May 2026 Summaries

Dynatrace is a popular full-stack observability and application performance monitoring platform for large-scale enterprise environments, integrating various monitoring capabilities and AI-assisted root-cause analysis. However, changing observability needs, such as real-time debugging and predictable pricing, have led engineering teams to explore alternatives. Competitors like Honeycomb, Datadog, Grafana, New Relic, Splunk, Elastic, and SigNoz offer varied strengths, such as Honeycomb's high-cardinality debugging and transparent pricing, Grafana's open-source flexibility, and Datadog's extensive integrations. These alternatives address the constraints of proprietary instrumentation and complex pricing models associated with Dynatrace. Organizations now prioritize OpenTelemetry compatibility, transparent pricing, and flexibility in deployment and analysis, with Honeycomb noted for its developer-centric observability and rapid debugging capabilities. While Dynatrace is favored for automation and governance in large enterprises, alternatives offer enhanced real-time insights and reduced operational complexity, providing diverse options based on organizational needs and priorities.

O11yCon 2026 showcased the transformative impact of agent-based software development, emphasizing its inevitability and the challenges it poses in terms of performance and cost efficiency. A series of talks highlighted the shift in software creation dynamics, with discussions on AI's varying outcomes in different team environments and the bottlenecks in code review processes. Honeycomb introduced innovative tools like Canvas and Agent Timelines, enhancing collaboration and visualization in agent interactions. The conference also addressed the blurring roles of software engineers and SREs, the evolving nature of code reviews, and the necessity of robust observability practices. Speakers underscored the importance of trusted knowledge bases, the redefined roles of security as facilitators rather than gatekeepers, and the need for behavioral observability in AI-driven environments. The event concluded with insights on how traditional metrics and logs are becoming obsolete in favor of trace documentation, urging organizations to adapt swiftly to these evolving paradigms.

Cloudelligent, an AWS Premier Tier Services Partner, and Honeycomb.io, an observability platform, have announced a strategic partnership aimed at assisting engineering organizations in modernizing their operations on AWS and achieving enhanced production visibility. As AI-driven development speeds up release cycles, traditional monitoring tools struggle with increasing system complexity, creating a need for quicker learning from production environments. Through this collaboration, Cloudelligent becomes a certified partner of Honeycomb, offering delivery, managed services, and customer engagement across the Honeycomb observability journey on AWS, while aligning with AWS Well-Architected standards. This partnership enables joint customers to modernize legacy environments, adopt cloud-native and Kubernetes architectures, and responsibly integrate AI and ML workloads, ensuring robust observability infrastructure. By combining Cloudelligent's expertise in AWS modernization with Honeycomb's event-based telemetry and high-cardinality query engine, engineering teams gain better visibility into distributed systems, reducing operational noise and enhancing reliability, which shortens incident resolution and development feedback cycles. This alignment helps transform engineering velocity into meaningful business outcomes, offering a competitive edge in the rapidly evolving software delivery landscape.

Canvas, a major update to Honeycomb's investigation workspace, transforms from an AI co-pilot into a collaborative platform where entire teams, including AI agents, can address production issues collectively. This evolution was driven by the increasing complexity of investigations due to AI-driven development, where failures often occur without clear errors or rollbacks and require immediate, coordinated responses. Canvas facilitates these investigations by instantly triggering auto-investigations upon detecting anomalies, allowing findings, queries, charts, and recommendations to be shared and updated in real time. Built on Honeycomb's event-based telemetry, Canvas allows multiple users and AI agents to collaborate on the same interface, leveraging pre-built and custom skills to follow complex investigation threads. The platform aims to streamline the investigation process by integrating human and AI expertise, offering a dynamic and interactive environment where the team's collective knowledge is readily accessible and actionable.

Agent Timeline is a new observability tool designed to enhance debugging of AI agent workflows in production by providing a conversation-first view of agent activities, which is likened to a flight recorder for AI agents. This tool addresses the challenges faced by current observability and application performance monitoring tools that stop at the model level and lack the ability to connect AI-layer visibility with full-stack observability. It organizes telemetry data around a conversation, capturing events such as LLM calls, tool invocations, retries, and failures, allowing users to easily trace and troubleshoot issues across distributed systems. Built on Honeycomb's high-cardinality, event-based architecture, Agent Timeline supports OpenTelemetry GenAI Semantic Conventions, enabling easy integration and ensuring that sensitive data is handled safely. By highlighting failures and providing comprehensive traceability from the AI layer to backend systems, Agent Timeline aims to simplify the debugging process and improve incident response times for enterprises running AI agents in production.

