Cortex

Executive Summary

Cortex is a commercial internal developer portal (IDP) founded in 2019 by Anish Dhar, Ganesh Datta, and Nikhil Unni, and backed by Y Combinator.^[1] The core loop is catalog–score–act: catalog every service, resource, domain, API, and ML model; continuously score them against standards like code coverage, vulnerability SLAs, and package freshness via Scorecards; and drive remediation across teams via Initiatives.^[1] An Eng Intelligence module layers DORA metrics, cycle time, velocity, and MTTR on top of the catalog, positioning Cortex for engineering leaders and platform teams as well as developers.^[2]

For an AI-measurement evaluation, the precise framing matters: Cortex is first an IDP, and its engineering-intelligence and AI-measurement features are one part of that broader platform. AI measurement today consists of an AI Impact Dashboard launched in October 2025 that integrates with GitHub Copilot — and only Copilot — to compare cycle time, deployment frequency, merged PRs, and other metrics between AI users and non-users.^[3][4] The company raised a $60M Series C in September 2024 led by Scale Venture Partners, with Sequoia, IVP, Y Combinator, and Stripe founders Patrick and John Collison participating; named customers include Adobe, Grammarly, and Opendoor.^[5]

---

Product Overview

Cortex pitches itself as eliminating the "developer tax" — the cycle of finding, fixing, and waiting — by making the state of all software visible and the path to standards compliance actionable.^[1] Teams catalog services and resources through 60+ pre-built integrations, define standards in Scorecards, and run Initiatives (deadline-driven campaigns) to push every team toward those standards; Eng Intelligence centralizes metrics from Git, incident management, and project management tools and ties trends back to the catalog.^[7][2]

Key Capabilities

AI Measurement Today

The AI Impact Dashboard, launched October 15, 2025, integrates directly with GitHub Copilot. Cortex automatically attaches an "AI Usage" label to any team member active in Copilot within the last 7 days, then compares engineering metrics between AI users and non-users; Active Copilot Users and Adoption Rate metrics feed the Metrics Explorer.^[3][4] Coverage of other AI coding tools (Cursor, Claude Code, etc.) is not part of the dashboard as launched.^[3]

---

Technical Architecture

Cortex is closed-source, managed SaaS. Data flows in from 60+ pre-built integrations — Git providers, CI, incident management (PagerDuty, Opsgenie, FireHydrant), and project management — into the catalog, which Scorecards, Initiatives, and Eng Intelligence dashboards all read from.^[7][2] The entity model is opinionated and semi-rigid: some catalog entity types are fixed and resist customization, a deliberate trade for faster time to value.^[7]

Key Technical Details

---

Strengths

• Scorecards are the defining feature — continuous, automated measurement of every service against defined standards, with Initiatives to enforce them; third-party evaluation calls this the strongest standards-enforcement model among commercial IDPs^[7]
• Fast time to value — opinionated design plus 60+ pre-built integrations gets a functional portal deployed within weeks, versus the multi-engineer ongoing investment Backstage typically demands^[7]
• Metrics tied to action — Eng Intelligence dashboards are natively integrated with the catalog, Scorecards, and Initiatives, so an insight ("cycle time regressed") connects directly to an owner and a remediation campaign^[2]
• Copilot ROI evidence, not just adoption stats — the AI Impact Dashboard compares Copilot users against non-users on cycle time, deployment frequency, and PR throughput, a concrete attempt at AI attribution rather than seat-count reporting^[4]
• Well capitalized with blue-chip backers — $60M Series C (September 2024) led by Scale Venture Partners with Sequoia, IVP, Y Combinator, and the Collison brothers participating^[5]
• Enterprise proof points — Adobe, Grammarly, and Opendoor named as customers using the catalog and continuous standards monitoring^[5]

