How DAC works.
DAC is a measurement system for product operations. It watches your team's work through six lenses, measures what matters across twenty-seven dimensions, and generates intelligent improvement items every sprint.
This page explains exactly how it works. No mystery. No black box. No "trust us, it is AI."
Read it in 90 seconds or eight minutes. Both options are here.
DAC runs a five-step loop every sprint.
Watch. Measure. Detect. Generate. Coach. Then the sprint runs, and the loop begins again.
Watch
DAC watches Linear, GitHub, PostHog, URL crawl, and public web sources. It reads the signals your team is already generating. No extra work. No new tools. No surveys.
Measure
27 dimensions across 6 functions. Scoring engine v3.0 with stage-adaptive weighting, cross-function interaction effects, coherence scoring, and anomaly detection. Every dimension is measured against benchmarks from real teams at your stage.
Detect
What improved. What declined. What drifted. Translation Gap direction. Compound effects that are unlocking or stalling. The scoring engine identifies patterns a human reviewing dashboards would miss.
Generate
2 to 5 intelligent items per sprint. Each one includes evidence, benchmarks, expected impact, and priority. Items are pushed directly to Linear alongside your feature work. No separate backlog. No transformation stream.
Coach
IC signal coaching for individual contributors. Team development guidance for leads. System coaching for VPs. Portfolio intelligence for executives. Same data model, different language for each audience.
Three frameworks. 88 dimensions. One unified score.
Three original frameworks under CC BY 4.0. Openly published. Independently verifiable. Combined into a single score that captures team, process, and product maturity.
People
ProdOps Intelligence
27 dimensions across 6 functions: Strategy, Design, Development, Operations, GTM, Intelligence.
5 stages from Foundation (27-48) through Building (49-70), Practicing (71-92), Optimizing (93-114), to Compounding (114-135).
Measures: Is the team mature enough to compound?
Process
Dev Lifecycle
34 tasks across 6 stages: Specify and Constrain, Design and Validate, Build and Integrate, Test and Harden, Ship and Observe, Learn and Compound.
3 cross-cutting concerns: Token Economics, Role Fluidity, Cognitive Debt.
Measures: Is the process AI-native and sustainable?
Product
Product Assessment
27 dimensions across 6 attributes: AI Architecture, Intelligence Layer, Human-AI Interaction, Data Foundation, Business Model, Compound Mechanics.
5 stages from Wrapper (27-49) through Enhanced (50-72), Integrated (73-94), Native (95-117), to Compounding (117-135).
Measures: Is the product built to compound?
The cross-framework tension analysis is where the real insight lives.
DAC also supports overlay frameworks: POM, DORA, OKRs, ShapeUp, North Star, and Linear Method. These layer on top of the core three, adding contextual benchmarks without replacing the unified score.
Your first DAC score in two stages.
Stage 1: Fast URL Crawl
Enter a URL. DAC crawls the public-facing product and scores all 27 product dimensions in about two minutes. Free. No sign-up required. Instant baseline.
Stage 2: Public Web Enrichment
DAC searches up to 15 public sources in the background over 3 to 5 minutes. Every source is cited. Maximum 3 points per function delta.
URL Input
dacard.ai/try
Stage 1: Crawl + Score
~2 min, 27 dimensions
Baseline Score
Stage 2: Web Enrichment
Up to 15 cited sources
Enriched Score
What DAC watches.
Signals flow in from the tools your team already uses. DAC normalizes everything into a common taxonomy. No manual reporting. No data entry.
Linear
Project Management
Sprint velocity and completion rates
Cycle time from start to done
Review and approval patterns
Backlog health and grooming cadence
Estimation accuracy over time
GitHub
Engineering
Merge patterns and PR throughput
Review turnaround time
Test coverage trends
Dependency health and update cadence
Commit pattern analysis
PostHog
Product Analytics
Feature adoption curves
Behavior shift detection
Experiment results and statistical significance
Retention and engagement patterns
Funnel drop-off analysis
The intelligent backlog.
Every sprint, DAC generates improvement items. Not vague suggestions. Specific, actionable items with full context.
DAC analyzes signals from the previous sprint and generates 2 to 5 improvement items ranked by expected impact.
Each item includes: what to do, why it matters, what good looks like, expected impact, and priority. Human reviews and approves.
Approved items push directly to Linear as tickets. They land alongside your feature work. No separate board. No extra process.
DAC watches whether items are completed, measures the impact in subsequent sprints, and adjusts future recommendations based on what worked.
Items live alongside feature work. No separate transformation stream. No "improvement sprint." The compounding happens inside the work your team is already doing.
From suggestion to autonomy.
You control how much DAC does. Start with full human review. Graduate to autonomous operation as trust builds.
Suggest
DAC generates improvement items. A human reviews and approves every one before it enters the backlog. Full control. Full visibility. This is where every team starts.
Assist
DAC pushes approved categories of items automatically. Low-risk items (documentation, test coverage) flow without approval. Higher-risk items still require human review. Weekly digest instead of per-item approval.
Autonomous
DAC manages the improvement backlog end to end. Humans set guardrails (budget, scope, priority ceilings) and review outcomes monthly. The system runs itself within the boundaries you define.
The methodology is open.
DAC is not a black box. Every part of the scoring methodology is published and verifiable.
Scoring rubrics are published. You can read the criteria for every dimension at every stage.
Benchmarks come from real teams. Not synthetic data. Not theoretical models. Actual scores from actual product organizations.
Every signal cites its source. You can trace any score change back to the data point that caused it.
Every score is reproducible. Run the same inputs through the same engine, you get the same output. Deterministic where it matters.
Frameworks are licensed CC BY 4.0. Use them without DAC. Build on them. Teach with them. No lock-in.
"If you cannot explain how a score was calculated, the score is not useful. It is decoration."
What DAC does with your data.
DAC does
Read public URLs you submit for scoring
Ingest signals from tools you explicitly connect (Linear, GitHub, PostHog)
Measure patterns across sprint cycles
Generate improvement items based on your team's data
Store scores and trends for your organization only
Search public web sources for enrichment (cited, capped)
DAC does not
Scrape LinkedIn or private social profiles
Access paywalled or login-gated content
Share data between organizations, ever
Write to your tools without explicit permission
Train models on your proprietary data
Sell, share, or aggregate your data for benchmarks without consent
The math that makes this work.
A team that improves 0.3 points per sprint reaches top-quartile in six months.
12 sprints: Foundation (52) → Scaling (90)
Most improvement programs assume linear progress. Do the training. Ship the process. Hope it sticks.
DAC works on compound improvement. Each sprint builds on the last. Small gains accumulate. A 0.3 point improvement per sprint does not sound dramatic. But compound it over 12 sprints and the trajectory separates you from peers who are improving linearly or not at all.
The gap widens every sprint. Teams using DAC are not just better. They are getting better faster. That is the compounding effect.
See DAC in action.