Nordic APIs

When should you cache API calls at the edge? Today, Kristopher Sandoval breaks down edge, server-side, and client-side caching, and when each makes sense for API providers.

Nordic APIs Summit 2026 explored how API providers can prepare existing APIs, infrastructure, and governance practices for increasing consumption by AI agents. Here's what we learned.

With LLMs, non-determinism means the same input can produce different — and potentially unwanted — results. Here's how to add a dose of determinism when building AI agents and integrations.

Do you really need both an API gateway and an AI gateway? Today on the blog, Janet Wagner breaks down how the two differ and why most organizations building with LLMs will likely need both.

Do APIs need to be reinvented for AI agents? Not necessarily. One recurring takeaway from Nordic APIs Summit 2026 was to strengthen the foundations.

Nordic APIs Summit 2026 is a wrap! This year, our speakers explored many angles on how agentic AI is shifting the API experience. Here are some of the key takeaways.

What happens when an LLM has too much agency? Excessive permissions can turn prompt injection into real-world actions. The OWASP LLM Top 10 highlights why least privilege and human approval still matter.

Next week, we get IRL. Editor in Chief Bill Doerrfeld reflects on the importance of in-person events and why they still matter in a world saturated with online content, AI-generated information, and endless advice on what to do.

Only 5 days left till Nordic APIs Summit Time is running out! ⏰ Check the agenda and get your tickets last minute : https://nordicapis.com/events/platform-summit-2026/

The OWASP LLM Top 10 maps major LLM risks, from prompt injection and excessive agency to model theft. Today on the blog, J Simpson reviews them all and sees how developers can reduce the attack surface with familiar security controls.

MCP without connective tissue is just a list of endpoints. Here's how developer portals can evolve to support AI agents as a new consumer type while maintaining the governance, access management, and lifecycle controls APIs still need.

So much new infrastructure is emerging around AI. But how much of it is actually necessary? Rashmi Baberwal examines when API, AI, and MCP gateways need distinct runtime capabilities, and where shared governance and control make more sense.

Not to toot our own horn too much, but in-person tech conferences still matter! You take in a lot in person, face-to-face networking matters in the AI age, and getting hands-on with the technologies and their creators doesn't hurt either.

One week to go till Nordic APIs Summit, taking place in Stockholm on October 13–14 and unconference a day before. Bring your crew: 15% off two tickets, 25% off three or more. Grab your tickets: https://nordicapis.com/events/platform-summit-2026/

AI agents don't just call APIs differently — they introduce new questions around identity, permissions, security, and observability. Our latest article previews some of the challenges we'll be digging into at Nordic APIs Summit 2026.

Catering to agents doesn't mean reinventing the wheel. Arguably, it means leaning harder on the API standards and best practices we've stressed for years. Lorna Mitchell explains why in a new interview ahead of Nordic APIs Summit.

Why go to a conference when there's endless content to consume? Expert guidance, curated experiences, hallway discussions, serendipitous connections, and firsthand knowledge can still set in-person events apart.

An agent's lack of human intuition exposes weaknesses that developers may have learned to work around. That's part of the beauty of API standards, says Lorna Mitchell. Read our interview with her ahead of Nordic APIs Summit.

How does API architecture change when the consumer is an autonomous agent? Ahead of Nordic APIs Summit 2026, Art Anthony explores the shifts in authorization, governance, security, design, and observability that will shape many of the conversations at the event.

It's easy to get swept up in tech hype cycles. Faith Sodipe suggests platform teams start with a simpler question: "What problem are we actually trying to solve?" Read our interview ahead of his Nordic APIs Summit 2026 talk.

It's important to know when to use deterministic components versus probabilistic AI in software engineering. But where do you draw the line? Janet Wagner explores the differences and shares industry perspectives.

As AI agents become the new API consumer, how does that alter API architecture? Ahead of Nordic APIs Summit 2026, we look at how agentic consumption is evolving API architecture, security, identity, governance, and observability — and what to expect in Stockholm.

12 days to Stockholm! 🚀Oct 12: UnconferenceOct 13–14: Nordic APIs Summit What happens when AI agents call your APIs? Join us to explore. Bring your crew: 15% off 2 tickets, 25% off 3+. 🎟️ https://nordicapis.com/events/platform-summit-2026/

"The speed at which things are progressing is nerve wrecking," says Isabelle Mauny. As AI agents gain access to more APIs and tools, she says enterprises need stronger governance, security, and evaluation frameworks.

"If you're automating the wrong process, it's just going to do the wrong process more quickly." Claire Barrett explores the cultural mindset, judgment, and decision-making required to truly become AI-ready.

