Allometry OS · Master PRD

Product Requirements Document

The underwriting and control layer for physical revenue. Layer 1 — 16 Action Modules. Layer 2 — The Platform. Single source of truth.

Version 2.0 Architecture Two-Layer Updated April 2026 Author Taylor Gendron Status Pre-seed · Design Partner Build

01 Product Overview

What is Allometry?

Allometry is the underwriting and control layer for physical revenue. We help asset-heavy operators (HVAC, EV, telecom, security, pallet manufacturing, smart lockers, solar) decide which work to pursue, how to price it, and how to execute it — with margin visibility at the address level, before the job starts. ABM × CPQ, applied to asset-heavy operators. Every account gets a Pulse.

Core Primitive — the Address

The address (location / site / deployment unit) is the atomic object. Every quote, job, deal, contract, and health signal relates to it. Unlike CRMs (contact-first) or ERPs (order-first), Allometry is address-first. A 50-location HVAC operator has 50 different unit economics, and the Pulse tracks every one.

The Pulse

One scoring primitive per account, per address. Composite score: margin · EV · risk · expansion. Continuously updated with every closed job, every contract change, every supplier shift. The moat: every module trains it. At 10K jobs we set the industry margin benchmark. At 50K we are the underwriting layer for every deployment in North America.

Two-Layer Architecture

The product is structured in two layers. Modules complete work. The Platform is how you see, call, connect, brand, host, and extend that work.

Layer	Role	What It Contains	Pricing
Layer 1	Action Modules — complete work	16 modules across 4 loops (Price · Deploy · Expand · Govern)	Per-module / per-loop / full OS
Layer 2	Platform — how you access work	Analytics · API · Integrations · Whitelabel · Infrastructure · Extensions	Included with every tier

Key consequence: Analytics, API, and Integrations are not modules. They are platform capabilities, included by default. You pay for work completion; dashboards, APIs, and integrations come with the product.

The 4 Loops

Loop	Risk Dimension	Question	Modules
Price	Financial	"Can you price it right?"	01 Quoting · 02 Margin · 03 Cost · 04 Contract
Deploy	Operational	"Can you deliver & keep it running?"	05 Location · 06 Job Costing · 07 Design · 08 Asset Intel
Expand	Growth	"Which accounts are worth winning?"	09 Deal Room · 10 Outbound ABM · 11 Health · 12 Forecasting
Govern	Market & Governance	"Is the system trustworthy?"	13 Market Intel · 14 Compliance · 15 Risk Scoring · 16 Vendor Intel

The Flywheel

Price → Deploy → Expand → Govern → Price. Each loop feeds the next. Quoting data trains cost models. Job costing reveals expansion-worthy accounts. Expansion signals refine targeting. Governance surfaces outliers. Every closed job sharpens the Pulse.

Target User (ICP)

Asset-heavy operators with 5-2,500 locations, $5M-$500M revenue, multi-address customers, variable margin per location, long sales cycles. Initial wedge: Quebec HVAC, solar, pallet manufacturing, smart locker networks. Expansion: SF, NYC; $10M-$50M operators as primary ICP.

02 Design Principles

Address-first, not contact-first. Every view, every decision, every metric anchored to the deployment unit. The address is the row in every table.
Modules complete work. Platform is how you access it. 16 units of work completion. 6 horizontal capabilities. Never confuse the two.
Every decision gets underwritten. Not reported. Not logged. Underwritten — scored against margin and EV before commit.
Before the job starts. Margin visibility must land before execution commits resources. Month-end is too late.
Agents handle 90%, humans validate 10%. Autonomous by default. Humans validate exceptions — relationships, strategy, outliers.
Loops, not features. Users think in workflows, not module lists. Navigation organizes around risk dimensions.
Data creates optionality. Every quote creates data that feeds every other module. Users start with 1 loop and expand naturally because the data is already there.
No drift. What you quote is what you deploy. Single source of truth from proposal to invoice. Address-level accuracy, always.
Embedded, not adjacent. Allometry sits on top of existing tools (CRM, ERP, FSM). We govern decisions the others can't make.