Honeycomb and Embrace have partnered to address the complex challenge of observability in modern application monitoring by seamlessly integrating frontend and backend data using OpenTelemetry as a common foundation. This integration resolves the issue of fragmented data by providing a unified view that connects frontend user experiences with backend service traces, eliminating the need for engineering teams to juggle multiple tools and manually correlate data. Embrace captures rich, structured telemetry from web and mobile applications, which flows into Honeycomb's platform, allowing teams to track trends and trace specific user experiences with context-rich data that includes metrics, network spans, and trace contexts. This collaboration enables a comprehensive understanding of application performance and user interactions, facilitating proactive issue detection and precise root cause analysis.

Honeycomb has achieved the Amazon Web Services (AWS) Financial Services Competency, highlighting its expertise in providing observability solutions for financial services organizations operating in high-stakes, regulated environments. This accomplishment underscores Honeycomb's ability to address the unique challenges faced by financial platforms, such as strict compliance requirements and the need for real-time, high-cardinality telemetry to effectively manage and understand complex transaction flows. By leveraging AWS services and OpenTelemetry-based distributed tracing, Honeycomb has enabled companies like Nubank, Salary Finance, and Moov to enhance their system reliability, reduce mean time to resolution, and ensure compliance-grade audit visibility. The competency emphasizes Honeycomb's commitment to helping teams achieve faster issue detection and maintain operational excellence in production systems while integrating seamlessly with AWS's foundational services.

Honeycomb.io and Embrace have announced an extended partnership to integrate their observability platforms, enhancing the ability of engineering and reliability teams to gain insights into both backend and frontend system health. This collaboration leverages Honeycomb's expertise in foundational observability and Embrace's focus on user experience, using OpenTelemetry to provide a unified data framework that captures high-cardinality, context-rich data. The integration enables teams to see a comprehensive picture of system performance, correlating technical signals with user impacts across web and mobile platforms, thus addressing longstanding challenges of visibility and collaboration between backend and frontend teams. By merging real user monitoring capabilities with backend observability, the partnership aims to deliver better user experiences and improved business outcomes, allowing joint customers to access and deploy the solutions flexibly via the AWS Marketplace.

AI is significantly transforming the Software Development Life Cycle (SDLC) by enabling faster code shipping with reduced human oversight and introducing non-deterministic behaviors in production environments through agents and large language models (LLMs). This shift has led to a compression of the traditional multi-stage development process into rapid cycles of intent and validation, often occurring in live settings rather than pre-production stages. Honeycomb addresses these changes with two key approaches: observability for AI, which enhances visibility into agent and LLM workflows in production, and observability with AI, which utilizes advanced tools to tackle complex problems quickly. The introduction of Agent Timeline provides a comprehensive view of agent workflows, while deeper LLM insights incorporate prompts and quality scores as essential telemetry. The Canvas tool facilitates collaborative and automated investigations when alerts are triggered, incorporating team knowledge through custom skills to streamline problem-solving. These innovations reflect a broader trend towards integrating AI capabilities into observability practices, allowing teams to respond more effectively to the unique challenges posed by AI-driven software environments.

Honeycomb.io has introduced new observability features specifically designed for AI agents in production, addressing the challenges posed by non-deterministic, multi-hop agent workflows. These features include Agent Timeline, Canvas Agent, and Canvas Skills, which enable engineering teams to track AI agent performance, behavior, and decisions in real-time without needing proprietary SDKs. The platform aims to mitigate uncertainty in AI systems by providing a coherent view of agent activities and enabling seamless investigation and collaboration through its Canvas tool. By integrating OpenTelemetry standards, Honeycomb ensures flexibility and adaptability for AI-generated telemetry data. The company will showcase these advancements during a three-day digital event, Innovation Week, followed by their flagship conference, O11yCon, in San Francisco.

