Product Vision

01 Thesis: the surface is the library, the substrate is the frameworks

Most assessment products fail because they confuse the artifact with the experience. They publish a framework, build a dashboard, and ask the buyer to internalize a vocabulary the buyer has no incentive to learn. The framework is the thing the company sells. The dashboard is the thing the customer touches. Adoption stalls in the gap between them.

Dacard inverts that order. The frameworks are not the product. They are the substrate. The product is the Pattern Library: a small, growing, named collection of cross-framework tension patterns that explain what is happening to a team in language a senior product leader recognizes on first read. Translation Gap. Fragility Signal. Compound Ready. Each pattern earns its place by being legible, shareable, and predictive. The 88 dimensions across three frameworks sit underneath, doing the calibration work that makes the surface trustworthy.

This inversion matters because it dictates everything else. The buyer-facing motion runs on patterns. The investor-facing moat runs on the substrate. The viral motion on LinkedIn runs on patterns. The defensibility against platforms runs on substrate fusion. Patterns travel. Frameworks calibrate. Both stay live, and the line between them is the architecture.

02 The diagnostic in plain language

A senior PM lands at a Series B company. Three months in, the board asks whether the AI rebuild is working. The PM has DORA scores from the engineering team, an NPS report from the customer team, and an OKR tracker from the chief of staff. None of those answer the question. The question is cross-functional, evidence-led, and has no instrument.

Dacard is the instrument. The PM connects Linear, GitHub, and the company URL. Within minutes, DAC returns a diagnostic across 88 dimensions and three frameworks, names the patterns firing, ranks the next moves, and pushes the highest-leverage actions into Linear. From that moment forward, DAC stays in the workflow as a continuous coach. The motion is "onboard, then coach." Not a one-time audit. Not a quarterly survey. A standing operating coach in the workflow the team already runs.

The category claim, at investor altitude, is decision intelligence platform for product teams in the agentic era. At buyer altitude, it is Agentic Product Operations. The two phrases coexist deliberately. The first travels on partner-meeting whiteboards. The second travels on LinkedIn.

03 The three-layer moat in detail

The product is one diagnostic on the surface, three layers of moat underneath. Each layer is defensible in isolation. The composition is what no adjacent platform can replicate, because the composition requires cross-function scoring at scale, outcome data per scored team, and the statistical pipeline tying the two together.

Surface: the Pattern Library

A pattern is a named cross-framework tension. Three live today.

Translation Gap. High Development Lifecycle scores, low Product Assessment scores, weak Team Operations evidence linking the two. The team ships fast and the product still feels off. The buyer concern is the one a CPO loses sleep over: we are efficient and not impactful. The pattern names that gap and points to the specific dimensions that need motion. It is the signature insight of the product, and the one that travels furthest on LinkedIn because every senior PM has felt it.

Fragility Signal. A composite ladder reading that looks healthy at the top while specific lower-tier dimensions show stress (high incident recurrence, weak rollback discipline, brittle context handoffs to agents). The team looks good on a dashboard and is one bad week from a regression. The buyer concern is the one a VP Product fields when the board asks if the team can absorb scale. Naming the pattern surfaces the risk before the bad week arrives.

Compound Ready. The constructive twin. The team has crossed a threshold across all three frameworks where adding agent leverage compounds rather than dilutes. The buyer concern is when to expand the pod, when to move from augmenting humans to orchestrating fleets. Compound Ready is the green-light pattern. It is also the one teams want to claim on LinkedIn, which is exactly why naming matters.

Each pattern is trademarkable thought-territory. Each addresses a distinct buyer job. Each can carry a category of content (founder-led posts, briefings, partner conversations) without needing to invoke the underlying framework vocabulary. The 8+ Q4 2026 roadmap adds patterns in the categories most evident in design-partner data: agent-leverage drift, decision throughput stalls, archetype-mismatch cases, post-restructure realignment patterns, and a small set surfaced by the discovery engine described in section nine.

