CapEx vs Expense Capitalization Helper
Build an internal tool that runs every purchase against your capitalization policy - cost threshold, minimum useful life, repair-vs-improvement - and recommends capitalize or expense with a written rationale, so you apply the policy the same way every time and an accountant signs off before anything hits the asset register.
A login-protected helper: import purchases, have the tool apply your capitalization policy and recommend capitalize or expense with a plain-English rationale, flag borderline and aggregate items, then have an accountant approve each treatment before exporting the asset additions for your register, the expense items, and a complete decision log - with a hard rule that nothing is recorded without human sign-off.
Before you start
- A free Vercel account
- A free Supabase account
- A free Resend account (and a sender address you can use)
- A purchases CSV (description, amount, category, expected useful life)
- Your written capitalization policy (threshold, useful-life minimum, repair-vs-improvement rules)
The problem this kills
A box of purchases lands at month-end and someone has to decide, line by line: does this get capitalized (recorded as a fixed asset and depreciated over years) or expensed (written off now)? The $480 monitor - expense. The $5,200 server - capitalize. But what about the $900 office chair when your threshold is $1,000... unless you bought twelve of them on the same order? Is the new HVAC compressor a repair (expense) or an improvement that extends the asset's life (capitalize)? Does that "consulting" line actually belong in the cost of the equipment it configured?
In most teams these calls live in one experienced person's head. They're made slightly differently each month, rarely written down, and almost impossible to defend when an auditor asks "why did you expense this one but capitalize that one?" Borderline items get coded by gut feel. Bundled purchases that should be aggregated slip under the threshold individually. And every miscall either overstates this year's expenses or quietly inflates the balance sheet.
This tool turns your capitalization policy into something the whole team applies the same way - and produces a written reason for every single call, with a human approving before anything is recorded.
What you'll build
A small internal web app, just for your team, that:
- Imports a batch of purchases from a CSV (description, amount, category, expected useful life, purchase ID).
- Applies your capitalization policy - the cost threshold, the minimum useful life, and your repair-vs-improvement rules - to each line and recommends capitalize or expense.
- Writes a plain-English rationale for every recommendation ("$5,200 ≥ $2,500 threshold and 4-year life ≥ 1-year minimum → capitalize" or "compressor replacement restores existing function, not an upgrade → expense as repair").
- Flags borderline items (just over or just under the threshold) and aggregate items (many small units on one order that may need to be grouped) for a closer human look instead of guessing.
- Dedupes on purchase ID so the same purchase can't be processed twice.
- Routes every recommendation to an accountant who approves or overrides the treatment before it's recorded.
- Exports the asset additions (in the exact columns your fixed-asset register expects), the expensed items, and a complete decision log of every call and who approved it.
What's inside the Implementation Plan
The plan is a single markdown file you paste into Claude Code (a free AI coding agent). It walks the agent through building the whole tool, step by step, each step ending with a ready-to-paste prompt.
The most important part: the plan opens by interviewing you about your business. Before it writes a single line, the agent asks for your real capitalization threshold, your useful-life minimum, how you draw the repair-vs-improvement line, the exact columns and asset categories in your data, how you handle bundled and low-value-pooled purchases, and your messiest edge cases (trade-ins, freight and install costs, software, leasehold improvements). It reads a short tailored spec back to you, you confirm it, and only then does it build - so you get a tool that encodes your policy, not a generic textbook one.
Inside you'll find:
- The discovery interview and how the agent turns your policy into the rules engine.
- The full build: database, login, CSV import with duplicate guards, the policy engine that recommends and explains, the borderline/aggregate flagging, the review-and-approve screens, and the exports.
- The hard human approval gate so the AI only ever drafts a recommendation - a person makes the call that gets recorded.
- Verification steps so you can prove it works, and the CSV-export fallback so it's fully usable even before you connect it to your accounting system.
The governance it includes (this is the point)
This isn't a toy. The plan builds in the controls a finance team actually needs:
- Login so only your team can see or touch anything.
- Row-level security so people only see the purchases and decisions that belong to your organization.
- A complete audit trail - every import, recommendation, approval, override, and export is logged with who and when, and the policy reasoning is stored alongside each decision.
- A hard human-in-the-loop gate - the AI recommends and explains, but an accountant must approve each treatment; nothing is ever auto-recorded, and no new asset reaches the register without sign-off.
- Duplicate guards so the same purchase ID can't be processed or recorded twice.
Who it's for
Staff and senior accountants, controllers, fixed-asset accountants, and ops/BPM folks who have to apply a capitalization policy consistently across a pile of monthly spend - and who want a real, auditable helper without hiring a developer or buying heavyweight fixed-asset software. You don't need to write code. You need your purchases CSV, your written cap policy, and an afternoon.
You've got this - paste the first prompt and let the agent interview you.