Release Notes: 2026-06-17

Engineering Insights now reads the way leaders think

What's new

Delivery, reliability, and CI health in Engineering Insights

• A Delivery and Reliability view tracks all four DORA metrics - deployment frequency, lead time for changes, change failure rate, and mean time to recovery (MTTR) - each scored against Elite, High, Medium, and Low performance bands, so you know exactly where you stand against industry benchmarks
• Failed production deployments and reverts are detected automatically and shown alongside successful deploys, so change failure rate reflects what actually happened
• Teams that do not formally tag deployments still get delivery numbers through deploy-workflow and merge-to-release-branch detection, and the view tells you when a number is being estimated rather than read directly
• A CI/CD Health view reads GitHub Actions for build success rate, run and failure counts, duration, queue time, and a per-workflow breakdown with automatic flaky-test detection
• A redesigned PR Flow view reports cycle time, time to first review, self-merge rate, open-PR aging, and PR size distribution
• You can define what counts as production per repository, and record recovery times by hand when automatic detection is not enough

A security and AI-assisted engineering picture

• A Security Posture view consolidates open Dependabot and code scanning alerts, with a prioritized "fix these first" list ranked by CVSS score and EPSS exploit probability, the GHSA/CVE identifier, the affected repository, and whether a patched version is available
• Open alerts over time, a severity mix, and time-to-remediate help leaders triage real risk instead of raw alert counts
• An AI-Assisted Engineering view measures the ROI of AI coding assistants - GitHub Copilot, Cursor, and Claude Code - showing adoption, AI-assisted versus human lines, acceptance rate, and cost, plus an AI-versus-human cohort comparison of who ships and reviews faster
• A new AI Leverage pillar feeds the Engineering Velocity Score (EVS) when AI coding data is present, and leaves the score unchanged when it is not, so teams without AI tooling are never penalized
• Executive reports in Custom Views can now include these delivery, CI, review, and AI measures, scoped to specific repositories and branches

Bloomy answers from live engineering work

• Ask Bloomy about pull requests, issues, commits, workflow runs, and projects across connected GitHub data
• Search and summarize tasks across Jira, Linear, and Asana with consistent assignee resolution
• Run codebase questions like "where is this implemented" against connected repositories, answered in language a leader can forward to their team
• Admins set each live capability to on, off, or admin-only, and people see a clear message when a capability has been turned off
• Gemini 3.5 Flash is available in Bloomy's model picker

Connect your AI tools to Abloomify knowledge

• Connect Cursor, Claude Code, Claude Desktop, and Codex with copy-ready setup snippets
• Employees manage their own scoped connections under My Connectors, while admins issue and revoke organization-wide keys under External AI Access
• Admins can revoke company keys but not an employee's personal connection, so ownership stays clear
• Setup snippets point at a branded address such as mcp.abloomify.com, so the URL your team copies looks like Abloomify
• Connecting Cursor or Claude by key now saves reliably for admins

A clearer device agent rollout for IT

• Download a recommended signed macOS installer or generate a fresh signed build when you need one, with version gating so an older build is not rolled out by accident
• Generate a per-company Chrome profile and download central configuration profiles, with links to the deployment guide
• Set menu bar visibility, turn on silent auto-update, and pin or roll back to a supported version from one place, with settings applied on the agent's next check-in
• Chrome profiles now enforce the correct minimum extension version, and a saved version pin below the current minimum stays visible so admins can change it
• Coding-session context captured on enrolled machines can become searchable company knowledge, subject to scope and admin settings

Who benefits

• Engineering and delivery leaders: see deployment frequency, lead time, failure rate, and recovery time with performance bands and period-over-period change
• Security and risk owners: review open vulnerabilities by severity, CVSS, exploit probability, and patch availability in one view
• Teams adopting AI coding tools: connect Cursor, Claude Code, and similar assistants to Abloomify knowledge with scoped keys, and see adoption signals in Engineering Insights
• IT and operations admins: roll out, configure, and version-control device agents across macOS and Chrome from one console
• Anyone using Bloomy: ask about live GitHub, tasks, and code, within the limits each admin sets

Try it

• Open Engineering Insights to explore the new Delivery, CI/CD Health, Security Posture, and AI-Assisted Engineering views
• Set what counts as production per repository, then build an executive report in Custom Views scoped to the repos and branches you care about
• In Connected Apps, set each Bloomy live capability to on, off, or admin-only, then ask Bloomy about a recent pull request or task
• Connect an AI coding assistant from My Connectors, or issue organization-wide keys from External AI Access
• Manage your fleet from Integrations > Device Agents, including signed installers, auto-update, and version pinning