Workflow: Ranked Actions

A pattern names what is happening. A ranked action says what to do next. The ranking is not a flat priority list. It is LNO-classified (Leverage, Neutral, Overhead, after Shreyas Doshi) and archetype-conditioned, so the same dimension may carry a different priority for a startup pod than for a scaled function. The ranking is conditioned on the patterns currently firing, the team's archetype, and the outcome history of comparable teams in the calibration set.

Ranked actions push to where the team works. Linear is the primary push target: actions become typed cards with framework context attached, ready for the next planning cycle. Slack carries the pulse: a brief weekly read tied to the cycle the team runs. Agent fleets receive context push that teaches Claude Code, Cursor, and other coding agents the framework state, so the agent's next decision is informed by the team's maturity, not just the team's tickets.

The flywheel is the data. Every action a team takes (accepts, rejects, defers, marks complete) returns telemetry to the calibration pipeline. The ranking model updates. The next team scored sees a better ranking. This is the calibration data moat in motion: replicable only by competitors who have an equivalent customer base, which they do not.

Engine: the Calibration Pipeline

The engine sits beneath the surface and the workflow. Four functions:

1. Outcome data capture. Every score, every action accepted or rejected, every cycle outcome (shipped, not shipped, retired, reverted) returns to the pipeline. The capture is automatic via adapter telemetry, not survey-based. By Q4 2026, target coverage is 50% of Pro+ customers (Linear, GitHub, and Stripe connected).
2. Predictive-validity testing. The pipeline continuously asks: do the dimensions still predict outcomes? Dimensions that stop predicting are flagged for retirement. Dimensions that begin predicting new outcomes (new agent-fleet patterns, new restructure patterns) earn their place. The framework is empirically defended, not ideologically defended.
3. Pattern discovery. Co-occurrence analysis across 88 dimensions, archetype, and outcome telemetry surfaces candidate patterns. Each candidate is gated by power thresholds, then by qualitative review. New named patterns ship quarterly.
4. Archetype recalibration. Archetypes are not fixed. As AI capability and team structure shift, archetype boundaries shift with them. The engine recalibrates archetype weights as new outcome data accrues. The first archetype recalibration ships Q1 2027.

The engine is where the moat compounds. Every customer makes the next customer's diagnostic better. Every cycle of outcome data tightens the ranking. Every quarter of pattern discovery adds inventory the surface can speak to.

The engine is also where the company stops being a scoring product and becomes a research instrument. Every quarter, the calibration pipeline surfaces drift: which dimensions are losing predictive power, which archetypes are blurring, which patterns are emerging. The output of that drift analysis is the basis for the public pattern-validity report. Investors do not have to take the moat on faith. They read the report. The report is grounded in the customer base. The customer base is grounded in the buyers who already needed the diagnostic. The loop is empirical at every step.

The cost guardrail on the engine is deliberate. Pattern-discovery analytics run against snapshot data, not against live LLM calls. Predictive-validity tests are statistical, not generative. The expensive AI work happens on the surface (the diagnostic, the coaching prose, the ranked-action narrative); the engine work happens on warehouse-grade infrastructure where unit costs scale sub-linearly. The 78 to 82 percent blended gross margin holds because the engine costs do not scale with customer count the way inference does.

04 Three frameworks underneath the surface

Beneath the named patterns sit three calibrated frameworks, 88 dimensions in total. The frameworks are the substrate that makes the patterns trustworthy. Customers do not need to learn the framework names to use the product. Buyers do not need to internalize 88 dimensions. The frameworks earn their keep by ensuring that every named pattern is grounded in observable, measurable practice.

The three frameworks measure three different functions. Team Operations measures how the cross-functional product team operates: rituals, decision quality, feedback loops, talent density, agent-augmented practices. Development Lifecycle measures how work flows from idea to shipped value: specification depth, context handoffs to agents, validation discipline, ship cadence, and the compounding effects of an agent-leveraged pipeline. Product Assessment measures the product itself: how AI-native it is, where it sits on the wrapper-to-compounding ladder, what the user gets that they could not get from a non-AI alternative.

Each framework has its own ladder, calibrated to the function it measures. Team Operations climbs from Foundation to Compounding because team practices accumulate. Development Lifecycle climbs from Specify to Compound because work practices form a pipeline. Product Assessment climbs from Wrapper to Compounding because AI products evolve from feature to platform.

