AI-assisted monthly close

AI-assisted monthly close

Truewind. AI Agentic Sass (Tools: Cursor, Claude, Figma, V0)

Truewind. AI Agentic Sass (Tools: Cursor, Claude, Figma, V0)

Truewind is an automated accounting platform that pairs AI agents with human review to close the books faster. Today, closing the books means accounting teams manually reconciling transactions, chasing status across spreadsheets and tools, and repeating the same close checklist every month with no analytical value added. The challenge was building an AI system accountants would actually trust with financial data, in a domain where a wrong automated action has audit consequences.

Truewind is an automated accounting platform that pairs AI agents with human review to close the books faster. Today, closing the books means accounting teams manually reconciling transactions, chasing status across spreadsheets and tools, and repeating the same close checklist every month with no analytical value added. The challenge was building an AI system accountants would actually trust with financial data, in a domain where a wrong automated action has audit consequences.

My Role
Founding Product designer

Problem

Manual close process leaves accounting teams buried in work

Manual close workflows were disrupting accounting operations. Disconnected tools and a lack of real-time visibility created compliance risk, increased reporting cycle times, and inefficiency across every close. Accountants spent 40+ hours per close on manual process work, with no centralized view of what was done, what was pending, or what needed their attention.

Solution

An intelligent close where AI
proposes and humans decide

I structured the product around a Workflows → Tasks → Steps model so accountants always know where they are and what's left. Some areas I focused on:

  • Making AI thinking visible — showing what was matched and why, not just the result

  • Keeping people in final control — every automated action reviewed before it posts

  • Building trust incrementally — visibility first, then suggestions, then automation

  • Structuring every close the same way, so no two closes run differently again

Reducing manual work without sacrificing auditability

It was critical for accountants to move through hundreds of AI-proposed matches per close without missing the ones that mattered. The agent auto-matches the bulk of transactions and routes only the exceptions to a human-review step — so people spend their time on judgment, not repetition, and every action stays defensible at audit time.

A system built for trust,
not just consistency

I established the design patterns and principles for the platform, status, priority, review states, audit history, so an accountant can tell a workflow is reliable before they've read the detail. In a tool making financial decisions, consistency is what earns trust

Take Aways

Take Aways

Trust is a design problem

Accountants didn't distrust the AI because it was wrong often, they distrusted it because they couldn't tell when it might be wrong. Making the reasoning visible mattered more than making the AI more accurate. Investing early in transparency and human-control patterns was what let the automation ship at all.

Time to close books

Time to close book

50%

reduction in time to close books.

Behavioral Impact

35%

fewer manual reconciliation errors.

Next Project

Investify