How it fits together

The four things this service is made of, and what happens when a workflow runs.

The pieces

Application
Lives in: Your automation platform

The identity a platform authenticates as: Client ID, Client Secret, redirect URLs, the permissions it may request.

Authorization
Lives in: Our servers

One application allowed to use one connection with a set of permissions. Revoking it stops that application at once.

Connection
Lives in: Our servers

One TallyPrime company on one computer you have paired. This is what an operation actually runs against.

Connector
Lives in: The computer running Tally

The small Windows app that carries requests to TallyPrime and nothing else.

Applications and connections are separate on purpose
So you can cut off one platform without disconnecting TallyPrime — and vice versa.

What happens when a workflow runs

  • Your platform calls us

    It sends the step data to our API with its access token.

  • We work out who is calling

    The token resolves to one application, one connection and a set of permissions.

  • We check before we write

    Ledgers, stock items, godowns and bill references are confirmed to exist in your company, and the entry is checked to balance.

  • We build the entry

    The request becomes a properly formed TallyPrime voucher — the part that is genuinely difficult, and the reason this service exists.

  • The connector delivers it

    It travels down the connection the connector already holds open, and is handed to TallyPrime.

  • We confirm it landed

    TallyPrime reports success even when it recorded nothing, so we read the saved record back. Your response is built from that read.

Why nothing has to be opened on your network

The connector makes an outbound connection to us and holds it open — like a browser, or a chat app waiting for messages. Requests travel down that connection. No port forwarding, no static IP, no firewall rule, no VPN.

What makes an automated entry safe

Every write is verified

TallyPrime says “success” even when it saved nothing. So after every write we read the record back, and your response is built from what was actually saved.

A retry never doubles an invoice

Platforms re-send a step after a network hiccup. Each write carries a fingerprint, so a repeat updates the same record instead of creating a second one. Send an Idempotency-Key header to control it yourself.

Checks happen before anything is written

Every ledger, item, godown and bill named in a request is confirmed to exist first — so you get “the ledger X does not exist”, not a bare rejection.

Permissions

ScopeWhat it allows
tally.readRead ledgers, stock items, outstanding bills and transactions from the connected company.
tally.writeCreate invoices, payments, notes, journals and ledgers in the connected company.
connection.readSee which company and computer this connection points at.

The person approving an application sees exactly these sentences. Anything outside them is refused with forbidden.

Something missing on this page? Tell us and we will fix it.