API Gateways and Identity Enable Agents

The bottleneck for enterprise agents is not model intelligence, it is getting safe, reliable access to real systems. Coding agents can already work inside bounded environments, but a finance or support agent has to log into SaaS tools, call the right internal APIs, respect permissions, and leave an audit trail. That makes gateways, identity, and tool governance more important than the model itself for broad enterprise rollout.

• Kong sits in the traffic path between apps, APIs, event streams, LLMs, MCP servers, and other agents. That position matters because every agent step is really another API call that needs authentication, routing, rate limits, logging, and sometimes cost controls before it can touch an internal system.
  1 sacra 2 sacra 5 konghq
• The near term proving ground is coding agents, because the tool set is tighter and permissions are easier to scope. Broader business workflows are harder, since an agent may need to read Salesforce, update ServiceNow, check Slack, and trigger an internal system, all under different access rules.
  4 openai 6 konghq 7 sacra
• The market is already shifting from raw model access to agent management infrastructure. OpenAI now markets agent creation, monitoring, and workflow execution for enterprises, while MCP and adjacent tooling are being pushed as a standard way to connect models to company systems more consistently.
  3 openai 4 openai 6 konghq 8 sacra

Over the next few years, the winners in enterprise agents will be the companies that turn scattered internal systems into clean, permissioned tools an agent can actually use. That pushes value toward API gateways, identity layers, and orchestration software, because those are the pieces that convert a smart model into a dependable worker.