Devs Just Wanna Have Fun! Enhancing Developer Experience With Natu...
Talkat Open Source Summit Japan + AI_dev: Open Source GenAI & ML Summit + Automotive Linux Summit 2025 · 2025-12-08..12-10·
summary
Attempts to let developers self‑provision via natural language use Goose recipes, an MCP client, and MCP servers for Argo CD and Dynatrace to create a Kubernetes cluster, install ArgoCD and OpenTelemetry components, and send telemetry into Dynatrace; Argo CD’s MCP server lacks needed features so application YAMLs had to be hand‑made, a prompt ran successfully nine times then failed on the tenth due to model nondeterminism, sessions log per file and metrics flow via OpenTelemetry, recommended guardrails include Goose recipes, Argo CD MCP server, and 0.5 temperature.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 2:48Developer mental load today
- 4:35Introducing MCP and AI workflows
- 5:38What is Goose
- 7:50Natural language workflows with Goose
- 10:17Challenges in prompt writing
- 13:26Tips for prompt writing
- 17:45Choosing the right LLM
- 19:50Monitoring AI tools
- 24:25Selecting MCP servers
- 27:16Prompt workflow viability
- 28:40Key takeaways
- 31:07Future directions
- 35:05Experimentation in tech
- 36:49Closing remarks
related talks
44mSupercharging DevOps with MCP (Without Opening a Security Hole)
25mCruising the AI Highway: Standardized MCP Deployment and Governance With Backstage
26mObservability for Your Internal Developer Platform: Instrumenting Bac...
30mFrom Experiment to Enterprise: How Block Operationalized MCP at Scale
34mChoose Your Own Adventure: The Dignified Pursuit of a Developer Platf...
35mCloud Native Development Without the Toil
39mCI/CD, APIs, and Scaling: What Every Cloud Native Developer Needs To Kn...
26mLet the Platform Build Itself: Using AI To Construct an Internal Developer Platform With CNCF Tools