Need to Align Enterprise AI Rollout with Product GA Timing
Enterprise contracts only cover generally-available features, so AI tooling rollouts at scale must wait for GA even when developers are already asking for preview capabilities.
Take Control Of Your Signals — Become a Naftiko Design Partner Today!
Persona Story:
Maya, the developer experience & AI engineering lead, cannot roll out preview AI features across a developer population of thousands. The enterprise agreement with the vendor covers GA features only. Developers see MCP and agent skills announced at developer events and want them immediately — but the governance answer has to wait until product graduates from preview. Maya has to manage that gap without losing developer trust or letting shadow usage fill the vacuum.
Problem Context
- AI vendors are shipping MCP, agent skills, and adjacent features to preview in weeks, while enterprise contracts are renewed in years
- Enterprise legal and procurement terms typically cover GA features, support SLAs, and production indemnification — none of which apply to preview
- Developers see preview features at conferences and in documentation and do not naturally know what is off-limits for their employer
- Rolling back a preview feature that has been casually adopted is harder than waiting for GA in the first place
- Vendor lifecycles (preview, public preview, GA) do not map cleanly to enterprise adoption phases
Problem Impact
- Developer pull for new AI features outruns the governance clock, creating a gap where the answer is “not yet” with no visible roadmap
- Shadow adoption fills the gap — developers try preview features on personal accounts or unmanaged tooling
- Enterprise loses the narrative of being AI-forward even when the intent and roadmap are there
- Rollout readiness work (allow-lists, registries, documentation) happens twice — once against preview behavior, again when GA changes
- Trust erodes when a feature is rolled out, then reverted when preview-era assumptions don’t survive GA
Naftiko Today
- Executable YAML capability specs abstract over upstream feature lifecycle — a capability definition can stay stable even as the underlying MCP spec or vendor surface evolves from preview to GA
- MCP exposure with Streamable HTTP and stdio transports gives the enterprise a stable integration point so developer-facing behavior does not have to change as upstream products graduate
- VS Code Extension (Fleet) provides a governed surface that the enterprise controls end-to-end, decoupling internal rollout timing from any one vendor’s GA calendar
- Spectral ruleset (15 rules) lets Maya define what “ready to ship internally” means independent of any vendor’s notion of GA
Naftiko Tomorrow
- Tool annotations for readOnly/destructive/idempotent (Second Alpha) would give Maya a way to ship capabilities with explicit maturity posture so preview-equivalent and GA-equivalent tools can live side by side under clear labels
- Naftiko Shipyard MVP (Fleet Second Alpha) would provide a staged publication model so capabilities can be piloted with a subset of teams before full allow-listing, mirroring the preview-to-GA pattern internally
- Enterprise security integration with Keycloak and OpenFGA (v1.1) would let Maya scope preview-equivalent capabilities to specific teams or cost centers without blocking the whole population
- Fabric capability discovery (v1.1) would surface maturity metadata at discovery time so developers see the expected posture before they adopt