search — the client engagement later became a co-founded partnership.
A film for every day’s news
MoviesEverywhere gets up to three film ideas a day for the news, each with a ready post. The agent learns from his votes. About $15 a month.
Tell us what you're building.
The problem, roughly what you think it needs, and where the data lives today. We will tell you what we would actually build — including when the answer is less than you asked for.