Two Days at DASH 2026: The Shift, the Signal, and Where Eden Stands

Last week, the team spent two days at Datadog DASH in New York. DASH is Datadog's flagship annual conference, and this year it brought together engineers, platform leaders, security teams, and founders from across the industry to talk about how observability, security, and AI are converging in production (not to mention an appearance from Adam Scott).

We came in with a clear question: when AI moves from research budgets into production infrastructure, what does the platform engineering stack look like?

We walked out with hundreds of conversations, a notebook full of follow-ups, and a much sharper view of where the category is heading.

This is what we saw.

What DASH Is, and Why This Year Was Different

DASH has always been Datadog's primary platform. It is where the company sets its product direction for the year, brings customers on stage to share how they run things, and gives the partner ecosystem a place to show how it connects to the platform.

This year the center of gravity shifted decisively toward AI. Not as a feature, but as the through-line for the entire keynote.

Datadog announced Bits AI across the stack: Bits Detection for autonomous monitor coverage, Bits Memories for retaining team operational knowledge, Bits Remediation for closing the loop from investigation to fix, Bits Code as a platform-wide coding agent, Bits Release for AI-driven release validation, Bits Testing for synthetic test generation, and Bits Data Analysis for natural-language business questions. They also announced AI Guard for protecting agents against prompt injection and tool misuse, Agent Console for tracking coding agent usage across an organization, Patterns and Bits Evals for understanding production LLM behavior, and Journey Monitoring for unified user-flow visibility.

The pattern is clear. Observability is moving from passive (what happened) to active (what to do about it). Datadog is building agents that investigate, validate, and remediate alongside engineering teams.

That same pattern showed up on the floor.

The Buyer for AI Has Moved

Two years ago, AI sat with research teams or innovation budgets. At DASH, the buyer was unambiguous. AI is now a platform engineering problem.

The questions we heard at the booth were not about model quality. They were about cost attribution, audit trails, routing, fallback logic, credential isolation, and per-tenant rate limiting. These are the same shape of problems platform teams have spent the last decade solving for databases and APIs. They are now being asked about model calls, agent actions, and tool use.

That shift matters because it changes who is in the buying conversation. The people walking up to our booth were platform leads, infrastructure architects, and engineering directors. The conversations were about how to govern AI traffic in production, not how to evaluate a model.

What We Learned

A few specific takeaways from the two days.

Governance is procurement-grade now. Conversations with fintech, healthcare, and insurance teams kept landing on the same point: their security and compliance organizations are starting to ask hard questions about AI infrastructure. Audit per call, credential isolation, sanctioned model lists, and per-tenant data residency are moving from nice-to-have to required. The teams ahead of this curve are buying it. The teams behind it are about to be told to.

Open source is the wedge. We open-sourced Eden under Apache 2.0 two days before DASH, and the response on the floor was sharper than any pitch we could have written. Engineers want to read the code. Platform leads want to evaluate it without a procurement cycle. Companies want the option to run it themselves in their own VPC. Releasing Eden in the open turned a lot of cold conversations into warm ones.

Consolidation is the conversation at scale. At every large bank, payer, and platform we talked to, the platform team is fighting fragmentation more than building anything new. Different business units have different AI infrastructures. Different teams have different observability. Different agents have different governances. The work for the next two years is consolidating what already exists, not adding more.

The AI-native companies are setting the bar. Companies like EliseAI, Harvey, AlphaSense, Daloopa, and Suno are operating at a sophistication level that makes the bar for what production AI looks like much higher than it was a year ago. They are the ones asking the sharpest questions about agent governance, audit, and cost attribution. They are also the ones most likely to ship the patterns the enterprise will adopt next.

The ecosystem story is its own conversation. Some of the sharpest signal from DASH was how the partner companies on the floor are shaping the rest of the stack around AI as production infrastructure. That story deserves its own write-up, which is coming next.

Where We Are

A quick state of Eden as of this week.

Eden is open source. Released under Apache 2.0 on June 11. The repository includes the Rust gateway and service runtime, endpoint schemas for databases and AI providers, shared auth, telemetry, analytics, and the control-plane primitives. Redis is the production-ready gateway protocol in this release. Postgres, Mongo, LLM, and agent gateway surfaces are included for development, evaluation, and community contribution as they continue moving toward production.

Eve is in production. The Rust data plane is sitting on critical paths today. In our Redis benchmark set, Eve matched a 400k requests per second workload with lower tail latency and lower CPU cost per completed request than Envoy's Redis proxy. The AI gateway path reached 12k requests per second for buffered 64 KiB OpenAI-compatible responses with zero errors.

Adam is in private preview. The AI-powered conversational interface for enterprises, built on zero knowledge, least-privilege access, and PII redaction by default. Early access is open.

Exodus is live on the Azure Marketplace. The ACR-to-AMR migration product, bundled with Datadog observability, available for one-click deployment. Customers can scope the effort, estimate savings, and run the migration without signing new agreements.

What's Next

A few things we are focused on through the rest of the year.

Production readiness for more gateway protocols. Postgres and Mongo are moving toward production status. LLM and agent gateway surfaces are getting hardened against real customer workloads.

Deeper integration with the partner ecosystem. More on this in the next post. Platform teams should be able to compose Eden with the rest of their stack without writing glue code.

Continued investment in the open-source community. The repository is the front door now. We are watching issues, taking contributions, and using community feedback to shape what becomes production next.

How to Get Started

The open-source repository is at github.com/eden-dev-inc/eden, released under Apache 2.0. You can run it, inspect it, modify it, and contribute. Documentation, benchmarks, and deployment guides are linked from the README.

If you are migrating off Azure Cache for Redis, Eden is on the Azure Marketplace today, bundled with Datadog observability. The migration that traditionally takes two to three months can be completed in a day.

If you are working on AI infrastructure governance more broadly and want to compare notes, reach out.