Dost launches AskDost, letting finance teams run AI agents on their own data

Dost, a finance automation platform built around native AI, has launched AskDost, a business intelligence assistant for finance and operations teams. According to Finextra, the tool connects to a company's ERP system and other financial data sources, answers questions in plain language and carries out finance workflows only after a person approves them. Each team can also run its own AI agents on its own data.

What is known about the launch

The announcement was published on 2 November 2026. The Finextra article describes the product in general terms and does not give pricing, availability dates or customer details. Any conclusions about adoption or market impact would therefore be premature.

How AskDost is designed

An ERP (enterprise resource planning) system is where a company keeps its accounting, financial and operational records. Getting a clear answer from one often requires reporting skills or help from an analyst. According to the description, AskDost is meant to change that:

• it connects to a company's ERP and other financial data sources;
• it answers questions asked in everyday language, without manual query building;
• it executes finance workflows, but only after an employee approves them;
• it lets each team run separate AI agents on its own data.

This makes business analysis more accessible to staff who do not work with databases directly.

Why the human approval step matters

The key distinction in the product description is between answering and acting. An assistant that only answers questions carries less risk than a system that can itself send a payment or change a record. Requiring human sign-off keeps the decision with a responsible employee, but it does not remove the risk of error. If the connected data is incomplete or wrong, both the answers and the suggested actions can be wrong too.

What it means for traders

This news is about the infrastructure of finance teams, not about trading directly. For a retail trader, it is a signal of how AI is gradually moving into accounting, reporting and operations inside companies. A few practical points:

• Answer quality depends on the quality of the connected data. This is a general principle for any analytical tool.
• Human approval reduces operational risk but does not eliminate it.
• It is worth looking at access terms, data protection and how the product performs in practice, not only at the press release wording.
• Conclusions about enterprise software should not be carried over into trading decisions. The announcement contains no market, asset or price data.

The Finextra report is useful as an example of how AI tools are being designed with human control built in. Further details, including pricing and user feedback, are not yet available.