A10 Networks has released the A10 AI Gateway, a centralized control plane for managing how organizations use AI agents, applications, and large language models. The software aims to give companies a single point for routing AI requests, tracking costs, and enforcing governance as AI usage spreads across teams and departments.
AI adoption has created a management problem. Developers increasingly build agents and embed AI into business systems, drawing on a growing roster of models. Without a central control point, teams have little visibility into which models are in use, how they perform, what they cost, and whether workloads run efficiently.
DHrupad Trivedi, president and CEO of A10 Networks, framed this as a demand for both speed and control. "Enterprises are being asked to move fast on AI and stay in control at the same time. AI use is spreading faster than teams can discover, govern, or secure it, driving up cost and leaving no single place to set the rules." Trivedi said the gateway offers "one intelligent control plane to see how AI is being used, to govern who can access which models, and match every request to the right model at the right cost."
Smart routing and identity-based controls
The first generation of AI gateways centralized access, simplified API management, and metered requests across providers. Enterprise AI is now more dynamic: different teams require different models, and every request varies in complexity. The A10 AI Gateway evaluates each request and routes simple tasks to more cost-efficient models while sending complex, reasoning-intensive tasks to more capable ones.
Access control works at the individual and group level. Organizations can apply routing policies by user profile, such as engineering versus finance, and keep them in sync with existing directory systems. That means access rules follow the org chart rather than living in a tool only an engineering team remembers to update.
Cost visibility per request
The gateway tracks real-time costs per request in dollars, sets token budgets per model and team with hard limits and soft-limit alerts, and enforces business-layer rate limiting per key or team. This central view lets managers see how AI is being used and governed across the enterprise, rather than handling spreadsheets of usage logs from security teams or submitting requests through procurement.
Deployment and the broader portfolio
Customers can run the gateway as software or with integrated hardware entirely within their own environment, whether that is on-premises, in a private cloud, or air-gapped. The full A10 portfolio also includes the security tools TrojAI and ThreatX, which cover AI security before deployment, at runtime, and at the network and API layer. Keeping data, models, and policy under customer control preserves data sovereignty, an advantage the company says a cloud-hosted gateway cannot match.
For managers responsible for AI budgets and governance, the key takeaway is that the tool tries to make cost and usage visible at the level of individual requests and team profiles. That offers a concrete mechanism for the balance between control and AI speed - at least until the next wave of models and agents arrives. AI for Management courses can help managers build the policy frameworks to deploy this kind of oversight, though the technology alone does not replace them.
Your membership also unlocks:
