Ascerta raises $18M to show enterprises which AI projects make money

The question comes up in a budget review: which of these AI projects is making money. The dashboard on screen shows token counts and agent runs. According to SiliconANGLE, most companies can count token use and lines of AI-generated code, but few can say which AI projects create value ^[3]. Ascerta co-founder and chief executive David Tepper said what most businesses see is "meaningless vanity metrics" ^[4]. "They're struggling to derive the real impact AI has on their business," he said ^[9].

Tepper said conventional FinOps tools, built for cloud infrastructure spending, do little to show what an AI agent or coding assistant achieves for the business ^[12]. Much of what Ascerta does on day one is still FinOps, pointed at the AI bill. It tracks AI costs down to individual users, teams and applications ^[5]. Tepper said it finds hidden fees, enterprise discounts and sub-token costs that other tools miss ^[6]. Convoy, the compute component, is built to consolidate AI workloads ^[7].

The value claims rest on the other two components: Atlas, which handles return on investment, and Forge, which covers coding agents ^[7]. A return figure needs an outcome to divide by. For Atlas to report one, the revenue or savings number has to come from the buyer's own systems. A buyer evaluating it needs to know which systems those are.

Ascerta's own results show why the two halves need separating. Return on investment rises when cost falls, even if the value delivered stays flat. That means the reported 47% average improvement could come mostly from the same work behind the 86% cut in wasted spend ^[8]. The company also reports 24% faster agent launches ^[8]. The report does not say how many customers the averages cover or what baseline they were measured against.

Ramana Khanna, managing director of Dell Technologies Capital, described the round as a call on where budgets are going. "Most enterprises are moving beyond broad AI experimentation and focusing their investments on what delivers measurable business value," he said ^[10]. The round is most of Ascerta's capital. It had raised $4.9 million before this $18 million ^[1]. The published buyer record is two named customers, and the Wipro name appears on both sides: Wipro Ltd. as a customer and Wipro Ventures as an investor in the round ^[2][8]. Atos Group chief AI officer Florin Ratar said the platform was "instrumental" in scaling the company's Sovereign Agentic Studios initiative ^[11].

I think the cost half is the part a pilot can check quickly, and the value half is the part that would justify a new contract. The trade-off is that the checkable half overlaps with FinOps tooling a team may already pay for. The valuable half depends on outcome data the buyer has to supply. Two facts sort the case. The first is whether the team already sees AI cost by team and application, discounts included. The second is whether each AI project has a named owner and an outcome number it was bought to move.

- Hidden cost, defined outcomes: the strongest case for a combined tool. Both halves of the return figure exist, and the only missing step is joining them. - Hidden cost, no defined outcomes: the purchase is a cost ledger. The fair price comparison is the existing FinOps contract, and any ROI figure it reports is Ascerta's. - Visible cost, defined outcomes: joining the two may take one query in the BI tool analysts already use, so a pilot has to show attribution they could not build. - Visible cost, no outcome owners: an ownership problem. Atlas would add a return-on-investment column with nothing to put in it.