03 System Architecture

Tech Stack

Framework	Next.js 16 · React 19 · App Router
Language	TypeScript
Database	PostgreSQL · Drizzle ORM · pgvector
Auth & SSO	WorkOS (SAML, OIDC, Directory Sync)
Hosting	Vercel · edge-first · multi-region
Styling	Tailwind CSS 4 · shadcn/ui
AI	Anthropic Claude (via SDK) · multi-stage RAG · tool use
Maps	Leaflet · react-leaflet
Charts	Recharts · D3 · Tremor
Email	Resend · Nodemailer
Payments	Stripe
Monitoring	Sentry · OpenTelemetry
Compliance	SOC 2 Type II · Quebec Law 25 · GDPR · data residency (CA/US/EU)

Multi-Workspace Isolation

Each customer gets an isolated workspace. Workspace-scoped database queries. Row-level security enforced at the data layer. Users can belong to multiple workspaces with distinct roles. Tenant_id is injected from session context on every read/write.

Core Data Model — Address as Root

Location (address) → has many Quotes · Jobs · Deals · Contracts · Health Scores · Signals
Account → has many Locations · has one Pulse
Pulse → per-account composite score (margin, EV, risk, expansion) · time-series
Quote → belongs to Location · has many LineItems · linked to Contract
Job → belongs to Location · has actual costs · linked to Quote
Asset → belongs to Location · tracked through lifecycle (install → uptime → decommission)
Deal → belongs to Location · has engagement tracking · powered by Pulse score
Contract → belongs to Account · applies terms to all Quotes
Signal → belongs to Location or Account · feeds Pulse

04 6 Data Primitives

The 16 action modules are surfaces on 6 shared engineering systems. Adding a module = adding a lens, not a codebase.

System 1 · Pricing Engine

Margin models, cost structures, volume tiers, contract terms. Powers Modules 01-04 (Quoting · Margin · Cost · Contract).

Tables: pricing_rules · cost_items · contracts · quote_line_items · margin_models

System 2 · Territory / Geo

Geocoded addresses, route optimization, coverage analysis, deployment scheduling, asset lifecycle tracking. Powers Modules 05, 06, 08 (Location · Job Costing · Asset Intelligence).

Tables: locations · territories · deployments · jobs · assets · asset_telemetry

System 3 · Intelligence Layer

Multi-stage RAG, scoring models, predictive analytics, account intelligence. Powers Modules 10, 11, 12, 13, 15, 16 (Outbound ABM · Customer Health · Demand Forecasting · Market Intel · Risk Scoring · Vendor Intel).

Tables: signals · health_scores · demand_forecasts · enrichments · risk_flags · vendor_scores

System 4 · Content / Render

Proposals, branded portals, floorplans, AI renders. Powers Modules 07, 09 (Deployment Design · Deal Room).

Tables: deal_rooms · proposals · designs · assets_uploaded · renders

System 5 · Governance Plane

Policy engine, approval flows, regulatory frameworks, audit trails. Powers Module 14 (Compliance) — all governance, approval, audit, and regulatory concerns fold into one module surface.

Tables: policies · approvals · audit_log · compliance_events · regulatory_frameworks

System 6 · Pulse Core Primitive

The scoring primitive beneath every module. Composite scores (margin · EV · risk · expansion) per account, per address, per commercial decision. Trained on every closed job. Retrained on realized outcomes.

Tables: account_pulse · address_scores · decision_log · training_signals · model_versions

Platform capabilities (Analytics, API, Integrations, Whitelabel, Infrastructure, Extensions) are not data primitives. They are Layer 2 — horizontal capabilities on top of the 6 systems. See Section 9.

05 Layer 1 · Loop 1 — Price Financial Risk

"Can you price it right before you commit?" — Every deployment starts with a pricing decision. This loop ensures no quote leaves the system without being underwritten.

Quoting Engine

P1 · Wedge

Purpose: Configure, price, quote in 60 seconds. Every quote scored against margin and EV before it leaves the system. The door-opener — the demo moment — the entry wedge.

