Humanizing Artificial Intelligence for Platform Engineering: Helping Internal Developer Platforms Become Easier to Use

Short summary

Internal developer platforms fail on usability, not capability: developers facing 40-page wikis route around the paved road. AI's real value isn't replacing platform teams but grounding to approved templates and golden paths—mapping plain English service requests to constrained, policy-compliant infrastructure scaffolding. Use retrieval against live template repos and Backstage catalogs instead of fine-tuning, which freezes old policies and becomes stale within weeks.

• •AI lowers friction to self-service by mapping plain English to approved templates, not bypassing governance
• •Grounding via retrieval (live template repos, Backstage catalogs) outperforms fine-tuning, which locks in stale policies
• •Concrete example: scaffold Go service with Postgres/Kafka from predefined modules with constrained, policy-compliant variables

Generated with AI, which can make mistakes.