For board-altitude reading, the three frameworks collapse into one composite verb ladder: React, Augment, Orchestrate, Lead, Compound. The composite is what shows up on dashboards, in board summaries, in the rail header that names where the team is right now. The per-framework ladders show up in coaching output, when the conversation moves to a specific function and the operator-coach voice needs to name the maturity of that function in particular.

The split between composite and per-framework vocabulary is deliberate. Tight UI contexts (rail rows, triad labels, dashboard headers) use the composite verb ladder and the abbreviated framework labels: Team, Operation, Product. Coaching prose uses the full names. The grammar holds across every surface.

05 The 30-60-90 onboarding ritual as the structural product motion

Every senior product leader who joins a new company runs a 30-60-90. Day 1 is the read of the room. Week 1 is the early deliverables that prove the new hire knows what they are doing. Day 30 is the first check-in with the boss. Day 60 is the first pushback, the first time the new VP Product disagrees with a path the team had locked in. Day 90 is graduation: the new leader has earned the right to set direction.

The ritual is universal. The artifacts are not. Most senior PMs build the artifacts from scratch every time they change companies, which means every restructure under the agentic operating model triggers a fresh round of artifact-building. Coaches exist (Reforge, Lenny+, executive coaching networks). Diagnostic services exist (consultancies, advisory firms, internal People-Ops audits). Neither answers the same question Dacard does, which is: applied to the work this team is actually doing this week, in this product, on this stack, what is the right Day 1 read, the right Week 1 deliverable, the right Day 60 pushback?

The Dacard onboarding ritual is the structural product motion. Day 1 ships an evidence-led read of the team and product, generated from the operational adapters and a public-web pass. Week 1 ships three deliverables: a maturity diagnostic, a ranked action list, and a coaching brief. Day 30 ships the first check-in artifact: a delta read showing what changed, what the new leader has earned, where the team has moved on the composite ladder. Day 60 ships the pushback brief: the patterns the team is firing, the moves that need to be made, the trade-offs the new VP Product needs to surface to the board. Day 90 graduates: the team has a calibrated baseline, a working coach, and a continuous loop into Linear and Slack.

The structural innovation is not the ritual itself. The ritual is older than software. The structural innovation is the ritual applied to an AI coach. Every senior hire under the new agentic-era operating model needs an evidence-based ramp. DAC is that ramp. Day 90 is when the buyer converts from Free to Pro because graduation is when the leader stops needing diagnostics and starts needing the standing coaching loop.

06 DAC's three product layers (different cut from the moat)

The moat layers (Surface, Workflow, Engine) describe how the product defends itself. The product layers describe how a customer experiences the product across time. The two cuts are not the same thing, and conflating them is a common reading error.

The three product layers map to the three pricing tiers in a clean way. Free covers the DAC-score layer for 30 days plus the new-hire ritual artifacts. Pro at $299 keeps DAC past graduation and unlocks the DAC-intelligence loop with full MCP access, Linear push, Slack pulse, and agent context push. Business at $1,200 extends DAC-intelligence and DAC-coach across the full product organization with 25 products and 10 coach seats. Enterprise at $2,500+ runs DAC-coach across a portfolio with custom benchmarks, SSO, and dedicated success.

The reason this cut matters: a customer's relationship with DAC matures across the three layers. They start with a diagnostic. They graduate into intelligence. They settle into coaching. The retention motion is the product layer transition, not the pricing-tier transition. NRR comes from customers moving up the product-layer ladder, expanding seats, expanding products under coach, expanding to portfolio view.

07 Substrate fusion: the moat at the data layer

The Calibration Pipeline runs on three substrates fused together. Each substrate alone is replicable. The fusion is not.

Operational adapters. 54 integration providers across 12 categories, normalized to a single signal taxonomy of 178 signal types. Today GA-live: GitHub and Linear. OAuth registered: Slack, Jira, PostHog, Figma, Attio. Each adapter implements the same ProviderAdapter interface, normalizes to the shared taxonomy, and feeds the same scoring engine. Adding a 55th provider is mechanical. The taxonomy is the durable asset.

