Nebuly announced a $21 million Series A led by Planven on November 12, 2026, to expand software that estimates the business value of enterprise AI agents. The Turin- and New York-based company’s financing brings attention to a question facing research administrators evaluating AI: whether recorded use translates into useful work, additional capacity or financial savings.
A funding round for AI value measurement
In its funding announcement, Nebuly named Exor Ventures, Tenacity Capital, TGVG Capital and Fastweb+Vodafone, investing through Founders Factory, among the participants, alongside returning investors and angels. CDP Venture Capital SGR also supported the round through its Artificial Intelligence Fund and Digital Transition Fund, with NextGenerationEU support identified in the release.
The company says it has signed multiple six-figure enterprise contracts and names Oura, Iveco Group, Fastweb+Vodafone, CNH and Zeiss as customers. These are company statements about commercial adoption. They do not establish how much productivity improved at those organisations or how closely an estimated return matched an observed financial outcome.
What the platform is designed to measure
Tech.eu’s November 12 report describes two measurement settings. For employee-facing AI, Nebuly assesses time saved, capacity created and tasks employees can undertake with assistance. For customer-facing agents, it tracks revenue influenced by AI and signals associated with customer churn or further sales opportunities. The software also identifies differences in how teams use AI and where training might improve its use.
The company says its platform connects to custom-built agents and assistants based on Claude, ChatGPT and Microsoft Copilot. It describes deployment on an organisation’s own infrastructure or private cloud, with models that can analyse agent behaviour without sending enterprise data to external model APIs. Those are deployment capabilities described by the vendor; they do not, by themselves, establish the suitability of a particular research-data workflow.
How an interaction becomes an estimated value
Nebuly’s published calculation method makes the role of assumptions explicit. It analyses conversations that meet the customer’s definition of success, uses a proprietary model to estimate how long the task would have taken manually, and multiplies estimated time saved by what it calls net hourly cost. Customers can adjust the assumptions and exclude unsuccessful conversations.
The resulting monetary figure is therefore an estimate based on task classification, a manual-work baseline and an assigned value for time. Altering the hourly assumption can change the reported monetary value even if the underlying interaction stays the same. Nebuly’s report also includes hours saved, completed tasks and task completion rate, allowing those components to be considered separately.
For a research office, the distinction matters: capacity released for other work and money removed from an operating budget are different outcomes. A time-based valuation does not establish a reduction in expenditure. The success definition also determines what enters the calculation, making it central to interpreting the headline figure.
The wider measurement problem
In SiliconANGLE’s reporting, chief executive and co-founder Francesco Signorato described the difficulty of assessing AI across tools from several providers and internally developed systems. He argued that counts of users, prompts and responses do not show whether an employee recovered useful time or a customer received the answer needed. He also warned against making hiring and headcount decisions on assumed productivity gains.
A separate measurement-science discussion from NIST, published December 2, 2025, identifies related questions: whether evaluations measure the intended concept, how uncertainty is reported, which human or non-AI baselines are appropriate, and how downstream outcomes are measured after deployment. That discussion supplies context for evaluating AI claims; it is not an assessment or endorsement of Nebuly.
Supportive reactions come from customers and investors
The announcement includes comments from Fabio Ricciato, Iveco Group’s head of Digital, Data, AI and Architectures, who said the group had worked with Nebuly since its early days. He described understanding and optimising AI’s value as increasingly important alongside deployment. Planven managing partner Giovanni Canetta Roeder said the ability to connect agent activity with business value helped motivate the investment.
These comments describe the expectations of a named customer and the lead investor. Neither statement supplies a controlled comparison of work completed with and without AI. Their support helps explain the commercial rationale for the financing while leaving the accuracy of particular value estimates as a separate question.
Why research offices will scrutinise the assumptions
For universities, the significance is an evaluation issue rather than evidence of a campus rollout. Applying NIST’s questions to the method Nebuly describes, a research administrator would need to distinguish a successfully completed conversation from a successfully completed institutional task. This is a construct validity question: whether the observed signal represents the outcome the institution actually values.
The same distinction affects comparisons between teams. Differences in task mix, success criteria or the value assigned to an hour can affect the figures without establishing a difference in research quality. CASRAI’s explanation of measurement and design validity provides background to that distinction. These are implications of the measurement approach, not reported findings about university users.
Because the method analyses user-agent conversations, the input data also makes research data governance relevant to any institutional evaluation. A self-hosting option answers a deployment question; decisions about which conversations enter an assessment and who can inspect the resulting outputs remain part of the institutional context.
What Nebuly plans next
Nebuly says the capital will support further platform development and expansion of its go-to-market team across the United States and Europe, a plan also reported by The Next Web on November 12. The announcement describes a commercial expansion plan rather than a dated product launch. The development confirmed this week is the financing; claims about the value generated by customers’ AI still depend on how that value is defined and measured.
Sources
- Nebuly: Series A announcement, November 12, 2026. Checked 2026-11-16.
- Tech.eu: Nebuly raises $21M Series A, Tamara Djurickovic, November 12, 2026. Checked 2026-11-16.
- Nebuly: How to measure the ROI of enterprise AI, undated methodology. Checked 2026-11-16.
- SiliconANGLE: Nebuly’s enterprise AI measurement funding, Mike Wheatley, November 12, 2026. Checked 2026-11-16.
- NIST: Accelerating AI Innovation Through Measurement Science, December 2, 2025. Checked 2026-11-16.
- The Next Web: Nebuly raises $21M Series A to measure enterprise AI ROI, Ana-Maria Stanciuc, November 12, 2026. Checked 2026-11-16.







