Alpha Suite · AI-Stack Maturity Assessment

AlphaStack

Score AI maturity against the actual service blueprint. Two stages: evaluate the gap, then close it.

Stage 1

Evaluate

Stage 2

Enable

Axes

Scale

0–1,000

Version

v0 Strawman

Date

May 2026

AAA AA A BBB BB B CCC Seven bands, same mechanic as AlphaRating v1.3

01 · How to Read This Pack

This is the AlphaStack methodology pack: what the SKU is, how the rubric works, and what Stage 1 intake produces. Not a live assessment.

AlphaStack is a productised AI-stack maturity assessment with two stages. Stage 1 (Evaluate) scores AI maturity against the subject's actual service blueprint and produces a band rating (AAA–CCC), a 10-axis scorecard, and an opportunity matrix. Stage 2 (Enable) builds workflow products that directly address the highest-leverage gaps Stage 1 identifies. A downstream re-score closes the loop with a quantified uplift.

The scoring mechanic is inherited verbatim from AlphaRating v1.3: 10 axes, 0–1,000 scale, two weight profiles (GP-PE and GP-VC), AAA–CCC band breakpoints, Signal + Score + Gap output per axis, atom-backed evidence. What changes is the subject matter: AlphaStack axes score AI-stack maturity as it operates against the actual service delivery blueprint, not company health broadly.

Primary buyer: VC/PE GPs commissioning assessments of portfolio companies for pre-investment AI diligence, value-creation review, or M&A target evaluation. The external sales partner (identity TBD) owns Phase 0 (C-suite education). AlphaStack scope begins at Phase 1 (Evaluate) and Phase 2 (Enable).

Document status

This pack is an internal strawman awaiting Ant review. Rubric axes are candidates, not calibrated against a live engagement. The scoring mechanic is correctly inherited; axis definitions require first-engagement validation before external use.

Primary: locked decisions Strawman: axis definitions, weight profiles Synthesis: intake mechanics

What AlphaStack is not A generic AI maturity audit against industry benchmarks. Commodity tools (McKinsey AI Maturity Assessment, BCG AI Readiness, IBM AI Maturity Model) produce abstract axis scores with no workflow grounding. AlphaStack scores each axis against the specific steps of how the subject actually delivers value. The opportunity matrix is a build brief, not a list of recommendations.

02 · Headline Claims · Read This First

Five things that are true about AlphaStack that are not true of commodity AI assessment tools.

The score is grounded in the subject's actual service blueprint, not an abstract capability checklist.

For each step in how the subject delivers value, AlphaStack asks whether AI is deployed and how effectively. The output maps AI gaps to real workflow steps, which means the opportunity matrix is directly actionable rather than a generic recommendation list. No commodity tool does this.

Stage 2 is scoped from Stage 1's output. Nothing is speculative.

The opportunity matrix is the build brief. Workflow products in Stage 2 directly address the Priority 1–2 gaps identified by the scorecard. The re-score at 3–6 months closes the commercial loop with a quantified uplift number, not a consulting summary. GP gets a before-and-after with a score delta.

The rubric is institutional-grade. AlphaRating v1.3 mechanic, same axes-weights-bands-evidence architecture that rated HedgeServ.

Every score traces to evidence atoms from the intake corpus. Inferred claims are explicitly flagged as INFERENCE with a confidence score below 0.7. The Signal + Score + Gap format per axis provides a structured output that can be cited, not a narrative summary that cannot.

Axis 10 (AI Opportunity Density) is the commercial signal no other tool produces.

Most assessment tools measure current state. AlphaStack Axis 10 measures the forward-looking ROI of closing identified gaps: a company with 30% AI coverage of highly addressable workflows scores higher on Axis 10 than one with 90% coverage and few remaining high-leverage gaps. This is the direct feed for Stage 2 scoping and pricing.

Stage 1 is deliberately lean. One Zoom session. Two-week turnaround.

The intake is one structured Zoom (60–90 min), a pre-session questionnaire, and targeted document uploads. No system installs, no 6-week engagement, no client-side burden beyond the session and uploads. The lean mandate is locked and enforced: if Stage 1 expands, it surfaces to Ant immediately.

03 · The Differentiator

"AlphaStack scores AI maturity against the actual service blueprint, not an abstract capability checklist. The output is an opportunity matrix grounded in real workflows."

Generic axes scored against industry benchmarks

Axes scored against the subject's actual service delivery blueprint

Abstract capability inventory

Workflow-step coverage: which steps have AI and how effectively

Generic recommendations

Opportunity matrix ranked by uplift ROI, directly mapped to workflow gaps

Report delivered, engagement ends

Stage 2 builds workflow products against the matrix; re-score quantifies uplift

No institutional scoring pedigree

AlphaRating v1.3 mechanic: same rubric that produced institutional-grade ratings on live engagements

04 · Rubric Overview

10 axes. Two weight profiles. Seven bands.