Public web. Every URL provided by a customer triggers a public-web crawl that extracts product-assessment signals, marketing claims, pricing structure, integration set, and changelog cadence. The public-web substrate is the reason DAC can return a Day 1 diagnostic before any operational adapter is connected. It is also the reason the diagnostic can travel virally: the score is real before the customer connects anything.

Tribal knowledge. The substrate that no competitor has and few have considered. Every team carries undocumented context: "this is the cycle we always slip on," "this is the integration we never trust," "this is the customer segment we keep promising and missing." DAC captures that context as Tribal Notes, classifies the underlying signals (Slack messages, decision logs, planning notes, retrospective transcripts), and feeds the classification back into the scoring engine. Phase 5 Slack ingestion is the biggest substrate unlock unbuilt today: an opt-in path that reads the team's actual operating signal, not a sanitized post-hoc summary.

The fusion is what produces the moat. A single-substrate competitor (Jellyfish on dev signal, LinearB on workflow analytics, a consultancy on tribal interview data) covers one of three. Dacard fuses all three through one taxonomy into one calibration pipeline. The fusion is the architectural argument for why platforms cannot ship a competing product simply by adding a scoring dashboard.

Substrate fusion compounds in two directions. Horizontally, every new adapter adds signal coverage without adding new mappings (the 178-type taxonomy is the contract). Vertically, every new substrate (a future support-ticket adapter, a future calendar-density signal, a future agent-interaction trace) plugs into the same scoring engine through the same taxonomy. The architecture is built so that the substrate can grow without the surface having to be rebuilt. That is the discipline that lets the Pattern Library expand and the Calibration Pipeline tighten in lockstep.

08 Distribution surfaces as product

Distribution is not separate from product. Each distribution surface is a product in its own right, designed to compound the others.

MCP server (live). Programmatic tool access to scoring, signals, patterns, and coaching. DAC becomes invocable from any MCP-compatible host, which today means Claude Desktop, Claude Code, Cursor, and a growing list of agent fleets. The MCP server is the surface that makes DAC composable inside other workflows.

Agent Skill (live). A first-class Agent Skill installable into Claude Code and other coding agents. The skill teaches the agent the framework vocabulary, the pattern set, and the calibration model. Reference proof point: Neon hit 80 percent agent-originated provisioning inside 24 months, contributing to its $1B Databricks acquisition. Battery Ventures' April 2026 thesis ("Agent Skills Are the New SDK") is the macro frame. The Dacard agent skill is the operational expression of that thesis: every AI coding agent that installs the skill becomes a distribution channel for the framework.

REST API (live). The conventional surface for engineering teams that want to wire DAC into a custom workflow, a CI gate, or a portfolio rollup. The REST API is the surface that makes DAC institutional.

Claude Code plugin (in queue). The next surface. A first-party Claude Code plugin that exposes DAC's coaching loop directly inside the coding workflow. The plugin closes the loop between framework state and code-time decisions.

Distribution is sequenced for compounding. The MCP server and the Agent Skill make DAC available where senior PMs and engineering leaders already work. The REST API makes DAC institutional. The Claude Code plugin makes DAC ambient. Each surface has its own usage curve. Together, they form the substrate distribution argument: as agent fleets grow, every surface becomes a leveraged channel.

09 The pattern-discovery engine

Three patterns named today is the start, not the ceiling. The pattern-discovery engine is how new patterns are found, validated, named, and shipped. Four steps:

1. Co-occurrence surfacing. Across 88 dimensions, archetype assignments, and outcome telemetry, the engine surfaces co-occurrences passing power thresholds. Most candidate patterns die at this gate. The ones that survive carry statistical weight before any human reviews them.
2. Predictive-validity testing. A surviving co-occurrence is tested against outcome data: does this combination of dimension states predict the outcome it appears to predict? Patterns that pass are promoted. Patterns that fail are discarded.
3. Naming. The qualitative review step. A surviving pattern needs a name a senior PM recognizes on first read. Translation Gap. Fragility Signal. Compound Ready. The naming bar is high. Patterns without a strong name stay in the candidate queue until a name lands.
4. Shipping. Validated, named patterns ship into the product. They appear in the diagnostic. They earn LinkedIn-shareable lines. They expand the addressable buyer-job count. Every shipped pattern is a piece of category creation.

