Community Spotlight

Cristina Flaschen is the CEO and Co-founder of Pandium, a code-first integration platform that helps B2B SaaS companies build and manage integrations at scale. Over the past fifteen years, she has worked across API strategy, partner ecosystems, implementation services, and integration architecture, giving her a front-row view into how integration requirements evolve as software ecosystems become more connected.

Historically, integration infrastructure was designed around human oversight. Developers interpreted documentation, investigated failures, and adapted to inconsistencies between specifications and production behavior. Integrations could succeed even when documentation was incomplete because engineers could fill in the gaps through experience, troubleshooting, and direct communication.

Agent-driven systems change that dynamic.

An AI agent cannot open a support ticket when an endpoint behaves unexpectedly. It cannot infer undocumented authentication requirements from previous experience. It operates on information exposed through API contracts, schemas, tool descriptions, and runtime behavior. When those elements are incomplete or inconsistent, failures occur.

This is why Flaschen argues that many of the challenges discussed around MCP and agentic AI are not entirely new. Instead, they expose weaknesses that already existed within API and integration ecosystems. Incomplete specifications, inconsistent authentication models, unclear error handling, and poorly documented workflows have long created friction for developers. Agent consumption simply makes those issues impossible to ignore. Rather than focusing on protocol adoption itself, she examines the underlying requirements that determine whether integrations remain reliable as new consumption models emerge.

The lesson for API teams is that agent readiness is becoming an integration requirement. Reliable machine-driven execution depends on explicit contracts, predictable behavior, and well-defined workflows. New protocols can improve how capabilities are discovered and consumed, but they cannot compensate for weak foundations underneath.

API Feed

Big Story

The challenge is that most enterprise APIs were not designed for this style of consumption.

Industry data already points to the scale of the problem. According to Postman's 2025 State of the API report, 93% of API teams report collaboration blockers, with inconsistent documentation cited by 55% of respondents and inconsistent API definitions by 43%. This shift is beginning to change how platform teams think about API architecture. The question is no longer simply whether an endpoint returns the correct response. It is whether APIs provide enough context, structure, and behavioral consistency for reasoning systems to make reliable decisions across complex workflows.

This is the foundation of what many teams are beginning to describe as the cognitive API stack. At the base remain the APIs and systems that execute business operations. Above them sit protocols such as MCP and other emerging coordination layers that expose those capabilities to agents. Above that sits the reasoning layer itself: planners, orchestrators, and AI systems that decide which tools to use, in what sequence, and under which conditions.

As reasoning systems become part of enterprise workflows, each layer places new requirements on the layer below it. APIs are no longer serving only applications. They are serving systems that must interpret, reason about, and act on the capabilities they expose.

The problem is that most enterprise API estates were designed for the developer-experience model, well-documented enough for a motivated engineer to integrate, consistent enough for major workflows, and flexible enough to handle edge cases through manual intervention.

An agent operating through MCP cannot message the integration team when something is unclear. It either succeeds, retries, or fails. If it fails repeatedly, the resulting issues may not become apparent until an entire downstream workflow breaks.

The organizations that navigate this transition effectively treat agent readiness as a design criterion. That means schema completeness, not only covering the happy path but also documenting edge cases, error conditions, and boundary behaviors in machine-readable form. It means authentication models that support agent identities as well as human users. It means rate-limit semantics designed for sustained programmatic consumption.

It also means expanding observability beyond individual API calls. Platform teams need visibility into workflow completion rates, dependency chains, retry behavior, and cross-service execution patterns. Understanding whether an API responded successfully is useful. Understanding whether a multi-step workflow produced the intended business outcome is becoming more important.

The protocols themselves, including MCP, are not shortcuts around this work. They are interface layers. An MCP server that wraps a poorly specified or inconsistently behaving API does not automatically become agent-ready. The quality of the cognitive API stack depends on the quality of the APIs beneath it, and the tooling emerging around agent consumption is making those dependencies far more visible.

That is what makes the cognitive API stack important. It reframes APIs as part of the enterprise's reasoning infrastructure. The teams that move fastest will be the ones that clean up the contracts, workflows, permissions, failure modes, and observability beneath those interfaces. Agent adoption will make weak integration foundations visible. API teams that treat that visibility as a design mandate will be better positioned for the next phase of machine-driven software consumption.

Resources & Events

apidays Amsterdam brings together API practitioners, architects, and platform leaders for two days focused on operating APIs as products in enterprise environments. The 2026 program includes sessions on API design, lifecycle management, governance, and platform strategy, as well as discussions on how APIs are evolving alongside AI-driven systems and automation. The event is designed for teams building and scaling API programs across organizations, with a mix of technical sessions and practitioner-led case studies. Details →

apidays Munich brings together API architects, platform engineers, and enterprise technology leaders for two days focused on API strategy, platform operations, and AI-driven enterprise systems. The 2026 program includes sessions on API lifecycle management, developer experience, event-driven architectures, and the operational challenges of managing APIs across distributed environments. Designed for teams scaling enterprise API programs, the event combines technical discussions, practitioner-led case studies, and platform engineering perspectives on how APIs are evolving alongside automation and AI-connected workflows. Details →

You can find a list of all Apidays events here

Apply to speak at Apidays Singapore, NY, London, Paris, and more here

Databricks' annual Data + AI Summit has become one of the largest gatherings of data engineers, AI practitioners, architects, and enterprise technology leaders. The 2026 event will explore how organizations are building data platforms that support AI applications, agents, analytics, and operational workloads on a common foundation. Sessions span data engineering, governance, AI infrastructure, lakehouse architecture, model operations, and agent development. For API leaders, the conference offers useful insights into how data systems, AI platforms, and application interfaces are converging as organizations move toward production-scale AI deployment. Details →

Based on a survey of 327 security professionals, this report finds that 92% of organizations lack the advanced security maturity needed to defend agentic environments, and two-thirds (66%) reported API growth of over 50% in the past year. Notably, nearly all attack attempts now originate from authenticated sources, and the report argues API security has become a foundational discipline in its own right rather than a subset of application or cloud security. Read →

Insight of the Week

API Evangelist scored 83 out of 100 and reached Level 5 in Cloudflare’s isitagentready.com scan, which checks whether sites expose the signals agents need to discover and use web resources. The site received full marks for discoverability, markdown content negotiation, and bot access controls. The remaining gaps were that OAuth discovery was not relevant to a public static site, and that Web Bot Auth and Agent Skills formatting needed cleanup. The audit covered well-known paths, RFC 8288 Link headers, API catalogs, MCP server cards, markdown responses, and access rules.

For the Commute

Drawing on leadership roles at Microsoft, Amazon, and Thoughtworks, Natalie Hollier talks about the realities of scaling sustainability programs within large technology organizations. Topics include change management, platform adoption, supplier engagement, AI's role in sustainability initiatives, and the challenges of turning long-term strategic goals into measurable operational outcomes. While focused on sustainability, the lessons on organizational transformation and platform-scale execution are relevant to technology leaders across industries.

Stay tuned for bold ideas, fresh perspectives, and the next wave of API innovation

-The Apidays Team