Splitto

Splitto is a personal finance tool built as a hackathon project by one developer (David Yong). The author states that Splitto uses AI to extract structured financial data from receipts and payments, but emphasizes that the AI does not write money directly — instead, deterministic code validates and finalizes transactions after human review. The product is described as a way to remove tedious manual entry while maintaining user control over financial decisions.

The project appears to be an experimental prototype focused on personal finance organization using AI for data extraction and structured validation. It is not evidenced to have any revenue, customers, or traction beyond the author's own development effort.

The description states that Splitto:

• Turns receipt or payment images into strict, editable financial drafts
• Uses AI (GPT-5.6) to extract structured data from images via a JSON schema
• Requires human review and confirmation before transactions are finalized
• Does not write money directly; deterministic code validates integer MYR sen, discounts, tax treatment, rounding, and allocation
• Keeps four financial domains separate: Shared receipts, personal Money, Portfolio snapshots, and Things
• Prevents shared obligations from becoming personal spending or portfolio values from being mistaken for cash flow

The product is described as a "financial draft" system where AI proposes structured facts and uncertainty is made visible. Final commitment remains deterministic and reviewable.

Inference The tool appears to be an experimental prototype focused on personal finance data entry automation, not a full financial management platform.

The author states:

• Splitto addresses the problem of inconsistent spending recording across multiple sources (receipts, shared purchases, cards, e-wallets, investments, durable goods)
• The core question is: "what would personal finance look like if AI removed the tedious entry without taking control away from the user?"
• The product aims to provide "zero-entry, not zero-control" — AI proposes structured facts but users retain control
• It is positioned as a way to make financial correctness, authorization, reconciliation, and final commitment deterministic and reviewable

Inference This is a self-described personal finance tool that attempts to solve the problem of fragmented financial data entry by combining AI with deterministic validation. The positioning is not yet market-tested or validated.

The description states:

• Splitto was built because the author was not consistently recording their own spending
• It targets individuals who manage personal finances and deal with multiple sources of income and expense (receipts, shared purchases, cards, e-wallets, investments, durable goods)
• The tool is described as a way to organize financial data from various domains

Inference The target customer is likely an individual user managing personal finances across multiple platforms or payment methods. No evidence suggests a specific ICP beyond this general category.

The description states:

• There is no mention of pricing, subscriptions, or monetization
• The author says the next step involves a "short weekly review inbox for statements, screenshots, and eventually authorized financial APIs"
• Bank sync, automatic card/provider data, and personalized financial advice are mentioned as future work, not current offerings

Inference No business model or pricing information is evidenced. The project appears to be an experimental prototype with no commercialization strategy described.

The description states:

• Built with React, TypeScript, Next.js-compatible Vinext routing
• Uses Cloudflare Workers, D1, R2, Drizzle ORM
• Leverages OpenAI Responses API and GPT-5.6 for image processing
• AI output is constrained to a strict JSON schema
• Uses Codex as an engineering collaborator during development
• Includes typecheck, lint, build, tests, and responsive QA before changes are treated as complete

Inference The technical stack indicates a modern, serverless approach with strong emphasis on validation and quality control. However, this is a prototype built in a hackathon context.

The description states:

• The project was submitted to the OpenAI 2026 hackathon
• It is described as an experimental prototype built by one person (David Yong)
• No revenue, customers, or adoption data are provided
• The author notes that real receipts exposed unexpected edge cases and required iterative fixes

Inference There is no evidence of traction, revenue, or customer adoption. This is a single-developer hackathon project with no commercial history.

The description states:

• No mention of competitors or market positioning
• The author references the problem that most finance apps still expect repetitive manual entry
• The tool aims to address fragmentation in financial data sources (receipts, shared purchases, cards, e-wallets, investments, durable goods)

Inference There is no evidence of competitive landscape analysis. No competitors or market positioning beyond general personal finance tools are mentioned.

The description states:

• The tool does not write money directly; it only proposes drafts
• The author notes that real receipts exposed edge cases like rounding errors, tax inclusion, large photos hitting framework limits, and mobile touch targets
• The hardest product decision was refusing an easy shortcut: shared receipts do not automatically become personal expenses
• The project is described as a hackathon prototype with no commercial traction

Inference Key risks include lack of commercial viability, limited functionality beyond prototype stage, and potential technical edge-case handling issues. No evidence of market validation or scalable business model.

1. What specific financial data sources does Splitto currently support?
2. How does the AI extraction process handle ambiguous or unclear receipt details?
3. Are there plans to integrate with existing financial institutions or APIs?
4. What is the expected user journey from receipt upload to final transaction confirmation?
5. Has the author considered how shared spending would scale beyond personal use cases?
6. What are the technical limitations of the current prototype that would need to be addressed for production use?