Core Features

Product selector — Search/browse SKU catalog, add to quote, set quantities
Dynamic pricing — Auto-applies volume tiers, regional adjustments, contract terms
Margin calculator — Real-time margin % per line item, color-coded (green/amber/red)
Margin floor enforcement — Hard blocks below threshold; soft nudges near threshold
Pulse scorecard — Margin · EV · risk · expansion per quote, per address
One-click proposal — Generates PDF + hosted deal-room link
Shareable URL — Unique link with engagement tracking

Agent Behavior

Auto-checks inventory availability and supplier lead times before pricing
Applies active MSA/contract terms automatically
Flags competitive price matching opportunities
Routes for approval if margin below threshold

Inputs

SKU catalog · pricing rules · cost data · contract terms · inventory · ICP score · account Pulse

Outputs

Quote object · line items · margin calc · Pulse scorecard · proposal PDF · deal room · approval request

Key Metrics

Time to quote (<60s) · quotes/user/day · avg margin · quote-to-close rate

Powers

System 1 · Pricing Engine · writes to Pulse Core

Margin Protect

Purpose: Defend margins on every deal. Real-time guardrails at quote-time, not month-end.

Core Features

Real-time guardrails — Flag margin erosion as it happens, not after
Price floor enforcement — Hard block below configurable threshold
Leakage pattern surfacing — Identify systematic margin leaks across book
Exception routing — Sub-threshold deals auto-route to approvers
Margin trend analytics — Per customer, product line, region, rep, time period

Inputs

Line items · cost data · target margins · rep/region overrides · historical realized margins

Outputs

Margin flags · approval triggers · leakage reports · adjusted floors

Cost Engine

Purpose: Unified cost truth. SKU-level + labor + logistics + supplier intelligence. The real number behind every line item.

Core Features

SKU-level inventory intelligence — Costs, lead times, quantity-on-hand, reorder points
Supplier scoring — Reliability, price volatility, lead time variance (feeds Vendor Intel)
Cost curve tracking — Forward cost projections, commodity/material trends
Labor cost modeling — Per region, per crew type, per deployment complexity
Reorder signals — Predictive triggers based on deployment pipeline

Inputs

Supplier feeds · material prices · labor rates · logistics costs · vendor scores

Outputs

Cost floor per SKU · forward cost projections · reorder signals · supplier recommendations

Contract Intelligence

Purpose: Contracts that quote themselves. Active MSAs drive quoting logic automatically — no manual term lookups.

Core Features

MSA parsing — AI extracts pricing terms, volume tiers, SLAs, payment terms from uploaded PDFs
Auto-apply at quote-time — Quoting Engine inherits active contract logic automatically
Renewal tracking — Surface expiring contracts 90/60/30 days out
Negotiation signals — Flag under-performing contracts vs peer benchmarks
Clause library — Reusable templates (pricing, termination, SLA)

Inputs

Uploaded MSAs · usage data · contract templates · benchmarks

Outputs

Structured contract terms · renewal alerts · negotiation recommendations

06 Layer 1 · Loop 2 — Deploy Operational Risk

"Can you deliver it — and keep it running?" — What happens after the quote is signed. No drift between what you promised and what you execute.

Location Intelligence

Purpose: Territory as a first-class object. Geocoded addresses, scored, mapped, optimized.

Core Features

Geocoded coverage zones — Address-level territory boundaries
Location scoring — Per-address attractiveness (margin potential, complexity, access)
White-space identification — Where you should be but aren't
Overlap detection — Reps stepping on each other
Route optimization — Day-of routing for install/service crews

Job Costing

Purpose: The truth loop. Actual vs. estimated on every job, every location. Tells you whether your quotes were right.

Core Features

Real-time P&L per job — Live cost tracking during deployment
Actual vs estimated — Variance flagged per SKU, per labor line
Margin retraining signal — Realized margins feed back into Pulse scoring
Job-level profitability — Per location, per rep, per crew, per product
Cost category variance — Materials vs labor vs logistics vs overhead

Deployment Design

Purpose: Design it, visualize it, sell it. AI-generated floorplans, equipment layouts, deployment previews.

