Companies have spent years investing in becoming data-driven, accumulating sales, marketing, financial and product information across an expanding collection of systems. The harder problem now is getting artificial intelligence to make sense of those systems together. Euno, an enterprise AI context platform built for that problem, closed a $23 million Series A led by N47 on Sept. 9, bringing its total funding to $29 million, according to the company.
That data can be “fragmented and complex and chaotic,” full of duplications, inconsistencies and conflicting information, Sarah Levy, co-founder and CEO of Euno, told PYMNTS. Other knowledge, like which spreadsheet a team uses or which metric executives trust, may never have been formally documented. “A lot of the information, who is using what, how things are computed, what things mean, they are stored in so many systems,” Levy said. “There is also a unique knowledge that is actually in people’s minds.”
That creates a different infrastructure requirement for AI. Connecting a large language model to a database gives it access to information, but it doesn’t tell the model which information is authoritative or how different pieces of data relate to each other.
Levy said a typical enterprise runs 10 to 12 different data platforms and business intelligence (BI) tools, and waiting to consolidate them before deploying AI could take years the company doesn’t have. Euno’s platform sits across those existing systems rather than replacing them, integrating with data lakes, warehouses, BI and transformation tools to collect metadata and operational signals including usage, calculations, definitions, ownership and data flows. It then maps the relationships among them so an AI system can understand how information was created and used, without collecting or storing the underlying enterprise data itself, according to Levy.
Siloed AI Agents Create Siloed Context
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
The context problem becomes more pronounced as companies deploy different AI applications across departments. Many enterprise AI implementations remain tied to individual workflows such as customer success, invoicing or warehouse management, Levy said, with each one developing its own glossary, rules and data context. When the underlying corporate data changes, those separate context layers require manual maintenance. “People find themselves continuously maintaining that and hand-holding that,” Levy said. “If you put it in simple words, people work for the robots.”
Levy pointed to Euno customer AlphaSense as an example. The market intelligence company has years of reports, dashboards and legacy logic across systems including Tableau and Snowflake as it works to make Anthropic’s Claude available to business users, Levy said. Euno maps the existing Tableau logic and helps translate it into governed metrics and semantics in Snowflake, so when employees query Claude, the system can provide context around an answer, including where the information originated and how it was calculated.
For enterprise users, the answer itself is only the start, Levy said. “Where did you bring it from? How did you come up with it? Who created that analysis?” she said, describing the kinds of follow-up questions business users ask. “This is the true conversation. It’s not enough to just get an answer.”
Regulators Want to Know What AI Agents Can See
Giving agents deeper institutional knowledge creates another problem: an AI system shouldn’t know everything a company knows. The same context that makes an agent more useful can expose information outside the scope of its job.
Euno addresses that through role-based personas, Levy said. An agent supporting financial reporting, for example, could be limited to a governed collection of financial key performance indicators and tables, while a product-focused agent could receive a broader set of context, with boundaries set by a central governance team. “You only get access to the context that is relevant to your role and your task,” Levy said.
In banking and payments, that boundary is a supervisory question. A June KPMG survey found 51% of banks were piloting AI agents, PYMNTS reported. That same month, the Office of the Comptroller of the Currency and the Federal Reserve were pressing banks with detailed questions about their AI controls, including whether a system can access or infer data beyond its authorized limits, PYMNTS reported. An agent scoped the way Levy describes, one that can name the governed table behind its answer and the team that owns it, hands the examiner and the analyst the same record.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Add as Preferred Source