What will it take to build an AI-native developer portal? Machine-readable outputs, semantic metadata, agent-facing discovery mechanisms, and more. Kristopher Sandoval explores how portals must evolve for AI agent consumers.

"You're much more AI-ready if you've followed best practices around API design and documentation," says Lorna Mitchell. Today on the blog, Art Anthony talks with her ahead of Nordic APIs Summit about standards, agents, and good API design.

Mapping and measuring your API portfolio is a first step toward organizational coherence. Ahead of Nordic APIs Summit, Gareth Faull talks with Art Anthony about "business gardening" and aligning APIs with organizational intent.

Are enterprises repeating the same governance mistakes with AI agents that they made with APIs? Isabelle Mauny explores why organizations should prepare now for rapidly expanding agent and MCP ecosystems.

For over two decades, API design and architecture remained remarkably consistent. That's starting to change as we've radically recontextualized how APIs are used and who, or what, consumes them.

What are determinism and non-determinism, and when should you apply each in software engineering? Today on the blog, Janet Wagner breaks down the difference and why AI agents often need both.

What does it really mean to be AI-ready? We catch up with Claire Barrett on the mindset shift required for AI-enabled strategies and integration, and the dangers of overlooking human factors.

AI agents are forcing API design to evolve, but they're not the only factor. Changing security models, system dependencies, regulations, and new consumer types are reshaping traditional API design too. Here are six major shifts.

Data access and security are only becoming more important in the agentic AI age. Learn how If P&C Insurance used identity federation to manage partner access and protect sensitive data across a complex B2B environment.

What should API providers expect from the move to stateless MCP? Slightly larger payloads, new security considerations, and simpler scaling. Janet Wagner pulls together industry perspectives on what the shift means in practice.

Interestingly, MCP and traditional developer portals are complementary. One is the "phone book" of endpoints, while the other handles authentication, billing, monitoring, governance, and more. Here's why both matter in the AI agent age.

As AI agents and LLMs take on more autonomous, mission-critical roles, the "determinism gap" makes traditional APM insufficient. Adriano Mota outlines observability patterns for tracing, evaluation, cost, and security.

Sure, you *can* collect user data. But *should* you? Art Anthony explores what happens when every company becomes a data company and the protections needed to handle customer data responsibly.

Traditional APM can miss semantic failures, retrieval blind spots, runaway agents, and uncontrolled costs. Adriano Mota explores what AI observability requires in the age of agentic AI.

Today on the blog, WSO2's Isabelle Mauny explains why AI agent security can't be an afterthought. Existing API vulnerabilities, rogue MCP servers, privilege drift, and unchecked tool access all become more pressing as agent adoption grows.

Preparing APIs for AI agents requires more than exposing endpoints. Here are ten areas to review, from workflow documentation and traffic handling to rate limiting, error responses, access control, and MCP support.

Less than a month to go. Time to rally your API crew! 🚀 Nordic APIs Summit is back in Stockholm, October 13–14. Plus a Day 0 Unconference to get the ideas flowing. Bring the team: 15% off two tickets, 25% off three or more. https://nordicapis.com/events/platform-summit-2026/#register

Today on the blog: AI agents, LLMs, MCP, and regulations are changing the rules of API design. J Simpson looks at six key areas that have evolved since REST was initially devised, likely for good.

What's a token, what kinds are out there, and what purpose does each serve? Janet Wagner breaks down authentication and authorization tokens, AI and LLM tokens, crypto tokens, and other application and API tokens.

In the agentic AI era, maintaining a coherent API portfolio matters more than ever. London Stock Exchange Group's Gareth Faull shares ideas for aligning APIs with organizational goals and avoiding structural fragmentation.

What will multi-agent infrastructure require at scale? Caching, observability and tracing, runtime governance, identity, and inter-agent communication are all imperative, writes Kristopher Sandoval.

Authenticated API traffic can become a serious security risk when valid credentials mask malicious behavior. Here are five signs that API providers should watch for.

Good agent-ready API design isn't entirely new. Clear documentation, predictable behavior, explicit contracts, and strong security still matter — but now APIs also need to explain themselves directly to autonomous software.

Being AI-ready is as much psychology as technology, says Claire Barrett. Ahead of her talk at this year's Nordic APIs Summit, we catch up with Claire on what AI readiness actually means in practice.

API observability for AI agents goes beyond runtime heuristics. It requires assigning intent, correlating API workflows, tracking tool usage, and monitoring how agents interact with systems over time.