← Back to research
·10 min read·company

DX

DX is the developer intelligence platform behind the AI Measurement Framework — the category's reference for tracking AI utilization, impact, and cost — combining system metrics with surveys for 300+ companies including Dropbox, Adyen, Vanguard, and Booking.com, where it measured a 16% throughput gain across 3,500+ engineers.

https://getdx.com/ ↗
engineering-intelligenceenterprisegithub

Key takeaways

  • DX's AI Measurement Framework (July 2025), authored by Abi Noda and Laura Tacho, has become the reference framework for measuring AI-assisted engineering — three dimensions (utilization, impact, cost) that deliberately exclude acceptance rates and lines of code
  • The research bench is the moat: founded by Abi Noda (Pull Panda, acquired by GitHub) with Dr. Nicole Forsgren (DORA creator, SPACE co-author) as strategic advisor and investor, plus SPACE co-authors Margaret-Anne Storey and Thomas Zimmermann as collaborators
  • Enterprise proof at scale: 300+ customers including Dropbox, Pinterest, eBay, Pfizer, Vanguard, and BNY Mellon; Booking.com used DX to measure a 16% throughput gain for daily AI users across 3,500+ engineers, with 65% adoption and 150,000 developer hours saved in the first year

FAQ

What is DX?

DX is a developer intelligence platform that combines system metrics from development tools with survey data from developers to measure developer experience, engineering productivity, and — via its AI Measurement Framework — the utilization, impact, and cost of AI assistants and agents.

How much does DX cost?

Pricing is not publicly listed. DX sells through enterprise sales with custom quotes.

What is the AI Measurement Framework?

A research-based framework published by DX in July 2025 for measuring AI-assisted engineering across three dimensions — utilization, impact, and cost — using a mix of telemetry from AI tools and self-reported developer data, deliberately excluding shallow metrics like acceptance rates.

How is DX different from Swarmia or Jellyfish?

Swarmia and Jellyfish lead with system metrics (git, Jira) for team visibility and resource allocation; DX leads with a research-backed mixed-methods model — surveys plus system data — and owns the framework layer (DX Core 4, AI Measurement Framework) that competitors benchmark against.

Executive Summary

DX is a developer intelligence platform that measures engineering organizations by combining system metrics from development tools with self-reported survey data from developers.[1] It was founded by Abi Noda — who previously founded Pull Panda, acquired by GitHub in 2019 — and Greyson Junggren, and built in collaboration with the field's leading researchers: Dr. Nicole Forsgren (creator of DORA, co-author of SPACE) joined as strategic advisor and investor, alongside SPACE co-authors Dr. Margaret-Anne Storey and Dr. Thomas Zimmermann.[2]

DX matters to this category for one reason above the rest: its AI Measurement Framework, published July 9, 2025 by Noda and CTO Laura Tacho, has become the reference framework for measuring AI-assisted engineering — three dimensions (utilization, impact, cost) that deliberately exclude acceptance rates and lines of code as misleading.[3][4] The enterprise proof is unusually concrete: Booking.com used the framework across 3,500+ engineers and measured 16% higher change throughput for daily AI users versus non-users, with 65% adoption and 150,000 developer hours saved in the first year.[5][6] ICONIQ Growth invested in March 2025, when DX reported 300+ customers including Pinterest, eBay, Pfizer, Vanguard, BNY Mellon, and Dropbox.[7]

AttributeValue
CompanyDX (getdx.com)
FoundedAbi Noda (CEO; Pull Panda, acquired by GitHub 2019) and Greyson Junggren[2]
Research affiliationNicole Forsgren (strategic advisor/investor), Margaret-Anne Storey, Thomas Zimmermann[2]
FundingNot publicly disclosed — ICONIQ Growth investment, March 2025[7]
Customers300+ companies as of March 2025[7]
HeadquartersSalt Lake City, UT (~80 employees at investment, doubling planned)[7]
Open SourceNo — closed source, proprietary SaaS

Product Overview

DX positions itself as the system of record for engineering measurement in the AI era: a "developer intelligence platform" whose pitch is that neither system metrics nor surveys alone tell the truth, so it instruments both.[1] The platform spans four product areas — developer experience (DXI index, benchmarking, DevSat surveys, experience sampling), engineering productivity (DX Core 4, SDLC analytics, TrueThroughput™), AI measurement (usage analytics, AI Code Metrics, impact analysis, vendor evaluation), and AI enablement (Agent Ops tooling, AI readiness assessment, systems catalog, internal developer portals) — plus DX AI, an "AI copilot for engineering leaders" that surfaces automated insights and risk flags.[1]

