← Back to all announcements
★★★★★ 20/07/2026

Amazon CloudWatch announces coding agent insights

Track AI coding agent ROI, token spend, and PR velocity across teams—all in CloudWatch, with no extra instrumentation needed.

View original announcement →

Visual Summary

graph TD A{{CloudWatch Coding Agent Insights}}:::announced B((Developers)):::external C([Claude Apps Gateway]):::feature D(Claude Code):::compute E(OpenAI Codex):::compute F(GitHub Copilot):::compute G([OpenTelemetry Metrics]):::feature H(Amazon CloudWatch):::compute I([Pre-built Dashboards]):::feature J([Token Billing Alerts]):::feature B -->|"use"| D B -->|"use"| E B -->|"use"| F D ==>|"routes via"| C C ==>|"emits OTLP"| G E -->|"emits OTLP"| G F -->|"emits OTLP"| G G ==>|"ingests"| H H ==>|"powers"| A A -->|"displays"| I A -.->|"triggers"| J classDef announced fill:#ff9900,stroke:#ec7211,color:#fff,font-weight:bold classDef compute fill:#e3f2fd,stroke:#1565c0,color:#1565c0 classDef storage fill:#e8f5e9,stroke:#2e7d32,color:#2e7d32 classDef feature fill:#fff3e0,stroke:#e65100,color:#e65100 classDef external fill:#f5f5f5,stroke:#616161,color:#616161

What's New

Amazon CloudWatch has launched Coding Agent Insights, a new capability that gives engineering leaders unified visibility into how AI coding tools—including Claude Code, OpenAI Codex, and GitHub Copilot—are driving value across their organizations. The feature ingests OpenTelemetry metrics emitted by these agents and surfaces them in pre-built CloudWatch dashboards alongside existing operational data, requiring no additional instrumentation when using the Claude apps gateway for AWS. Engineering teams can now track token spend, correlate agent adoption with software delivery metrics like commit throughput and pull request velocity, and optimize model selection by cost-to-output ratio.

How It Works

  • OTel-native ingestion: Coding agents emit OpenTelemetry (OTel) metrics—covering token consumption, per-turn latency, tool calls, API requests, and approvals—which are sent to CloudWatch's native OTLP metrics endpoint without requiring custom instrumentation.
  • Claude apps gateway integration: The Claude apps gateway for AWS acts as a self-hosted control plane that sits between developers' Claude Code clients and the model provider, automatically relaying OTLP telemetry to CloudWatch (or other collectors) as part of every request.
  • Identity and org attribute enrichment: Identity and organizational attributes such as user.email, team.id, department, cost_center, and organization must be sent as OTel resource attributes; the gateway handles this automatically for Claude Code users.
  • Pre-built dashboards: Coding Agent Insights dashboards appear automatically under the GenAI Observability section in the CloudWatch console and populate as soon as agents begin sending metrics in the expected shape—no dashboard creation or import is needed.
  • Multi-agent support: Beyond Claude Code (via the gateway), OpenAI Codex and GitHub Copilot are also supported, each with their own setup guide for configuring OTel emission to CloudWatch using bearer tokens or enterprise SSO paths.
  • Slicing and dicing: Dashboards allow filtering and comparison across organizational dimensions (organization, department, team, cost center, user), and underlying data can be exported as CSV for further analysis.
  • Alerting: Proactive token billing alerts can be configured using standard CloudWatch alarms on the ingested OTel metrics.

