Runtime Overview: Client Pulse Responder

The ecosystem.Ai Runtime is the Java 25 scoring engine (Amazon Corretto 25). It loads campaign properties, runs pre-score / model / post-score / reward plugins, and exposes the closed loop: POST /invocations then POST /response.

Why a runtime?

The runtime is the production prediction process. Workbench and Server configure models and deployments; the runtime serves them with millisecond latency, writes ContactLog / ResponseLog, and (for Dynamic Engagement) runs rolling learning.

What it does

• Score campaigns and return ranked final_result rows
• Learn from accepted offers on /response (singular)
• Load Java plugins (plugin.prescore, plugin.postscore, plugin.reward)
• Expose MCP (POST /mcp), ontology (/ontology/runtime/*), and campaign BDD (/test/campaign*)
• Host the Runtime Dashboard tester UI

Default HTTP port is 8091. Local run_local.sh often uses 8099. Hosted demo: https://medemo.ecosystem.ai .

Start here

• API Access
• Runtime Dashboard
• Campaign testing
• MCP Support
• Ontology for agents