The scoring mechanic is inherited verbatim from AlphaRating v1.3. What changes is the subject matter: AI-stack maturity, grounded in the service blueprint.

#	Axis	What it measures	Wedge
01	AI-Workflow Integration	AI coverage and depth across each step of the service delivery blueprint. The primary differentiator axis.	Wedge
02	Data Infrastructure & Moat	Data quality, labeling discipline, governance, and proprietary training advantage over competitors.	Wedge
03	AI Stack Architecture	Model layer choices, build vs. buy strategy, infrastructure maturity, vendor lock-in risk.
04	Team AI Fluency	Adoption breadth and depth across the organisation; AI ownership structure; upskilling maturity.
05	AI Velocity	Iteration speed, experimentation cadence, time from identified opportunity to live deployment.
06	AI Governance & Risk	Hallucination controls, data privacy posture, IP risk, regulatory exposure (AI Act, sector rules).
07	Commercial AI Leverage	Measurable business impact: cost reduction, revenue enablement, quality improvement, throughput.	Wedge
08	AI Strategy Coherence	Coherent AI roadmap vs. opportunistic adoption; vision-to-execution gap.
09	AI Durability	Structural defensibility of AI advantage against commoditisation and AI-native entrants.
10	AI Opportunity Density	Forward-looking: how much unlocked AI value sits in identified workflow gaps, relative to capture effort. Stage 2 feed.	Wedge

Weight Profiles

Overall Bands (0–1,000)

AAA / AA

950–1,000 / 850–949

Exceptional or institutionally strong across all weighted axes. No material structural gaps.

A / BBB

750–849 / 650–749

Strong or investment grade. Gaps present but not structural. Manageable with targeted intervention.

BB / B

550–649 / 400–549

Below investment grade. Gaps are structural. Significant risk across multiple axes without enablement.

CCC

0–399

Fundamental concerns. High risk. AI posture is a liability, not an asset, at current state.

Per-Axis Colour (0–100)

Blue

80–100

Category-leading. No material gaps.

Green

60–79

Solid. Gaps present but not structural.

Amber

40–59

Mixed. Gaps are meaningful, addressable.

Red

0–39

Structural weakness. Material risk.

05 · Axis Deep-Dives

Definitions, scoring logic, and what good vs. weak evidence looks like per axis. All 10 axes shown. Status: strawman, awaiting first-engagement calibration.

01 AI-Workflow Integration Wedge

The primary differentiator axis. For each step in the subject's service delivery blueprint, is AI deployed and how effectively? Scores breadth (what percentage of workflow steps have AI embedded), depth (peripheral/experimental vs. optimised/measured), and evidence of actual productivity or quality impact. Level 0 = no AI. Level 4 = AI is the delivery mechanism for that step.

Signal: what scores up

Verified AI deployment at named workflow steps. Toolchain inventory + workflow documentation confirming coverage. Measured productivity or quality impact at deployed steps. High breadth (majority of delivery steps have AI embedded at depth).

Gap: what reduces the score

AI deployed at peripheral steps only (formatting, summarisation). No measurement of productivity impact. AI-dark steps in high-value parts of the delivery blueprint. Self-reported claims with no verifiable artefact backing.

Self-reported claims discounted; verifiable artefacts (toolchain inventory, workflow docs, sample outputs) weighted up. Stage 1 lean: infer from interview + uploads.

02 Data Infrastructure & Moat Wedge

Data quality, labeling discipline, governance maturity, and availability for model training or retrieval. A strong data moat = historical, proprietary, clean data that creates an AI training advantage competitors cannot replicate. Gaps weighted by how AI-addressable the workflows are: a data-poor company in a high-AI-leverage workflow is a structural risk.

Signal: what scores up

Proprietary structured data accumulated over time. Active use of data for model training or RAG retrieval. Governance and labeling discipline. Data advantage competitors cannot replicate quickly.

Gap: what reduces the score

Reliance on generic public datasets or third-party APIs with no proprietary layer. Unstructured or poorly governed data. No data flywheel. Competitor data moat is as strong or stronger.

03 AI Stack Architecture

Model layer choices (frontier API vs. fine-tuned vs. local), build vs. buy strategy, infrastructure maturity (observability, CI/CD for model updates, evaluation rigor, cost controls). Vendor lock-in risk. Is the architecture aligned with the subject's workflow requirements (latency, privacy, scale) or generic/accidental?