Core Features

Floorplan optimization — AI generates optimal equipment placement
AI render generation — Visual mockups of completed deployment
Site mockups — Proposal-grade visuals for deal rooms
Design libraries — Reusable templates per vertical / configuration
Export to proposal — Embeds into Deal Room + PDF proposals

Asset Intelligence

Purpose: Deployment lifecycle — from install through uptime through decommission. What you deployed, how it's performing, when to act.

Core Features

Install scheduling — Intelligent sequencing by margin priority, crew skill, resource availability
Activation tracking — Confirm every asset came online; flag slow-to-activate
Uptime scoring — Per-asset utilization rates, rolled up to account and territory
Maintenance triggers — Predictive signals flagging service before failure
SLA risk flags — Detect trends that will breach contracted SLAs before they do
Decommission & renewal signals — Time-to-end-of-life + upgrade/renew recommendations
Calendar sync — Bi-directional Google/Outlook for install + service windows

Parallel module: Asset Intelligence (08) tracks deployed assets. Vendor Intelligence (16) tracks suppliers. Two lenses on the physical world — internal vs upstream.

07 Layer 1 · Loop 3 — Expand Growth Risk

"Which accounts are worth winning — and keeping?" — How you win more and grow what you have. Turns deployment data into pipeline weighted by margin potential.

Deal Room

Purpose: Where the quote becomes a sale. Branded portal, engagement tracking, e-signatures — all tied to the Pulse.

Core Features

Auto-generated proposals — AI creates branded one-pagers from quote data
Hosted deal room — Unique URL per deal with all assets
Engagement tracking — Views, time-on-page, downloads — real-time
Embedded ROI calculator — Prospect projects their own returns
E-signatures — Built-in signing (no external tool)
Magic link auth — Prospects access without creating an account

Outbound ABM

Purpose: The agent tells you who to call. Account-based prospecting scored by margin potential, not just firmographic fit.

Core Features

EV-scored accounts — Pulse projects margin potential per target account
Address-level enrichment — Every target account's locations geocoded, sized, mapped
Stakeholder mapping — Decision-makers, influencers, champions per account
Multi-channel sequences — Email → LinkedIn → Call, AI-personalized
Intent detection — Job postings, funding events, contract expirations as buy signals
Retrain loop — Won deals + realized margin feed back into the targeting scorer

Customer Health

Purpose: Know before they churn. Predict before they expand. Composite account health from quoting frequency, margin trends, engagement, support load, renewal timing.

Core Features

Composite health score — Updated daily, per account, time-series
Churn prediction — 90-day advance warning with intervention recommendations
Expansion signals — Ready for multi-site rollout? Pulse flags it
Account timeline — Full history of interactions, quotes, deployments, health changes
Agent intervention recs — Specific next-best actions per at-risk account

Demand Forecasting

Purpose: Predict before they ask. Turn historical deployment patterns into forward-looking demand signals per region, product, season.

Core Features

Seasonal demand curves — Historical patterns by region, product, industry
Pipeline prediction — Revenue forecast with confidence intervals
Inventory pre-positioning — Where to stage inventory before demand spikes
Replenishment triggers — Auto-alerts when demand indicates restocking
What-if modeling — Scenario planning for territory expansion, pricing changes

08 Layer 1 · Loop 4 — Govern Market & Governance Risk

"Is the system trustworthy and accountable?" — The layer that monitors external signals, governs internal decisions, scores portfolio risk, and tracks supply-chain health. Four corners: External · Policy · Portfolio · Supply.

Market Intelligence

Purpose: Outside-world signals. Competitive pricing, win/loss patterns, market shifts, answer engine visibility. The eyes and ears of the OS.