The traditional software development lifecycle, characterized by distinct stages such as development and operations, is undergoing a rapid transformation with the integration of AI, leading to compressed cycles of intent and validation. The old model, which viewed production as the final stage, is being dismantled by AI's non-deterministic nature, necessitating real-time validation during production. This shift emphasizes the importance of bridging the gap between development and operations to prevent loss of value through recurring issues. Observability, particularly in the context of AI-driven systems, is becoming crucial, as traditional methods fall short in capturing the unpredictable nature of AI agents. New observability practices, like Honeycomb's Agent Timeline, aim to provide comprehensive visibility into AI operations by integrating real-time data and facilitating collaborative investigations. Organizations that prioritize observability from the start, such as those using Mixpanel and Intercom's systems, are better equipped to handle the complexities introduced by AI, ensuring efficient problem-solving and system reliability.

OpenTelemetry Weaver is a toolset designed to enhance the clarity and utility of telemetry data by using semantic conventions to define, validate, and evolve telemetry schemas, treating them like public APIs. This approach allows for the creation of a layered schema registry, aiding both human analysts and AI agents in understanding the attributes within a dataset. At Honeycomb, a three-layer registry is employed: the OpenTelemetry base layer for standard attributes, a Honeycomb-specific overlay for platform attributes, and an optional customer domain overlay for organization-specific attributes. This layered approach ensures that each attribute's meaning is documented and accessible, facilitating informed data analysis and decision-making. Weaver also supports schema dependencies, automatic column descriptions, and tools for deeper exploration, while offering capabilities such as live telemetry validation, type-safe SDK generation, and schema version diffing. Through these features, Weaver aims to make telemetry data more comprehensible and actionable, ultimately improving observability and analytics processes.

OpenTelemetry tracing offers a powerful way to enhance visibility and correlation in distributed systems by efficiently adding attributes to spans, which are units of work, to encapsulate critical event data. Attributes, such as request parameters and feature flag values, provide a cost-effective method for enriching trace information, particularly when using providers like Honeycomb, which charge per event. While spans are preferred for marking significant operations, especially across network or asynchronous boundaries, logs can still be useful for capturing fixed-time events or exceptions without overwriting existing attributes. The strategy for structuring traces, whether by adding attributes or creating new spans, should align with the desired level of detail and correlation needed for observability, allowing developers to effectively tell the story of their application's performance and behavior.

Grafana is a popular open-source visualization and monitoring tool, widely used in the observability space, but it often requires managing multiple backends and plugins, leading to operational complexity as systems grow. As an alternative, several platforms offer more unified solutions, each with different strengths, such as Honeycomb's single datastore for logs, metrics, and traces, and Datadog's fully managed platform with extensive integrations. These alternatives prioritize features like OpenTelemetry compatibility, high-cardinality data handling, and transparent pricing. Honeycomb, for instance, supports high-cardinality data and exploratory querying, eliminating the need to manually correlate data across multiple tools, while Datadog offers out-of-the-box dashboards and AI-driven anomaly detection. Other platforms like Dynatrace and New Relic provide strong automation and APM capabilities but often come with higher costs and complexity. The document emphasizes the importance of choosing a platform that supports future-proofing through vendor-neutral instrumentation, unified signal coverage, and robust debugging capabilities, particularly in complex, distributed systems.

The text introduces the concept of log clustering as a solution to managing vast volumes of log data, which often obscure critical errors and anomalies due to noise from routine processes like health checks and connection messages. Log clustering, particularly using the Drain algorithm, groups log lines by structural similarity to derive templates that capture fixed message parts, allowing teams to focus on broader log classes instead of individual lines. This approach is incorporated into the OpenTelemetry Collector through the drain processor, which annotates logs with templates rather than filtering them, enabling integration with other processors for filtering, deduplication, or routing. The drain processor offers configuration options like merge_threshold and tree_depth, impacting the specificity and memory usage of templates, and supports seeding templates for consistent pattern recognition across multiple instances. The processor is currently at alpha stability and aims to improve log observability by allowing users to identify and manage noisy patterns effectively, reducing log volume while maintaining essential visibility into system operations.