The AI Measurement Framework is the connective tissue. It tracks AI assistants and agents across utilization (are developers actively using the tools), impact (downstream productivity and business outcomes), and cost (spend versus return), and explicitly excludes acceptance rates because they say little about real impact.[3][4]

Key Capabilities

CapabilityDescription
AI Measurement FrameworkResearch-based measurement of AI utilization, impact, and cost across assistants and agents[3]
AI Code MetricsUsage analytics and AI-impact analysis on code output, plus AI vendor evaluation[1]
Agent Ops / AI enablementAgent operations tooling, AI readiness assessment, systems catalog, context management[1]
DX Core 4Unified productivity framework (speed, effectiveness, quality, impact) with industry benchmarks[1]
DXI + surveysDeveloper Experience Index, DevSat surveys, targeted studies, experience sampling[1]
SDLC analyticsSystem-metric pipelines: team dashboards, sprint analytics, R&D capitalization, executive reporting[1]
DX AIAI copilot for engineering leaders — automated insights and risk flagging[1]

Technical Architecture

DX is closed-source, enterprise SaaS. Its defining architectural choice is mixed-methods measurement: telemetry connectors into development tools (git, CI/CD, AI assistants) feed system metrics, while structured surveys and experience sampling capture what telemetry cannot — and the frameworks (Core 4, AI Measurement Framework) define how the two combine into comparable, benchmarkable numbers.[1][3]

Key Technical Details

AspectDetail
DeploymentManaged SaaS, enterprise sales[1]
Data sourcesDevelopment-tool telemetry + periodic surveys + experience sampling[1]
AI attributionFramework-level (utilization/impact/cost via tool telemetry and self-report), not line-level code attribution[3]
Open SourceNo

Strengths

  • Owns the category's reference framework — the AI Measurement Framework (July 2025) is the benchmark competitors and enterprises alike cite when defining how to measure AI-assisted engineering, covered in depth by The Pragmatic Engineer[3][4]
  • Research credibility no competitor matches — Forsgren (DORA), Storey, and Zimmermann (SPACE) are directly affiliated; DX frameworks descend from the same lineage as DORA and SPACE[2]
  • Quantified enterprise outcomes — Booking.com: 16% higher throughput for daily AI users across 3,500+ engineers, 65% adoption, 150,000 developer hours saved in year one; WorkHuman raised AI assistant ROI 21% using DX measurement[5][6][4]
  • Blue-chip customer base — 300+ companies including Dropbox (with a Drew Houston testimonial on the homepage), Adyen, Vanguard, Pinterest, eBay, Pfizer, and BNY Mellon[7][1]
  • Honest about AI hype — DX's own framing concedes that coding is only 20–25% of developer time, so a 10% coding-time saving is 2–2.5% of total time; this realism builds trust with skeptical engineering leadership[4]
  • Backed by ICONIQ Growth — the investor behind Snowflake, Datadog, and GitLab, with headcount doubling planned post-investment[7]

Cautions

  • Proprietary black-box index — the DXI has been called the most problematic metric in the Core 4 framework: the formula and exact survey questions are proprietary, making it hard to know what moves the score and creating dependency on the vendor[8][9]
  • Survey-based measurement has known limits — self-reported data is subjective and sampling-dependent; competitors argue Core 4 quantifies experience without fully embracing its complexity, and the broader metrics genre drew heated scrutiny after McKinsey's 2023 developer-productivity piece[9][8]
  • No line-level AI attribution — DX measures AI at the framework level (utilization, impact, cost), not per-line code provenance; teams wanting ground-truth attribution of which lines came from which tool need a different layer[3]
  • Funding and pricing opacity — the ICONIQ investment amount and round are undisclosed, revenue is undisclosed, and there is no public pricing[7]
  • Vendor grades its own homework — the headline case studies (Booking.com) are published by DX about its own framework; the 16% figure is a DX-measured correlation between daily AI use and throughput, not an independent controlled study[5]

Pricing & Licensing

TierPriceIncludes
EnterpriseCustom (not publicly listed)Developer experience, productivity, AI measurement, and AI enablement modules; sold via enterprise sales[1]

Licensing model: Proprietary closed-source SaaS, annual enterprise contracts.