Core Features

Competitive monitoring — Track competitor deployments, pricing changes, territory moves
Win/loss analysis — Pattern recognition across deal history
Market signals feed — Aggregated from public sources, job postings, filings
Answer engine optimization — Monitor how you appear in AI-powered search (GPT, Claude, Perplexity)
Competitive pricing alerts — Flag when competitors undercut your territory

Compliance

Purpose: The policy plane. Every decision the Pulse makes is gated, logged, and compliant by construction. Replaces separate "Governance" + "Audit" + "Regulatory" concerns.

Core Features

Governance rules — Natural-language policies: "If margin < 25%, require VP approval" · configurable by role, territory, deal size
Approval workflows — Hard blocks, soft nudges, escalation paths · Slack/email integration
Audit trail — Every decision logged, every quote versioned, every approval tracked — immutable event log
Regulatory frameworks — Built-in Quebec Law 25, GDPR, SOC 2 · data residency controls · consent tracking
Contract lifecycle governance — Auto-enforce MSA terms, flag expiring contracts, trigger renewals
Policy enforcement engine — Runs policies at quote-time, deploy-time, invoice-time
Role-based permissions — Admin · Manager · Rep · Viewer, scoped by workspace, territory, account
Board-ready reporting — Governance posture exports for compliance audits and investor reporting

Absorbs: Governance rules · audit trail · regulatory posture · approval flows · contract lifecycle governance. One action module instead of three overlapping surfaces.

Risk Scoring

Purpose: Portfolio-level risk detection. Flag outlier accounts, concentration risk, and anomalies before they hit the P&L.

Core Features

Outlier detection — Accounts and deals whose economics drift from the portfolio norm
Concentration analysis — Revenue/margin/geo concentration alerts (10% customer rule, single-region exposure)
Anomaly flags — Unusual quote behavior, pricing deviation, margin drift per account
Stress testing — What-if modeling — "what happens if top-3 accounts churn?"
Account risk heatmap — Portfolio risk concentration by account, region, vertical
Trend surfacing — Rolling-window detection of degrading health before churn

Vendor Intelligence

Purpose: Upstream risk. Score supplier reliability, monitor SLAs, detect concentration and vendor failure risk before it affects deployments.

Core Features

Supplier scoring — Per-vendor reliability, on-time rate, quality rate, price volatility
SLA monitoring — Track delivery SLAs at line-item level; alert on breach risk
Concentration risk — Flag single-source exposure on critical SKUs
Lead time intelligence — Real-time lead-time tracking per supplier, per SKU
Alternative sourcing — Surface backup vendors when primary is at risk
Cost feedback loop — Vendor changes feed back into Cost Engine (03) for forward margin protection

Parallel module: Asset Intelligence (08) tracks deployed assets. Vendor Intelligence (16) tracks suppliers. Two lenses on the same physical world — internal vs upstream.

09 Layer 2 — The Platform Horizontal · Included

The 16 action modules are Layer 1 — what you buy. The Platform is Layer 2 — how you see, call, connect, brand, host, and extend those modules. Included with every tier, horizontal across all 16 modules.

Analytics See the work

Unified intelligence across all 16 modules — one dashboard, one dataset
Executive views (revenue · margin · pipeline · territory · risk on one screen)
Predictive forecasting with confidence intervals
Drag-and-drop report builder with filters, grouping, scheduled delivery
Drill-down from any metric to location, account, product, time period
Embed in external dashboards · export to PDF / CSV

API Call the work

RESTful endpoints for every entity across all 16 modules
Webhook subscriptions for real-time event notifications
OAuth 2.0 authentication · scoped API keys · per-workspace rate limits
Rate limits: 1K/day (Startup) → 5K/day (Growth) → Unlimited (Enterprise)
The same API that dispatches humans today dispatches humanoids tomorrow

Integrations Connect the work

Category	Integrations	Sync
CRM	Salesforce · HubSpot	Bi-directional
ERP	NetSuite · SAP	Bi-directional
FSM	ServiceTitan · Jobber	Read
Accounting	QuickBooks · Xero	Push
Calendar	Google · Outlook	Bi-directional
Communication	Slack · Teams · Email	Notifications
Enrichment	Apollo · Clearbit · Snitcher	Pull
Payments	Stripe	Bi-directional

