AI Expense Parser
Photo, voice or text to structured expense in Sheets
Takes any expense input and turns it into clean structured data. Handles text, voice transcripts and receipt photos in one model.
A system that works while you are not looking
This parser removes manual expense entry. Send a Telegram message like lunch 12.50, forward a voice note saying spent 40 on fuel, or send a photo of a receipt, and it extracts amount, description and category and logs it correctly. It chooses the right monthly tab in Google Sheets automatically, so finance stays organized without naming sheets by hand. It is built for small teams and founders who want clean books without opening a spreadsheet every time.
Expenses logged in seconds from chat, in the right month, with consistent categories.
Outcome
Telegram messages are classified by type with a Switch. Text goes straight to the agent, voice notes are transcribed and forwarded as text, and photos are sent for vision description. A single AI agent with a structured output parser extracts amount, currency, description and category from the natural language. A Code node determines the correct month tab name from the current date, then a Sheets node appends the row. A Telegram confirmation summarizes what was logged and in which tab, so the user gets immediate feedback.
Trigger
Inbound signal starts the flow
Orchestrate
Models, tools and branches run
Deliver
Result logged and returned
Who it is for
Ops and finance teams that automate inbox, calendar and sheets so humans handle exceptions only.
What you save
Hours become minutes per task, handoffs stay consistent in Sheets or CRM, and the workflow file is yours to own. Import, add your keys and run.
More from portfolio