Why It's Important

  • ROI visibility at scale: As AI coding agent adoption grows, organizations have lacked a standardized way to measure return on investment; Coding Agent Insights directly addresses this gap by correlating agent usage with engineering delivery metrics.
  • Cost governance: Engineering leaders can right-size token budgets by department or team, identify overspend, and compare cost-to-output ratios across different models—turning AI tooling from an opaque cost center into a manageable line item.
  • Accelerates adoption decisions: By showing which teams are benefiting most from AI coding agents, the feature provides data-driven justification for expanding or restricting access, removing guesswork from rollout planning.
  • Unified observability: Presenting AI coding agent telemetry alongside existing CloudWatch operational data (infrastructure, application performance) enables teams to draw correlations—e.g., whether increased agent usage coincides with faster deployments or fewer incidents.
  • Reduced instrumentation burden: The zero-instrumentation path via the Claude apps gateway lowers the barrier to collecting meaningful telemetry, making it practical for organizations without dedicated observability engineering resources.
  • Enterprise governance alignment: Integration with corporate IdP (via OIDC/SSO) means usage data is automatically attributed to real organizational identities, supporting chargeback, compliance, and audit requirements.

How It's Different

  • No custom instrumentation required (for Claude Code): Unlike generic OTel pipelines that require manual SDK integration, the Claude apps gateway automatically emits telemetry in the exact metric and attribute shape the dashboards expect, eliminating setup friction.
  • Pre-built, auto-populating dashboards: Rather than requiring teams to build their own dashboards from raw metrics, Coding Agent Insights provides ready-to-use views that appear automatically in the CloudWatch console as soon as data flows in.
  • Correlated with engineering delivery metrics: Most AI cost dashboards focus solely on token spend; Coding Agent Insights uniquely correlates agent usage with software delivery outcomes like commit throughput and PR velocity within the same observability platform.
  • Multi-agent, multi-vendor coverage: The feature supports agents from multiple vendors (Anthropic, OpenAI, GitHub) under a single dashboard, avoiding the need to maintain separate observability stacks per tool.
  • Organizational hierarchy awareness: The dashboards natively understand org-level dimensions (team, department, cost center), which is uncommon in general-purpose observability tools and critical for enterprise chargeback and access decisions.
  • Integrated with existing CloudWatch workflows: Because it builds on standard CloudWatch metrics infrastructure, teams can reuse existing alarms, anomaly detection, and dashboarding workflows rather than adopting a new tool.

When to Prefer It

  • Scaling AI coding agent rollouts: When an organization is moving from pilot to broad deployment of AI coding tools and needs data to guide access expansion, budget allocation, and team prioritization.
  • Engineering cost accountability: When finance or engineering leadership requires department-level or cost-center-level attribution of AI tooling spend for chargeback or budgeting purposes.
  • Measuring developer productivity impact: When you want to quantify whether AI coding agents are actually improving delivery metrics (PR velocity, commit throughput) rather than relying on anecdotal evidence.
  • Multi-agent environments: When your organization uses more than one AI coding tool (e.g., some teams on Claude Code, others on GitHub Copilot) and you need a unified view rather than siloed per-vendor dashboards.
  • Already invested in CloudWatch: When your observability stack is primarily CloudWatch-based and you want to avoid introducing a separate SaaS tool just for AI coding agent analytics.
  • Enterprise governance requirements: When your organization requires SSO-based identity attribution, audit trails, or policy enforcement for AI tool usage—capabilities provided by the Claude apps gateway integration.
  • Token budget optimization: When you suspect token budgets are either over-provisioned (wasting money) or under-provisioned (limiting productivity) and need per-team data to right-size allocations.

Availability

  • GA status: Generally available as of July 20, 2026.
  • Supported regions: Available in all AWS commercial regions except Middle East (UAE), Middle East (Bahrain), and Israel (Tel Aviv).
  • Supported agents: Claude Code (via Claude apps gateway), OpenAI Codex, and GitHub Copilot; each has a dedicated setup guide in the CloudWatch documentation.
  • Pricing: Standard CloudWatch OpenTelemetry metric ingestion pricing applies; there is no additional charge specific to Coding Agent Insights dashboards beyond the cost of ingesting the OTel metrics.
  • Free tier: The first 10 custom/detailed monitoring metrics per month fall within the CloudWatch free tier; beyond that, standard custom metrics pricing applies.
  • Prerequisites (Claude Code path): Requires Claude Code v2.1.195 or later, an OIDC-compatible identity provider, PostgreSQL 14+, and a TLS-enabled deployment environment running Linux in production.
  • Dashboard availability: Dashboards auto-populate under GenAI Observability > Coding Agent Insights in the CloudWatch console; no manual import is required.

