Platformatory
Consulting
As sprawl grows, governance turns pejorative—unless a few key platforms pave well-governed self-service.
Platforms
A few key platforms. Specialized enough to be safe. Self-service enough to be fast.
The traffic plane is where cross-cutting concerns have to be solved—auth, policy, tenancy, audit, rate limits, and identity for every caller. North–south and east–west are the same plane with different trust boundaries: edge/API management and service mesh are one story, not two product silos. That was always true for APIs and service-to-service calls; it is urgent now with the explosion of MCPs and agentic sprawl, where every agent principal is another consumer on the plane.
Web-scale traffic has escalated to botnet-scale—with agents in the mix. The observability platform has to take standardized telemetry across servers, services, and clients seriously, without breaking the bank. That pressure creates a specialized data domain of its own: high-cardinality, high-churn operational signal that must stay queryable, attributable, and cost-bounded while agentic operations act on it.
Almost nobody is coding only by hand anymore. Token-maxxing is the norm. The principles that govern internal development have shifted—code is no longer the scarce barrier—yet coding agents still have to be tamed to enterprise guardrails, domain models, and standards. The IDP has to solve the interplay between domains, context boundaries, runtimes, runbooks, and the humans who remain accountable for the software.
How we think about it
As enterprise infrastructure and sprawl grow, “governance” quickly becomes a pejorative. Two failure modes show up: team-level autonomy goes out the window, or build-it-run-it teams make increasingly siloed decisions—often outside their circle of competence—and open up every kind of risk.
Building a few key platforms paves a middle path: a tribe focused on enabling well-governed self-service, with deep specialization in a handful of system domains. That is where the traffic plane (API management + mesh), observability & agentic operations, and an internal developer platform earn their keep.
Start with the traffic plane. Cross-cutting—policy, identity, tenancy, audit—does not belong in every app and every agent. North–south (edge / API management) and east–west (service mesh) are the same plane with a different trust boundary—not two disconnected stacks. With MCP surfaces and agentic sprawl multiplying consumers overnight, that plane is where those concerns have to land, once, and hold.
The commitment is simple to say and hard to operate: self-service everything, with the platform carrying the governance that used to live in committees and ticket queues—including when the consumers and the producers are agents.
Next step
Tell us where autonomy is blocked—or where siloed decisions are creating risk. We will map which platform to pave first.