Whitelabel Brand the work · Enterprise

Custom domain · logo · colors · email templates
Fully branded client portal
Partner management with tier-based permissions
Revenue sharing & commission tracking for channel partners

Infrastructure Host the work

SOC 2 Type II — certified
WorkOS-powered SSO (SAML, OIDC, Directory Sync)
Multi-tenant isolation · workspace-scoped queries · row-level security
Data residency — Canada / US / EU
High availability — 99.9% SLA, multi-region failover
Encryption at rest (AES-256) + in transit (TLS 1.3)
Compliance frameworks (Quebec Law 25 · GDPR · HIPAA optional for enterprise)

Extensions Extend the work

SDKs — Python · Node.js · TypeScript · Go
Webhooks & Events — React to every Pulse update in real time
Sandbox — Dev environment with test data; iterate without hitting prod
Custom Modules — Build your own action modules on top of the Pulse primitive
Marketplace (future) — Partner-built vertical modules (beehives, palettes, EV, smart lockers)
The moat: the Pulse is the substrate; anyone can extend the work, no one else can replace the substrate

10 User Roles & Permissions

Role	Access	Typical User
Admin	Full: settings, users, billing, all modules, all data	CEO, Founder
Manager	All modules · approval authority · team management · reports	VP Sales, VP Ops
Rep	Quoting · deal rooms · territory view · own pipeline	Sales Rep, Account Exec
Ops	Job costing · asset intel · location mgmt · crew assignment	Operations Manager, Dispatcher
Finance	Margin analytics · cost engine · contract mgmt · compliance reporting	CFO, Controller
Compliance	Compliance module · audit log · regulatory posture	DPO, Risk Officer
Viewer	Read-only dashboards & reports	Board Member, Investor
Partner	Partner portal · own deals · commissions · co-branded deal rooms	Channel Partner
API	Programmatic access via scoped API key	Developer, Integration

11 UX Principles

Loop-first navigation. Main nav organizes by Price · Deploy · Expand · Govern — not by module name. Users think in workflows.
Location is always in context. Every screen shows which address(es) are active. The location picker is always one click away.
Pulse is persistent. A scorecard — Margin / EV / Risk / Expansion — is visible on every quote, deal, and account view.
Agent activity feed. A persistent sidebar shows what agents are doing: quotes generated, alerts fired, approvals pending. Makes the "work getting done" visible.
Progressive disclosure. Start with 1 loop (typically Price). Reveal complexity as the user expands. Don't show 16 modules on day one.
Platform is invisible. Analytics, API, Integrations live in "Settings" — not in the primary nav. The user's focus stays on action.
Mobile-ready for field. Rep view works on mobile: quick quote, location lookup, deal room access, asset status.
Dark mode default. Matches brand. Light mode available with one click.

12 Build Roadmap

Phase 1 — Wedge (Current → Month 3)

Goal: First design partners live. Prove the quoting wedge.

Module	#	Priority	Status
Quoting Engine	01	P1	Live (DPs)
Margin Protect	02	P1	Live (DPs)
Location Intelligence	05	P1	Live (DPs)
Deal Room	09	P1	Live (DPs)
Compliance	14	P1	Scaffolded
Platform · P1 Analytics	—	P1	Scaffolded
Platform · P5 Infrastructure (SOC 2 base)	—	P1	In progress

Phase 2 — First Loop (Month 3-6)

Goal: Complete Price loop. Design partners expanding into Deploy & Expand.

Module	#	Priority	Status
Cost Engine	03	P2	Planned
Contract Intelligence	04	P2	Planned
Job Costing	06	P2	Planned
Asset Intelligence	08	P2	Planned
Outbound ABM	10	P2	Planned
Customer Health	11	P2	Planned
Market Intelligence	13	P2	Planned
Risk Scoring	15	P2	Planned
Vendor Intelligence	16	P2	Planned
Platform · P2 API	—	P2	Planned
Platform · P3 Integrations	—	P2	Planned

Phase 3 — Full OS (Month 6-12)

Goal: All 4 loops live. Full platform. $1M ARR. Seed raise.