Quarterly pattern-validity reports begin Q3 2026 (internal cohort), public Q4 2026. The reports are the artifact that turns the engine from internal discipline into investor-grade evidence. By Series A, two public reports anchor the moat narrative. By Series B, the report cadence is a moat in itself: a public, recurring artifact that no competitor can match without the same customer base and the same outcome telemetry.

10 Three-year roadmap aligned to fundraise milestones

The roadmap is sequenced by what unlocks the next round of capital. Each phase has a small number of high-leverage milestones, each tied to a fundraise narrative. The Vanta analog holds at every step: continuous, calibrated, packaged-as-software measurement of an organizational property the buyer has no other path to answer.

Now to Q4 2026: seed extension or pre-A

• 8+ validated named patterns shipped, with at least three quantitatively validated and the rest qualitatively validated against design-partner interview signal.
• ≥100 paying customers across Free-to-Pro and Pro-to-Business motions.
• 50% of Pro+ customers have outcome data coverage (Linear, GitHub, Stripe connected).
• First public pattern-validity report shipped Q4 2026.
• Founding engineer hired at month 3, head of customer success hired at month 12.
• SOC 2 Type II in motion.
• 30-60-90 onboarding ritual end-to-end live across all four pricing tiers.
• Slack-as-signal-adapter shipped (the biggest substrate unlock).

Q4 2026 to Q3 2027: Series A

• 12+ validated named patterns shipped, with the discovery engine running quarterly.
• ≥300 paying customers, NRR > 110%.
• Two public quarterly pattern-validity reports.
• Archetype recalibration shipped: the engine adjusts archetype weights from outcome data on a quarterly cadence.
• Claude Code plugin live, in addition to MCP server, Agent Skill, and REST API.
• First AE hired at month 15. CRO discussion deferred until pipeline justifies it.
• Portfolio view in beta for Enterprise.

Beyond Series A: the platform

• Portfolio view at scale: VC firms and corporate-development functions running DAC across portfolios of 20+ companies.
• Custom benchmarks: Enterprise customers can publish private cohort benchmarks to their own boards.
• Programmatic decision-intelligence platform: DAC is the API for "give me a calibrated read on this team, this product, this cycle," invoked by every layer of the modern product organization.
• Public pattern-validity reports as a category artifact: an annual State of AI Maturity report grounded in DAC's customer base.

Each phase compounds the previous. Patterns expand. Outcome data deepens. The pipeline tightens. Distribution widens. The moat thickens at every layer simultaneously, which is what the three-layer moat is designed to do.

11 The long game

The long-game frame is structural. In the agentic-era organization, the senior PM is the player-coach running a cross-functional pod with agent fleets. That role exists at every Series B+ company in software within five years. Every one of those PMs needs an evidence-based ramp on every cycle, every quarter, every restructure. Coaches do not see the work. Frameworks do not adapt. Dashboards do not coach.

DAC becomes the operating coach inside the agentic-era organization. Embedded in every senior PM's workflow. Invoked on every cycle. Pinned in every pod. Surfaced on every restructure. Pushed into Linear, pulsed in Slack, instantiated as an Agent Skill on every coding agent the team runs. The Vanta analog is the right comparable for the category shape (continuous monitoring of an organizational property, packaged as software, sold to a buyer with no other path to the answer). The Vanta analog understates the surface area. Compliance is one ritual. AI maturity is the standing condition of the organization.

The compounding loop is the long game. Pattern Library grows. Calibration Pipeline tightens. Outcome telemetry deepens. Distribution surfaces multiply. Each turn of the loop makes the next turn more valuable. The product becomes the system of record for how the agentic-era team operates, the way Vanta became the system of record for how the modern company stays compliant.

For the strategy and category framing, see the investment memo. For the underlying systems architecture, see technical architecture. Questions: darren@dacard.ai.