The reason AI agents shouldn’t touch your source code — and what they should do instead

Key Concepts

• Observability: The ability to understand the internal state of a system by examining its outputs (logs, metrics, traces).
• Feature Flags (Feature Toggles): A software development technique used to turn certain features on or off without deploying new code.
• Progressive Delivery: Releasing software incrementally to a subset of users before a full rollout.
• AI Observability: Applying AI and machine learning to analyze observability data for insights and automated actions.
• Autonomous Agents: AI systems capable of taking actions in an environment with minimal human intervention.
• Open Feature: A CNCF project aiming to standardize feature flagging across different platforms.
• Agentic Workflows: Workflows driven by AI agents performing tasks and making decisions.
• DevSecOps: Integrating security practices into the DevOps process.
• Runbooks: Predefined procedures for responding to specific incidents or issues.

Dino Trace & Dev Cycle: Redefining Developer Experience Through Unified Observability & Feature Management

Introduction

This discussion centers on Dino Trace’s acquisition of Dev Cycle and the implications for the future of developer experience, observability, and software delivery. The conversation highlights a shift towards unified platforms encompassing logs, metrics, traces, AI, security, and infrastructure, with a strong emphasis on integrating feature management directly into the developer workflow and leveraging AI-powered automation.

I. The Evolution of Feature Management & the Birth of Open Feature

In 2020, Dino Trace recognized a fragmented feature flag market. Vendors were building similar SDKs, creating onboarding challenges for developers switching between solutions. This led to the creation of the Open Feature project (a CNCF project), aiming to establish a standardized set of SDKs for feature flagging. Dino Trace internally adopted Open Feature, which ultimately led to their engagement with Dev Cycle. Dev Cycle built a product on top of Open Feature, and the two companies collaborated on joint customer projects, making integration a natural progression. (“We started to work on a lot of joint customers. So it became for us at some point like very natural. So to integrate them into the stack.”)

II. The Convergence of Observability and Feature Management

The acquisition is driven by the increasing intertwining of observability and feature management. While feature flags define what features are available, observability is crucial for understanding how those features are performing. Customers struggled to determine if feature flag combinations were working effectively and where flags were being used across their entire stack. (“The biggest challenge for most customers was yeah have all those features like defined but how are they working? How they’re how’s the combination working or not working? where are certain feature flags actually used throughout my entire stack.”) This necessitates a unified platform. DataDog’s acquisition of Apple LaunchDarkly exemplifies this trend, with observability vendors recognizing the need to incorporate feature management capabilities.

III. Enabling Autonomous Operations with Feature Management

Dino Trace is moving towards “Dino Trace Intelligence,” utilizing autonomous agents to take action within customer environments. The safest and quickest way to address issues is often through configuration changes, making feature management a critical component. (“…the safest change to an environment is not rewriting the code and deploying it by an agent but changing the configuration.”) This allows for rapid mitigation of problems and enables a more proactive, automated approach to system management. Feature management is also essential for experimentation, particularly in AI-native applications.

IV. AI-Native Applications & Experimentation in Production

The rise of AI-native applications necessitates more experimentation in production. Testing prompts and model combinations in pre-production environments is often insufficient. Customers are requesting the ability to modify prompts and run combinations directly in production, requiring observability to track the impact of these changes. (“…it's very hard to test this in a pre-production setting you have your test data but you need to do it in production.”) Dino Trace’s AI observability solution is designed to facilitate this process, allowing users to identify optimal prompts and configurations.

V. Acquisition Details & Integration Roadmap

The acquisition of Dev Cycle closed two weeks prior to the interview. The immediate focus is ensuring a seamless continuation of the Dev Cycle business for existing customers. Initial integration steps include allowing Dino Trace users to log into Dev Cycle with their premium tier access. A product integration already existed, allowing users to leverage Dino Trace observability to optimize feature flags and targeting rules. The roadmap involves building a fully integrated solution throughout the year, with native feature management capabilities within Dino Trace, initially in preview in the second half of the year. Dev Cycle’s existing remote server capabilities for adding feature flags on-the-fly will be further integrated, particularly in the context of AI-assisted coding.

VI. Why Dev Cycle? A Developer-Centric Approach

Dino Trace chose Dev Cycle due to its strong focus on the developer workflow. The company differentiates itself from other feature flag vendors that cater primarily to experimentation (marketing-focused) or SREs. (“…we went with dev cycle because obviously as a feature flag is most important for the developers as they're releasing and writing code.”) The Dev Cycle team demonstrated a seamless integration into the developer’s coding environment, allowing feature flags to be added directly within the IDE. (“…the first demo I remember they showed us off the product was really somebody sitting in there white coding and the feature flag getting added there and we said this is it.”) The shared history and collaboration through the Open Feature project also played a significant role.

VII. AI, Security, and the Changing Landscape of Responsibility

Feature flags serve as a safety mechanism for AI-driven automation, allowing for quick rollbacks and configuration changes if issues arise. While full automation of production environments isn’t imminent, AI can accelerate decision-making and actions within defined guardrails. The conversation also touches on the increasing frequency and decreasing cost of security exploits due to the accessibility of AI-powered attack tools. Observability is crucial for informing AI agents about the real-world impact of their actions and for detecting and responding to security threats. (“…observability especially in this agentic work becoming more or less the eyes and ears of the agents.”)

VIII. Dino Trace’s Vision: Shipping Reliable Software

Dino Trace’s overarching goal is to enable customers to “ship software that works perfectly.” The company views observability, security, and automation as interconnected components of this vision. They emphasize the need to move beyond traditional DevOps/DevSecOps terminology towards a focus on “shipping reliable software.” The ability to detect vulnerabilities and prioritize remediation based on observability data is a key differentiator. (“If this is a backend service that nobody ever touches, it's different.”) Dino Trace can even generate instructions for coding agents to automatically fix vulnerabilities.

IX. Acquisition Strategy: Build vs. Buy

Dino Trace employs a dual strategy of internal development and strategic acquisitions. While they invest in building core capabilities internally (like Dino Trace Intelligence), they acquire companies to accelerate innovation and gain access to specialized expertise. (“You're investing in certain areas. You're pushing your product forward in certain areas. But then you realize you you see this new opportunity in the market.”) The acquisition process prioritizes team fit and cultural alignment, often involving collaborative coding sessions during due diligence. (“We bring them actually to a lind lab. We sit down and we do a product planning sessions.”)

Conclusion

Dino Trace’s acquisition of Dev Cycle signifies a major step towards a unified platform for observability and feature management. The company is positioning itself to empower developers with AI-driven automation, enabling faster, safer, and more reliable software delivery. The emphasis on developer experience, coupled with a proactive approach to security and a commitment to open standards (Open Feature), suggests a compelling vision for the future of software development. The core takeaway is a move towards a more active, intelligent observability that not only monitors systems but also acts on insights to ensure software reliability and performance.