Observing, Testing and Connecting Spring AI Applications: Microcks, Reshapr and OpenTelemetry

for more info visit: https://www.salaboy.com

Transcript

1. Mauricio Salatino @salaboy Laurent Broudoux @lbroudoux Testing and Observing MCP connected Spring AI applications with Microcks, Reshapr and OpenTelemetry
2. Who we are! Mauricio Salatino @salaboy Laurent Broudoux @lbroudoux CNCF TOC member, Java Champion, Ecosystem Engineer @ DashO Cloud native Architect @ Postman, Microcks & Reshapr co-founder
3. The AI/Cloud-native Spring Store Logical view Store 🛍 Shipping 🚚 LLM 🧠 󰠁 User ask Shipment place Order orchestrate process
4. Understanding & Testing our Agentic Apps
5. Observing Spring AI Store 🛍 LLM 🧠 󰠁 User place Order orchestrate process Otel Collector 👀 traces, metrics,logs Dash0 https://www.salaboy.com/2026/05/27/five-semantic-conventions-one-config- property-observing-spring-ai-with-arconia/
6. Discover app & cut dependencies Store 🛍 LLM 🧠 󰠁 User place Order orchestrate process ❌ LLM integration can be slow, costly, tedious to use in CI environment… We can use simulations/mocked responses to speed-up the feedback loop when testing non-LLM areas
7. Microcks Turn artifacts into light & smart simulation INGEST PROVIDE MOCKS WELL-KNOWN ARTIFACTS 🧠 SMART MOCKS API Consumer and others
8. Discover app & cut dependencies Store 🛍 󰠁 User place Order Dash0 👀 traces, logs, metrics https://github.com/anthropics/anthropic-sdk-java/pull/315 traces, logs, metrics
9. Wiring tools to existing APIs with MCP
10. Events and MCP Store 🛍 Shipping 🚚 LLM 🧠 󰠁 User ask Shipment place Order orchestrate process Shipping MCP Kafka ✉ ✉
11. So, you want to write an MCP Server? spring-ai-starter-mcp-server has everything you need! Protocol adapters + annotations: @McpTool @McpResource @McpPrompt But imagine you have an existing backend 🤔 • Itʼs a lot of boilerplate code: annotations + RestClient • It requires you to become an MCP expert and a network proxy and security expert • You need to deploy, operate and update a new component for each MCP Server
13. Reshapr Turn artifacts into MCP Server INGEST PROVIDE MCP SERVERS ENTERPRISE-GRADE MCP SERVERS 🛡 NO CODE GEN!
14. Reshapr is reshaping API to MCP RENAME INCLUDE EXCLUDE ENRICH SIMPLIFY CONDENSE FILTER FORMAT SECURE MCP ENDPOINT & API BACKEND ACCESS Adapt API to Agent needs. Control context.
15. Reshapr architecture FLEXIBLE HIGHLY DYNAMIC FULLY MULTI-TENANT SCALABLE OBSERVABLE
16. Making your system understandable Store 🛍 Shipping 🚚 󰠁 User ask Shipment place Order Shipping MCP Kafka ✉ ✉ Dash0 👀 traces, metrics and logs
17. Wrap-up
18. Key Learnings 🤔 • Mocking & Contract Testing come together for cutting dependencies and providing with fast feedback loops • Reshapr provides an easy, flexible and enterprise-grade way to connect your existing API as MCP to your AI-infused app. • Observability is mandatory for understanding how distributed AI-infused apps are expected to work • All of the above can run on your machine without a Kubernetes cluster!
19. THANK YOU! https://github.com/salaboy/spring-ai-agentic-webinar