Signal: what scores up

Architecture aligned with workflow requirements. Evaluation loops in place. Observability and cost controls active. Build/buy strategy with clear rationale. Low vendor lock-in or managed dependency.

Gap: what reduces the score

Accidental architecture (tools selected without strategy). No evaluation discipline. Vendor concentration risk. Latency or privacy constraints unaddressed. No observability into model performance or cost.

For lean Stage 1: inferred from toolchain list and workflow description. Flag as low-confidence if not verifiable from uploads.

04 Team AI Fluency

Who uses AI tools, how deeply, and at what levels of the organisation. Adoption breadth (across functions) and depth (beyond chat interfaces into embedded workflows). AI ownership and accountability structure. Upskilling and onboarding maturity. C-suite AI literacy signals. Score reflects the gap between AI capability available and AI capability actually absorbed.

Signal: what scores up

Broad adoption across functions. Embedded workflows, not just chat interface use. Named AI ownership. Active upskilling programme. C-suite with working knowledge of AI tooling, not just stated ambition.

Gap: what reduces the score

AI concentrated in one function or team. Adoption superficial (chat interfaces only, no embedded tooling). No named AI ownership. Upskilling ad hoc or absent. C-suite AI narrative disconnected from ground-truth adoption.

05 AI Velocity

How fast is the subject iterating on AI deployments? Learning cycles, experimentation cadence, time from identified opportunity to live deployment. Evaluation discipline: are they measuring whether AI is working? Leading indicator: a named process for trialling and retiring AI tools vs. ad hoc adoption.

Signal: what scores up

Named experimentation process. Short cycle times from opportunity to deployment. Active tool retirement (retiring things that don't work is a velocity signal). Evaluation rituals in place.

Gap: what reduces the score

AI adoption ad hoc, no process. Long lag from idea to deployment. Tools accumulate without evaluation. No mechanism to retire failing experiments. Evaluation discipline absent.

For lean Stage 1: inferred from team description and workflow documentation. Flag as low-confidence if not verifiable.

06 AI Governance & Risk

AI-specific risk management: hallucination controls and output validation processes, data privacy posture (PII in prompts, vendor data-handling terms), IP and confidentiality risk, regulatory exposure (EU AI Act, sector-specific rules). Vendor concentration risk. Score reflects maturity of the governance layer: policies exist vs. policies are enforced vs. policies are tested.

Signal: what scores up

Output validation processes at human-facing AI touchpoints. Reviewed vendor data-handling terms. Documented PII policy for AI usage. Regulatory exposure mapped and managed. Governance tested, not just documented.

Gap: what reduces the score

No hallucination controls at customer-facing outputs. PII entering third-party model prompts without review. Regulatory exposure unassessed. Vendor terms not reviewed. Governance exists on paper only.

07 Commercial AI Leverage Wedge

Is AI measurably moving business metrics: cost reduction, revenue enablement, quality improvement, throughput increase? Evidence: unit economics (cost-per-task AI vs. human), documented time savings, customer-facing AI features with adoption data. Scores based on verifiable commercial impact. Zero-impact deployments (AI running but not moving a metric) score below baseline.

Signal: what scores up

Documented cost-per-task comparison. Measured time savings with baseline. Revenue attributable to AI features. Customer-facing AI adoption metrics. Unit economics showing AI delivering at lower cost than human alternative.

Gap: what reduces the score

AI running but no metric instrumentation. Time savings described but no baseline. Revenue attribution absent. "AI is faster" without a number. Discipline gap: the capability exists, the measurement does not.

Stage 1 lean: inferred from workflow description and any metrics shared. Flag as low-confidence if purely self-reported with no documentation.

08 AI Strategy Coherence

Is there a coherent AI roadmap grounded in the service blueprint, or is AI adoption opportunistic and uncoordinated? Vision-to-execution gap: stated AI ambitions vs. observable evidence of progress toward them. C-suite alignment vs. ground-truth reality. Score discounts visions not supported by verifiable execution signals, same mechanic as AlphaRating v1.3 Vision & Feasibility axis.

Signal: what scores up

Named AI roadmap tied to service blueprint steps. Execution evidence matching stated ambitions. C-suite narrative consistent with ground-truth adoption. Investment in AI aligned with roadmap priorities.

Gap: what reduces the score

AI ambitions stated, execution absent. Opportunistic adoption with no coherent strategy. C-suite describes an AI-first company; delivery team describes ad hoc tool use. Roadmap exists but investment doesn't follow it.

09 AI Durability

Will the subject's AI advantage survive the next wave of AI commoditisation? Data moat quality, proprietary model investments, workflow lock-in from AI integration, defensibility against AI-native entrants. Model-layer independence vs. fragile dependency on a single vendor's capability cliff. Score reflects structural defensibility, not current capability.