Tags

Servicescloudwatch
Typenew-featureintegration
Conceptscoding-assistantmlops
Use Casesdeveloper-toolsdevopscost-optimization
Providersanthropicopenai
GeographyEMEA

Related Resources

AI Radar AWS

AWS AI/ML news — curated, researched, explained

An automated intelligence platform that curates, researches, and analyzes AWS AI/ML/GenAI announcements daily. Every report is backed by real research — the system reads linked blog posts and documentation to provide accurate, in-depth analysis.

How Each Report Is Generated

  1. Collection — Daily monitoring of the AWS "What's New" RSS feed
  2. Filtering — AI-powered relevance detection for AI/ML/GenAI topics
  3. Taxonomy Tagging — LLM-based classification across 6 dimensions
  4. Importance Scoring — Point-based system with tag bonuses (1-5 stars)
  5. Research Phase — Follows links to blog posts and documentation
  6. Report Generation — Claude Sonnet produces structured 6-section analysis
  7. Visual Summary — Claude Opus generates Mermaid diagrams for key items
  8. Publishing — Static website rebuilt and deployed via CloudFront

Features

  • Faceted filtering by service, type, concept, and more
  • Multi-dimensional taxonomy with 80+ tags across 6 dimensions
  • Geographic availability badges (Global, APJ, EMEA, AMER) with filtering
  • Timeline visualization of announcement volume
  • PDF export for offline reading
  • Mermaid visual summaries for key announcements
  • Daily automated updates — no manual curation
What makes this different: Each report involves a dedicated research phase where the system reads linked blog posts and AWS documentation pages. This produces analysis that goes beyond the original announcement text.

Technology

Built with Python, AWS Lambda, Amazon Bedrock (Claude Sonnet 4.6, Opus 4.6, Haiku 4.5), S3, CloudFront, WAF, EventBridge, and CDK.

Open Source

This project is open source. Fork it, customize it for your needs, and deploy your own instance.
📦 github.com/bbonik/ai-radar-aws

How Importance Scoring Works

Each announcement receives a point score based on multiple factors. The total score maps to a 1-5 star rating:

1★ < 2 pts 2★ ≥ 2 pts 3★ ≥ 3.5 pts 4★ ≥ 5 pts 5★ ≥ 6.5 pts

Point Breakdown

FactorPointsWhen
Core AI service (Bedrock, AgentCore, SageMaker AI)+4Service named in title
Key AI service (SageMaker, Kiro, QuickSight)+2Service named in title
Other AI-related service+1Default
Blog post link+3Link to aws.amazon.com/blogs/
GitHub samples link+2Link to github.com/aws*
Documentation link+1Link to docs.aws.amazon.com/
New model+1.5Tagged as "new-model"
New service+1Tagged as "new-service"
New feature+0.5Tagged as "new-feature"
Anthropic / OpenAI provider+2Provider explicitly mentioned
Instance / notebook announcement-2Hardware/capacity, not feature
Performance / pricing / security-0.5Incremental updates
Region expansion to APJ+1Expands to Asia Pacific
Region expansion (non-APJ only)-1.5Only expands to other regions

Geographic Relevance Badges

Each announcement card shows a small badge indicating whether the feature is available in your region:

🌐 Global Available in all regions
🌏 APJ Asia Pacific
🌍 EMEA Europe / Middle East / Africa
🌎 AMER Americas (US, Canada, South America)
No badge Geography unknown
How geography is detected: The system detects ALL geographies mentioned in each announcement. If the text mentions specific regions (Tokyo, Frankfurt, Oregon, etc.), the corresponding geography badges are shown. If it says "all regions" or is a new feature with no region specified, it gets the Global badge. Geography is also filterable — click a geo chip to see only announcements available in that region.