Adaly wants $11M for agent data plumbing. It still names no customers.

Adaly filed a Form D notice for an $11 million offering on August 13, 2026, less than three months after announcing the commercial launch of its federated DataOS ^[1][5]. For anyone evaluating the category of software that sits between AI agents and operational business systems, the number worth studying is not 11; it is the count of named customers, published prices and disclosed deployments, which remains zero ^[7].

Start with what the filing does and does not say. A Form D is a notice of an exempt securities offering, and filing one does not establish that the offering has been fully sold or that a financing has closed ^[4]. The notice does not establish how much Adaly has sold, and participating investors, the valuation and total funding to date all remain unverified ^[3]. The company is run by CEO Kyle Csik and CTO Aleksandar "Sasha" Grujicic ^[2]; The Org lists Csik as a co-founder since May 2024 and a former chief data and technology officer at Night Market ^[21]. That puts the raise roughly 27 months into the company's life and 78 days after the commercial launch ^[22][23].

On traction, the public record is a set of categories rather than a set of accounts. The May 27 launch announcement said Adaly was working with Fortune 500 businesses across healthcare, media, retail and consumer packaged goods, without naming them or disclosing revenue, pricing, deployment scale or technical performance ^[6][7]. A May 16 post titled "Thank You To Our Pilots" described a Fortune 500 food company, several large retailers, two media agencies and a major publisher, again unnamed, and did not establish how many became paying customers ^[8]. Pilot-to-paid conversion is the single most useful number in early enterprise infrastructure, and it is the one absent.

The product claims carry the same shape. Adaly says its DataOS queries data where it resides, connecting models, applications, agents and employees to operational records without first copying everything into a warehouse or lake ^[10], and that it preserves permissions, lineage, auditability and governance from the underlying systems while not training on customer data ^[11]. Those security and deployment claims have not been independently validated ^[11], and the company has not published independent benchmarks or production-reliability data ^[15]. That matters more here than in read-only tooling, because Adaly says the platform supports write access so an agent can act inside an operational system after retrieving information ^[13]. An agent that changes inventory, customer, financial or clinical records needs tightly enforced permissions and a reliable audit trail ^[14]. Adaly sells three packages: Vessel as an embedded connectivity layer, Studio as a hosted analytical workspace, and Atlas as a bespoke deployment for complex institutions ^[12] - the last of which is a heavy commitment for a company its LinkedIn page lists at roughly 15 employees ^[9].

The competitive set is not empty either. The product overlaps established data-virtualization and query platforms including Denodo, TIBCO Data Virtualization, Starburst and Dremio ^[16], and the adjacent agentic-data category already includes Reltio AgentFlow, Acceldata, Dataiku's agent platform and VAST Data ^[17]. Adaly's stated distinction is live federation plus semantic context, existing governance and agent write-back, which is positioning rather than measured advantage; it has disclosed no comparative latency, connector coverage, implementation time or customer retention ^[18][19].

Watch for three things: a named reference running write access in production, any published connector coverage or implementation timeline, and an amended Form D or closing announcement establishing what was actually sold ^[19][4]. The underlying constraint is real - agents need current information and, when permitted to act, controlled access to operational systems ^[20]. Meeting it at production scale inside corporate security controls is the part still unevidenced ^[20].