KeewanoDB Launches Agent‑Focused Database, Secures $12M Funding

KeewanoDB, a Tel Aviv‑based startup under the formal name Sandstorm Ltd., unveiled its new event‑oriented database, KeewanoDB, today. The launch is accompanied by a $12 million funding round led by Hetz Ventures and a16z Speedrun, with participation from other unnamed investors. The company claims its design delivers real‑time context for artificial intelligence agents, a feature it says is lacking in conventional relational databases.

The core proposition of KeewanoDB is that AI agents—software entities that autonomously gather data, make decisions, and act—require a data layer that can emit context as events occur, rather than relying on batch‑style queries. By structuring data around events, the database can push updates to agents instantly, reducing latency and enabling more dynamic decision loops. This contrasts with traditional relational models that emphasize schema rigidity and transactional consistency at the expense of real‑time propagation.

KeewanoDB’s architecture reportedly emphasizes event streams, allowing agents to subscribe to specific streams and receive updates in near‑real‑time. The startup argues that this model aligns with the operational patterns of modern AI workloads, where inference and learning cycles are tightly coupled to incoming data streams. The claim is that the database’s event‑centric approach can lower the overhead of context retrieval, a critical bottleneck in large‑scale AI deployments.

The funding round’s composition is noteworthy. Hetz Ventures, a venture fund focused on high‑growth tech, and a16z Speedrun, a program within Andreessen Horowitz that accelerates early‑stage AI companies, both bring deep expertise in scaling AI infrastructure. The presence of a16z signals confidence in the agent‑centric paradigm, as the firm has historically invested in foundational AI technologies such as data lakes, model training platforms, and inference engines.

KeewanoDB’s announcement arrives amid a broader wave of agent‑centric initiatives. Splunk, a data‑analytics giant, recently announced a platform overhaul that repositions the company as a data and governance foundation for the “agentic enterprise.” Splunk’s shift underscores the industry’s recognition that agents—whether software bots, automated monitoring tools, or autonomous analytics pipelines—are becoming the primary consumers of data. Concurrently, Eve Security, a runtime security startup, raised $4.5 million to stop malicious AI agents at runtime, highlighting the security challenges that accompany widespread agent deployment.

Another example of agent‑centric investment is Evvy, a precision diagnostics startup focused on women’s health, which closed a $40 million Series B round. While Evvy’s domain differs from KeewanoDB’s, its funding reflects a broader appetite for AI‑driven, context‑aware solutions across industries. The convergence of funding across diverse sectors suggests that the market is increasingly valuing infrastructure that can support autonomous decision‑making at scale.

KeewanoDB’s competitive advantage lies in its promise of low‑latency context delivery. In AI pipelines, the time between data ingestion and inference can be a critical performance metric. By pushing events directly to agents, KeewanoDB aims to reduce the round‑trip time that traditional databases impose. Additionally, the event‑oriented model can simplify scaling, as new agents can subscribe to streams without requiring complex schema migrations.

However, the transition to event‑centric databases is not without challenges. Maintaining consistency across distributed event streams can be complex, especially when agents perform write operations that must be reconciled with other agents’ reads. Moreover, integrating an event‑oriented database into existing enterprise ecosystems—many of which rely on relational databases for transactional workloads—may require significant architectural changes.

Potential beneficiaries of KeewanoDB include AI developers building autonomous systems, enterprises deploying large‑scale monitoring agents, and security firms that need to track agent behavior in real time. The database’s real‑time context could also aid compliance teams that monitor AI decision processes for bias or regulatory adherence.

The counter‑argument points to the entrenched position of relational databases and the maturity of event‑driven platforms such as Kafka. Critics may argue that the overhead of adopting a new database model outweighs the benefits, especially when existing pipelines can be adapted to consume events from legacy systems. Additionally, the lack of a proven track record for KeewanoDB’s performance at scale introduces uncertainty.

Nevertheless, the convergence of funding and strategic announcements from multiple high‑profile players suggests a shift toward agent‑centric infrastructure. KeewanoDB’s launch, backed by a notable $12 million round, signals that investors are willing to back specialized databases that cater to the unique demands of autonomous agents. The broader ecosystem—spanning analytics, security, and precision medicine—demonstrates that real‑time context is becoming a critical commodity for AI systems.

In sum, KeewanoDB’s entry into the market marks a tangible step toward a future where data layers are designed explicitly for agents. The company’s funding and positioning within the agentic AI narrative indicate that the industry is moving beyond traditional relational models toward architectures that can keep pace with the speed and autonomy of modern AI workloads.