Signal: what scores up

Proprietary data moat that compounds over time. Workflow lock-in: AI so embedded that competitors cannot easily replicate the delivery model. Multi-model strategy reducing vendor cliff exposure. Structural advantage, not capability that commoditises next quarter.

Gap: what reduces the score

AI advantage entirely from frontier API access (easily replicated). Shallow data moat. No workflow embedding; AI is an add-on, not the delivery mechanism. Fragile vendor dependency. AI-native entrants could match the posture within 12 months.

10 AI Opportunity Density Wedge

The forward-looking axis. How much unlocked AI value sits in identified workflow gaps, relative to the effort to capture it? A company with 30% AI coverage of highly AI-addressable workflows scores high (dense opportunity). A company with 90% coverage and few remaining high-ROI gaps scores lower. Directly feeds the Opportunity Matrix and Stage 2 scoping.

Signal: what scores up

High-AI-addressable workflow steps that are currently AI-dark. Clear effort-to-uplift ratio on identified gaps. Priority 1–2 gaps accessible without major infrastructure rebuild. Stage 2 engagement is commercially attractive to both parties.

Gap: what reduces the score

Already high AI coverage with few remaining high-ROI steps. Workflow gaps exist but are low-addressability (regulatory constraint, data unavailability, high infrastructure cost). Stage 2 opportunity is thin or very long-horizon.

This is the only forward-looking axis. Validity requires calibration across live engagements: confirm Axis 10 can be scored consistently without becoming subjective.

06 · Stage 1 Intake Mechanics

One Zoom. Two weeks. A scored pack.

The lean mandate is locked. Stage 1 is not a 6-week engagement. If it expands beyond the parameters below, it surfaces to the orchestrator immediately.

Format

Intake session

One structured Zoom session (60–90 min) with the subject: CTO, COO, or CIO. Structured interview per kickoff template. Pre-session questionnaire async (30–60 min subject time). No system install required.

Timeline

Two-week turnaround

From completed intake (session + uploads received) to pack delivery. Strawman estimate; confirm with Ant before first engagement. Pack delivered as a password-gated subdomain.

Document Uploads

Targeted artefacts

Process or workflow documentation
Toolchain inventory (AI tools in use, where, by whom)
Team structure and AI ownership
Sample workflow artefacts (optional, highest value if available)

Output: Stage 1 Pack

What the GP receives

AlphaStack band: AAA–CCC
10-axis scorecard: Signal + Score + Gap per axis
Per-axis colour: Blue / Green / Amber / Red
Opportunity matrix: gaps ranked by uplift ROI
Delivery: password-gated pack subdomain

Evidence Protocol

Atom-backed scoring

Every score traces to evidence atoms from the intake corpus. Inferred claims (not verifiable from artefacts) are explicitly flagged INFERENCE with a confidence score below 0.7. No fabricated quotes. All inputs under NDA.

Stage 2 (if enabled)

Workflow products + re-score

GP and subject agree which Priority 1–2 gaps to address. Each selected gap becomes a purpose-built workflow product on the Black Book framework. Re-score at 3–6 months: relevant axes re-scored, delta is the commercial proof point.

07 · Open Variables

Decisions required before any external use.

The following variables are open. None of these are resolved by this pack. AlphaStack cannot be presented to a partner or prospect until these are Ant-reviewed and confirmed.

Variable	Status	Notes
Final SKU name	TBD	Working name "AlphaStack" not locked. Ant decides.
Subdomain slug	TBD	Do not auto-pick. Awaiting Ant.
Brand house	Deferred	Alpha Suite is working housing. Alpha Suite vs. FullyBaked decision deferred until post-v0 one-pager.
Pricing model	TBD	Not defined. Ant + partner decision.
External partner identity	TBD	Not disclosed. Partner owns Phase 0 (C-suite education).
Weight profile selection logic	TBD	GP-PE vs. GP-VC decision at kickoff. Third profile possible based on partner buyer context.
Stage 1 timeline (2-week estimate)	Strawman	Confirm with Ant before first engagement.
Axis 10 validity	Unvalidated	Forward-looking axis. Confirm it can be scored consistently without becoming subjective across live engagements.
Intake fallback (no service blueprint)	Undefined	What happens when the subject has no documented service blueprint? Define the intake fallback before first engagement.

Verdict: Below Bar, Known Gaps This pack is an internal strawman. The rubric mechanic is correctly inherited from AlphaRating v1.3. Axis definitions are candidates, not calibrated against a live engagement. This document is ready for Ant review; it is not ready for partner or external-prospect use without his sign-off and first-engagement calibration.