Interfaces

Ask from wherever your team already works.

Kipper answers finance questions in plain English — in Slack, Microsoft Teams, and SMS today, and through ChatGPT, Gemini, and Claude via the Model Context Protocol. Same read-only, audit-logged access, wherever you ask.

read-only audit-logged encrypted ~5s answers
What's our total outstanding AR right now?
Kipper

Total outstanding AR is $446,490 across 38 invoices.

current
$320,400
30 to 60d
$84,200
over 60d
$41,890

NetSuite, read-only, answered in 3.8s

Choosing an interface

Two ways in, depending on who's asking.

The six interfaces split into two groups, and the difference isn't cosmetic — it changes who sets it up and where the conversation happens. The data access is the same in both.

Native apps

Slack · Microsoft Teams · SMS

Kipper is the assistant. An admin connects the finance system once and grants access, then anyone in scope asks a question where they already work — a Slack channel, a Teams chat, or a text message — and the answer comes back in the same thread.

Best for: Rolling out to people who don't have a finance-system login and never will: sales and account reps, service and support staff, warehouse and operations teams, executives away from a desk.

  • Nothing to install for the person asking
  • Admin controls who can ask and what they can see
  • Answers stay in the channel, so the whole team sees them

MCP connections

ChatGPT · Gemini · Claude

Your own AI assistant does the asking. Kipper runs a Model Context Protocol server, and once it's connected, the assistant can look up your finance records mid-conversation — alongside whatever else it's helping with.

Best for: People who already run their day inside an AI assistant and want finance data available in that same conversation, without switching tools.

  • Same read-only, permissioned access as the native apps
  • Finance answers land in a wider working conversation
  • Set up per user, in the assistant they already use

Which one should your team start with?

Start where the questions already get asked. If your finance team is fielding "can you just look this up for me" messages in Slack, put Kipper in Slack — the request and the answer end up in the same place, and the rest of the channel sees it too. If those questions arrive by phone from people in the field, SMS removes the app entirely: they text a question and get the number back.

Microsoft Teams is the natural fit where IT already runs on Entra ID and wants SSO and enterprise controls in place before finance data goes anywhere. And where someone already spends their day in ChatGPT, Gemini, or Claude, an MCP connection means they don't have to leave that conversation to check a balance.

You don't have to pick one. Every interface reads through the same connection to the same finance system, so the answer doesn't change based on where the question was typed.

Behind every answer

Connect your finance system once.

Every interface reads from the same live connection to NetSuite, QuickBooks, or Xero — so the answer is the same wherever you ask.

FAQ

Questions about asking from anywhere

How access, permissions, and accuracy work once finance data is reachable from Slack, Teams, SMS, and AI assistants.

That's the point of Kipper. Most people who need a number from NetSuite, QuickBooks, or Xero don't have a seat in it — and often shouldn't, either because per-seat licensing is expensive or because broad access to the finance system isn't appropriate.

Kipper sits in front of the finance system. A workspace admin connects it once and decides who can ask what. Everyone else just asks in Slack, Teams, SMS, or their AI assistant. Only the admin needs finance-system credentials.

Kipper is read-only by architecture — it can never create, edit, or delete a record in your accounting system, in any interface. Access is permissioned by the workspace admin and enforced on every query, no matter how the question is phrased or which interface it arrives through.

Every question and answer is written to an audit log that admins can review. The security posture doesn't change between interfaces: connecting through ChatGPT gives that assistant exactly the same scoped, read-only access a Slack user has.

In Slack, Teams, and SMS, Kipper is the assistant you're talking to. With ChatGPT, Gemini, and Claude, your assistant is talking to Kipper — Kipper acts as a tool it can call through the Model Context Protocol.

The data access is identical either way. The difference is where the conversation lives and who administers it: native apps are set up once for the workspace, while MCP connections are configured per user in their own assistant.

No. Every interface reads from the same connection to the same finance system, so the answer to a given question is the same wherever it's asked. Most teams end up using several — Slack for day-to-day questions in channel, SMS for people who are away from a desk, and MCP for individuals who live in an AI assistant.

Kipper doesn't predict your figures. It translates the question into a SQL query, runs it against a synced copy of your finance records, and returns what the query returned. There's no document retrieval and no RAG step in between.

If the data can't answer the question, Kipper says so rather than estimating.

Kipper answers direct questions about transaction-level records and the supported financial statements, plus bank balances on a QuickBooks connection. It is not a general-ledger, journal-entry workflow, or bank-feed tool, and it does not read bank feed transactions or reconciliation status.

It also isn't an analytics tool: it won't build budgets, calculate financial ratios, or produce board narratives. Those stay with your finance team and their finance system.

Ask your finance system anything.

Pick the interface your team already lives in. Connect a finance system, set who sees what, and start asking — in minutes.