Module	#	Priority	Status
Deployment Design	07	P3	Planned
Demand Forecasting	12	P3	Planned
Platform · P4 Whitelabel	—	P3	Planned
Platform · P6 Extensions (SDKs, Sandbox)	—	P3	Planned
Marketplace (Extensions P6)	—	P3	Future

13 Success Metrics

Product Metrics

Metric	Month 3	Month 12
Time to first quote	< 60s	< 30s
Quotes per user per day	10+	50+
% quotes above margin floor	90%	98%
Quote-to-close rate	baseline	+20%
DAU / MAU	40%	60%
Modules per customer	2-3	8+
Loops per customer	1	3-4
Pulse score accuracy (margin prediction)	±15%	±5%

Business Metrics

Metric	Month 6	Month 12
Design partners (paying)	10	20+
MRR	$10K	$83K
ARR	$120K	$1M
ACV	$15K	$50K+
Net revenue retention	—	150%+
Gross margin	80%	85%

14 Technical Risks

Risk	Impact	Mitigation
Pulse accuracy (margin prediction)	High — wrong scores = lost trust	Human-in-the-loop on all quotes in Phase 1. Confidence intervals. Progressive automation as accuracy improves. Realized-margin feedback retrains the scorer.
Data quality from integrations	High — garbage in, garbage out	Validation layer on all imports. Reconciliation checks. Manual override capability. Structured schema at connector layer.
Multi-tenant isolation	Critical — data leak = company-ending	Workspace-scoped queries. Row-level security. Regular penetration testing. SOC 2 audit on schedule.
Scale of address data	Medium — maps can be slow with 10K+ pins	Clustering, viewport-based loading, WebGL rendering for large datasets.
Agent hallucination in proposals	High — fabricated data in customer-facing output	Grounded generation only. All agent outputs cite source data. Validation layer before user-facing output.
Solo founder bus factor	High — if Taylor is unavailable, product stalls	Codebase well-structured. First engineering hire triggered by paying-customer milestones post-raise.
Regulatory complexity (multi-jurisdiction)	Medium — Law 25 (QC), GDPR (EU), SOC 2 (US), HIPAA optional	Compliance module (14) is built as a first-class product surface, not bolted on. Frameworks are configurable per workspace.

// Allometry OS · Master PRD v2.0 · Two-Layer Architecture
April 2026 · Taylor Gendron · taylor@allometry.com · allometry.com

Product Requirements Document

Table of Contents

Foundations

Layer 1 — 16 Action Modules

Layer 2

Ops & Delivery

01 Product Overview

What is Allometry?

Core Primitive — the Address

The Pulse

Two-Layer Architecture

The 4 Loops

The Flywheel

Target User (ICP)

02 Design Principles

03 System Architecture

Tech Stack

Multi-Workspace Isolation

Core Data Model — Address as Root

04 6 Data Primitives

System 1 · Pricing Engine

System 2 · Territory / Geo

System 3 · Intelligence Layer

System 4 · Content / Render

System 5 · Governance Plane

System 6 · Pulse Core Primitive

05 Layer 1 · Loop 1 — Price Financial Risk

Core Features

Agent Behavior

Core Features

Core Features

Core Features

06 Layer 1 · Loop 2 — Deploy Operational Risk

Core Features

Core Features

Core Features

Core Features

07 Layer 1 · Loop 3 — Expand Growth Risk

Core Features

Core Features

Core Features

Core Features

08 Layer 1 · Loop 4 — Govern Market & Governance Risk

Core Features

Core Features

Core Features

Core Features

09 Layer 2 — The Platform Horizontal · Included

10 User Roles & Permissions

11 UX Principles

12 Build Roadmap

Phase 1 — Wedge (Current → Month 3)

Phase 2 — First Loop (Month 3-6)

Phase 3 — Full OS (Month 6-12)

13 Success Metrics

Product Metrics

Business Metrics

14 Technical Risks