Hidden costs: Survey program operation (response rates require sustained internal championing), and consulting dependency risk given the proprietary DXI methodology.[8]

Competitive Positioning

Direct Competitors

CompetitorDifferentiation
SwarmiaSwarmia leads with git/Jira system metrics and working agreements for team leads; DX leads with mixed-methods research frameworks and sells higher in the org — Swarmia itself benchmarks against DX Core 4[9]
JellyfishJellyfish anchors on engineering-to-business alignment (allocation, capitalization) for the C-suite; DX anchors on developer experience plus AI measurement with a research pedigree
SpanSpan is an AI-native newcomer to engineering intelligence; DX is the established framework owner with 300+ enterprise customers[7]
LinearBLinearB pairs metrics with workflow automation and publishes pointed critiques of DX's proprietary DXI; DX counters with research affiliation and survey depth[8]
GitClearGitClear measures AI at the line level (Diff Delta attribution); DX measures at the framework level with surveys plus telemetry[3]

When to Choose DX Over Alternatives

  • Choose DX when: leadership wants the canonical, research-backed framework for AI measurement — utilization, impact, cost — and values survey signal alongside telemetry[3]
  • Choose Swarmia when: team leads want self-serve system metrics and working agreements without an enterprise survey program
  • Choose Jellyfish when: the primary buyer is finance/C-suite and the job is allocation and R&D capitalization
  • Choose GitClear or line-level tools when: you need ground-truth attribution of AI-generated code, not adoption and impact dashboards

Ideal Customer Profile

Best fit:

  • Enterprises (hundreds to thousands of engineers) rolling out AI assistants and agents who must justify spend with utilization, impact, and cost data[3]
  • Organizations that already run or want developer experience survey programs, not telemetry alone[1]
  • Engineering leadership that wants industry benchmarks and a defensible, research-based methodology for the board[2]

Poor fit:

  • Small teams without budget for enterprise sales cycles or appetite for survey programs
  • Teams wanting line-level AI code attribution or open, auditable metric formulas[8]
  • Buyers requiring public pricing or open-source deployment

Viability Assessment

FactorAssessment
Financial HealthSolid signals, limited disclosure — ICONIQ Growth investment (March 2025), 300+ customers, headcount doubling planned; amounts undisclosed[7]
Market PositionCategory framework owner — AI Measurement Framework and Core 4 are the benchmarks competitors compare against[3][9]
Innovation PaceStrong — AI Measurement Framework (July 2025), AI Code Metrics, Agent Ops, and DX AI copilot shipped as AI reshaped the category[1]
Community/EcosystemResearch-led — Engineering Enablement newsletter, published frameworks, Forsgren/Storey/Zimmermann affiliation[2]
Long-term OutlookPositive — the AI measurement question is getting bigger, and DX defined how enterprises answer it

DX converted developer-experience research into the default measurement standard for AI-augmented engineering at precisely the moment every CTO is being asked "what are we getting for our AI spend." The open risks are disclosure (funding and revenue are opaque) and methodology trust — the proprietary DXI invites exactly the black-box critique competitors are making.[7][8]

Bottom Line

DX is the most important tool in AI engineering intelligence: it owns the AI Measurement Framework that the rest of the category orients around, pairs it with the field's strongest research bench, and has the enterprise case studies — Booking.com's 16% throughput gain across 3,500+ engineers chief among them — to back the framing.[3][5][2] The trade-offs: survey-heavy methodology with a proprietary index formula, no line-level AI code attribution, and undisclosed funding and pricing.[8]

Recommended for: Enterprises that need a defensible, research-backed answer to "is our AI investment working" — measured across utilization, impact, and cost — with benchmarks leadership will accept.

Not recommended for: Small teams, buyers needing transparent metric formulas or line-level AI attribution, or organizations unwilling to run survey programs.

Outlook: Strong. The framework owner usually wins the measurement category, and DX's July 2025 framework arrived first with the right customers attached; watch whether Agent Ops keeps pace as measurement shifts from assistants to autonomous agents.[3][1]

Research by Ry Walker Research • methodology

Sources

Share on XShare on LinkedIn