Yes — a connected assistant reads Empire's own company data and answers a supported question directly, without changing how Empire itself works.
Empire and TallyPrime both work today
Dhandha GPT's first and current connector is Empire, the Windows ERP many Surat textile SMEs already run. The assistant reads a local snapshot of a business's own Empire data and answers a supported, read-only question — a customer balance, a sales total, a stock figure — on Telegram, in Hindi, English or Hinglish.
This is not a rebuild of Empire and it does not touch how Empire itself works. Entering, editing and closing records all stay inside Empire.
What a supported question looks like
The product supports a defined, growing set of questions rather than open-ended queries against the whole database. If a question falls outside that set, the assistant says so rather than inventing an answer from a similar-sounding report.
A worked example (invented)
Consider an invented Empire user, Radha Traders. Before releasing a new order to a dealer, the owner asks for that dealer's current outstanding balance. The assistant returns the balance and the date it reflects, matching what an Empire ledger report would show for the same company and period. The owner still decides whether to release the order.
Getting started
Read preparing Empire records for a trial for the non-secret setup details to bring, and reconciling a report total if a figure looks different from what Empire itself shows.
What a typical first trial covers
An Empire trial usually starts with one company, one or two supported questions — commonly an outstanding balance and a sales total — and a small set of named, authorised users. This mirrors the setup steps in the Empire records preparation guide: bring the exact product and version details, agree company scope, and pick one question you can already verify.
Customised Empire installations
Many Empire installations carry customisations built for a specific business over the years. A supported question is matched against the standard report definitions the connector understands; a heavily customised report may not have a direct match yet. Say so during setup rather than assuming every custom report is automatically covered.
A worked example, continued
Returning to the invented Radha Traders example: if the dealer credit-limit rule lives in a customised Empire field rather than a standard one, the assistant may not yet retrieve it directly. The owner's setup conversation should surface this gap so the team knows to check that figure manually until it is added to the supported question set.
Where this differs from a Tally trial
The overall shape — scope, one test question, freshness check, acceptance record — is identical to a Tally trial (see AI for Tally users), because both start from the same read-only, records-first design. The only real difference is which report definitions are already supported today.
Who typically runs this trial
Usually the owner and whoever manages the Empire installation day to day — often an accounts person or an external Empire support contact — run the trial together. Neither role alone has the full picture: the owner knows which question matters, and the Empire administrator knows the installation's specifics.
A note on timing
Empire and TallyPrime both work today, so either can start a 14-day free trial. Confirm current status before planning a specific start date.
A note on Empire versions
Empire installations vary in version and configuration across businesses that have run it for years. State your exact version and any known customisations honestly during setup rather than assuming an older or heavily modified installation behaves identically to a newer one — this single disclosure step avoids most early confusion.
Who to loop in on the business side
Besides the Empire administrator, loop in whoever regularly asks for the figures the trial will cover — often sales or accounts staff — so the chosen test question reflects a real, recurring need rather than one picked in isolation by the owner alone. A question nobody actually asks day to day is a weak first test, even if it is easy to verify.
What "supported" will likely mean over time
The catalog of supported Empire questions is expected to grow as more businesses trial it and surface real, recurring needs. A question that is not supported today is not necessarily excluded permanently — it may simply not have been prioritised yet. Ask your setup contact whether a specific gap is on a near-term roadmap before assuming it never will be covered.
Questions you might ask
Does the assistant change anything inside Empire?
No. It only reads a supported set of records and answers a question. It cannot add, edit or delete an Empire entry.
Can it answer every possible question about my Empire data?
No. It supports a defined, growing catalog of questions. Outside that catalog it refuses rather than guessing.
What if my Empire installation is customised?
Bring your exact product and version details during setup — see the Empire records preparation guide — so supported questions can be matched to your installation.