Dost launches AskDost to bring AI agents to finance teams
Dost, the Barcelona and London-based finance automation platform, has launched AskDost, an AI assistant built specifically for finance and operations teams. Unlike general-purpose tools that sit outside a company's data infrastructure, AskDost connects natively to ERP systems including SAP, Microsoft Dynamics 365 Business Central, Sage and Oracle, allowing finance leaders to query live transaction data and execute workflows from a single interface. The product is available as a standalone offering, requiring no existing Dost account, as well as inside the broader Dost platform.
The timing is pointed. Deloitte's latest UK CFO survey found that 67% of CFOs believe their organisations are only roughly a quarter of the way towards the AI transformation they expect to achieve. A separate CFO Connect Europe-wide survey from 2026 found that two-thirds of finance teams already use at least one AI tool, yet no specialist finance AI product named in the survey had reached even 1% adoption. The implication Dost draws is stark: finance professionals are routinely feeding company data into general-purpose assistants with no audit trail and no access to the structured records inside their ERP.
Human approval by design
AskDost addresses that gap through what the company describes as a human-in-the-loop architecture. The assistant prepares workflows across invoices, approvals, payments and reconciliation, but nothing executes until a person signs off. Every action generates a full audit record. A finance leader can begin with a single line in a profit-and-loss statement, drill through to the underlying invoice lines, and act on what they find, all within one conversation.
Personalisation is handled through Dost Studio, a no-code environment where individual finance teams build and deploy their own apps and agents on their own data. Queries built once are saved, shared with the team, and can refresh on a schedule, replacing the proliferation of duplicated exports and pivot tables that characterise finance operations in mid-market businesses.
Adam Barbera, CEO and co-founder of Dost, described the problem the product is designed to solve: "Ask something simple, like whether we're paying above contract or where cash will be in eight weeks, and the answer will be spread across three systems and someone's spreadsheet. Nobody should have to rebuild their finance function around a piece of software."
The agentic finance inflection point
The launch sits at a convergence point that extends beyond accounts payable software. The broader transition to agentic AI, where autonomous systems take sequences of actions rather than merely answering questions, is reaching financial operations faster than most enterprise functions because the data is relatively structured and the compliance requirement for auditability is already embedded in regulation. Dost's human-approval layer reflects the reality that finance chiefs and their auditors will not accept autonomous execution without an oversight mechanism, whatever the efficiency upside.
For capital allocators watching the enterprise software stack, the picture is shifting. The mid-market ERP ecosystem has long been dominated by legacy vendors, but a new cohort of native-AI platforms is now targeting the operational layer between ERP data and human decision-making. Dost's move to offer AskDost as a standalone product, decoupled from its broader platform, lowers the switching cost and widens the addressable market, a classic land-and-expand motion in a segment where procurement cycles are long. Investors tracking the agentic-AI wave in vertical software will note the approach: rather than displacing the ERP, Dost sits on top of it, which reduces implementation risk and accelerates time to value.
Dost reports that the underlying platform has processed more than five million accounting transactions at 95% data extraction accuracy for clients including HomeServe, Honest Greens, Housfy and Primaprix. AskDost builds the conversational and agentic layer on that foundation, and represents the company's first product designed to be adopted without the full platform in place.