---

Cautions

• An IDP first, an analytics tool second — engineering intelligence and AI measurement are modules of a service-catalog platform; buying Cortex for AI analytics means buying (and rolling out) a full developer portal^[2]
• AI attribution is Copilot-only — the AI Impact Dashboard launched October 2025 with a direct GitHub Copilot integration and no stated coverage of Cursor, Claude Code, or other AI coding tools, leaving multi-tool organizations with partial visibility^[3]
• Crude usage proxy — "AI user" is anyone active in Copilot in the past 7 days, a binary label that cannot distinguish heavy agentic usage from occasional autocomplete^[4]
• Semi-rigid data model — fixed catalog entity types resist customization; teams whose engineering topology doesn't match Cortex's assumptions "will fight the system rather than extending it"^[7]
• No public pricing, expensive at scale — third-party estimates put the full IDP plan at ~$69/user/month, roughly $78K–$83K/year for 100 engineers, with self-service actions less flexible than Port's^[7]
• Proprietary lock-in — migrating away requires re-exporting the entire service catalog, scorecards, and workflows^[7]

---

Pricing & Licensing

Cortex does not publish pricing; the figures above are third-party estimates, which put 100 engineers on the full IDP plan at roughly $78K–$83K per year.^[7]

Licensing model: Proprietary closed-source SaaS, per-user subscription with custom site licenses.^[7]

Hidden costs: Catalog population and integration rollout are a real platform-team project, and per-seat economics become prohibitive for large engineering organizations; migration away later carries re-export costs.^[7]

---

Competitive Positioning

Direct Competitors

When to Choose Cortex Over Alternatives

• Choose Cortex when: you want a service catalog with continuous standards enforcement, and engineering metrics that tie back to owners and remediation campaigns — with AI measurement as a bonus, not the centerpiece^[7]
• Choose Faros AI or LinearB when: engineering intelligence and AI-impact measurement are the primary purchase, not a portal add-on
• Choose Backstage when: you have platform engineers to staff an open-source framework and need full data-model control^[7]
• Choose Port when: developer self-service workflows and a flexible entity model matter more than opinionated scorecards^[7]

---

Ideal Customer Profile

Best fit:

• Platform engineering teams at organizations of 100–200+ engineers that want standards enforcement and compliance visibility without staffing a Backstage build^[7]
• Engineering leaders who want DORA/velocity metrics connected to a service catalog and ownership data, not floating in a separate analytics tool^[2]
• GitHub Copilot shops that want first-party evidence of Copilot's impact on delivery metrics^[4]

Poor fit:

• Teams buying primarily for AI-impact analytics across multiple AI coding tools — the dashboard covers Copilot only^[3]
• Organizations whose engineering topology doesn't fit Cortex's semi-rigid entity model^[7]
• Smaller teams for whom per-seat IDP pricing is hard to justify, or open-source/self-host requirements^[7]

---

Viability Assessment

Cortex is a well-funded, enterprise-proven IDP whose bet is that the service catalog is the right substrate for engineering metrics — an insight is only useful if it routes to an owner and a scorecard. That architecture is genuinely differentiated for platform teams. But its AI measurement is young and narrow: one Copilot dashboard with a binary 7-day usage label, launched October 2025.^[4]

---

Bottom Line

Cortex is a strong buy for platform teams that want a managed service catalog with teeth — scorecards and initiatives that enforce standards rather than report on them — and engineering metrics wired into that same ownership graph.^[7][2] As an AI engineering intelligence tool specifically, it is a partial answer: the AI Impact Dashboard delivers real Copilot-versus-baseline comparisons, but covers one tool with a coarse usage signal, inside a platform you must adopt whole.^[3][4]

Recommended for: Platform engineering teams of 100+ engineers standardizing on a commercial IDP, especially GitHub Copilot shops that want delivery-metric evidence of AI ROI alongside catalog and scorecards.

Not recommended for: Buyers whose primary need is multi-tool AI attribution and engineering analytics without a portal rollout — engineering-intelligence-first platforms fit better.

Outlook: Solid. The Series C and named enterprise customers underwrite the IDP franchise, and the catalog-plus-metrics architecture is a credible base for expanding AI measurement; watch whether AI attribution grows beyond the single Copilot integration.^[5][3]

---

Research by Ry Walker Research • methodology