Ascerta Raises $18M to Measure AI Business Value

Ascerta, the Bellevue, Washington company formerly known as Pay-i, announced an $18 million Series A on September 30, 2026, led by Dell Technologies Capital and joined by Hitachi Ventures, BGV and Wipro Ventures. The round brought its reported total funding to $22.9 million. Ascerta is expanding from AI cost tracking into measuring adoption and business outcomes.

The business question behind that shift is whether spending on models, copilots and agents produces enough value to justify more investment. Ascerta proposes to connect what AI costs and how people use it to the work completed and the results an enterprise cares about. The financing supports that plan; a useful ROI figure still depends on evidence that the measured result came from the AI initiative.

The funding and the broader Ascerta name

Pay-i began with a narrower problem: making AI consumption and its cost visible. A bill can show how much a model call or tool costs, but it cannot establish whether the underlying task improved. The Ascerta name reflects a wider pitch to technology and finance leaders who must decide which initiatives to expand, change or stop.

The public funding figures use different levels of precision. CEO David Tepper’s account of the round gives $18.1 million for the Series A and $23 million raised in total. That is consistent with describing the round as about $18 million and the cumulative amount as about $23 million, although the published $22.9 million total and the founder’s $18.1 million round figure use different rounding conventions.

Founded by Tepper, Doron Holan and Erik Winters, the company is using the new capital to expand its platform, team and integrations. Those plans matter to buyers because an AI value measure has to draw from tools already in use as well as from the systems that record business results. More integrations may widen the activity a platform can observe; they do not, by themselves, turn activity into a financial return.

What Ascerta proposes to measure

Ascerta’s approach has three connected layers: cost, adoption and outcome. Cost covers the resources consumed by a use case. Adoption shows who uses an AI tool and for what work. An outcome is the change in a business measure the initiative was meant to improve. Keeping those layers distinct prevents a rise in usage from being counted automatically as a rise in value.

In Ascerta’s announcement, CEO David Tepper calls tokens, generated code and agent runs “meaningless vanity metrics” when separated from business impact; the company also reports customer-level improvements of 47% in AI initiative ROI, 24% in agent launch time and 86% in wasted AI spend. The announcement does not provide the sample sizes, baselines or calculation methods needed to assess how widely those reported improvements apply.

Ascerta assigns different parts of the problem to its products. Atlas addresses value, adoption and ROI across individual workflows or a portfolio. Forge focuses on coding-agent use and engineering productivity. Convoy addresses AI capacity an organization provisions itself. The platform is intended to work alongside internal applications and tools including Microsoft Copilot, Amazon Bedrock AgentCore, Salesforce Agentforce, GitHub Copilot, Claude Code and Codex.

That range illustrates why one activity count is inadequate. A coding assistant can generate more code while leaving review effort or delivered quality unchanged. An agent can run more often because demand has grown, or because it repeatedly fails and retries. A capacity pool can be busier without supporting more useful work. Each interpretation requires a record of the task and its result, not just the volume of AI activity.

When an AI ROI figure becomes useful

A decision-grade return needs a traceable chain from an AI use case to its full cost and a defined business result. These are evaluation criteria for any platform making that claim, including Ascerta; they are not a claim that every Ascerta deployment already meets them.

Comparable cost: Put model usage, tool subscriptions, provisioned capacity and relevant operating expense against the same use case and period. A low-cost call may belong to an expensive workflow once retries and human oversight are included.
Task-level adoption: Identify the people or teams using the tool and the work they use it for. Licenses show access, while agent launches show activity. Neither establishes that the work was completed better.
Defined outcome: Specify the business measure before calculating a return. For a coding workflow, completed work and its quality may matter more than lines generated. The measure must match the reason the organization funded that particular initiative.
Credible comparison: Compare the outcome with a relevant baseline and account for other changes in staffing, demand or process. Otherwise, an improving business measure can be credited to AI when several factors moved together.
Reconciled value: Match the valued improvement and the complete cost over the same period. Time saved becomes a defensible dollar benefit only when the organization can explain how it used that time or which expense it avoided.

The hardest link is often between observed use and a result outside the AI tool. A platform may directly meter consumption and detect adoption, while an enterprise defines the KPI and supplies the operational record behind it. Even when those records move together, attributing the change to AI requires a sound comparison. That distinction is central to whether an ROI estimate can guide a spending decision.

The evidence buyers will need next

Ascerta’s reported customer improvements identify the outcomes it wants to measure, but the published aggregate figures leave buyers without a way to judge the range of results across different customers and workflows. A customer-level account would be more informative if it showed the initial cost and outcome, what changed during use, the period measured and how the resulting value was calculated.

The company’s next test will arrive as its broader platform and integrations reach more enterprise workflows. For a buyer deciding which AI initiative deserves more funding, the consequential result is a documented change in business performance that can be weighed against the entire cost of producing it.