Built in 24 hours: Transaction comments on Osome MCP

Built in a day, inspired by you

Bookkeeping questions are rarely solved by looking at numbers alone. Most of the real work happens in the conversation thread attached to a transaction — where your accountant asks for context, you explain it, and the details stay pinned to that payment for future reference.

Previously, our Model Context Protocol (MCP) server let AI assistants search for transactions, but those critical comment threads were isolated inside the web app. If you were working inside ChatGPT, Claude, or Gemini, you had to break your workflow, open Osome, find the transaction, and manually read or type your update.

We’ve closed that gap completely.

Why this matters

• No more tab-switching: Read and respond to your accountant's questions directly within your AI chat workflow.
• Centralised context: Every comment added via your AI assistant posts directly to the official transaction record inside Osome.
• Automated follow-ups: Power users and partners can now build seamless, automated reconciliation workflows around transaction inquiries.

How to use it

Once your AI tool is connected to the Osome MCP server, simply ask natural questions about your transaction notes.

Try asking your assistant:

• "What did my accountant ask about this payment?"
• "Add a note that this was the July software subscription."
• "Find the card payment from 12 July and tell my accountant I uploaded the receipt."

How it works for assistant builders

Your AI assistant uses a simple two-step process to handle transaction conversations:

1. It uses search-transactions to identify the numeric document ID of the payment.
2. It calls get-transaction-conversation to read existing thread comments, or add-transaction-comment to post a new note (and automatically create the thread if one doesn't exist yet).