Implementation Plan

Roadmap

26-week phased plan. Foundation first, then products in order of PMF urgency. Each phase has a clear milestone that validates before moving on.

26-Week Overview

Foundation
Weeks 1–3
Core Project Loop
Weeks 4–10
Client & Sales Layer
Weeks 11–18
Strategy Layer
Weeks 19–26
0

Foundation

Phase 0
Weeks 1–3
○Monorepo setup (Turborepo + pnpm)
○Shared Postgres schema + Prisma
○Auth layer (Clerk or NextAuth)
○Claude API wrapper with function calling
○Conversational intake engine v1
○Basic UI component library
Milestone

Shared infra deployed. Can create orgs, users, and run AI conversations.

1

Core Project Loop

Phase 1
Weeks 4–10
▲ Blueprint● Bench◎ Retro
○Blueprint: Conversational project scoping
○Blueprint: SOW + proposal generation
○Bench: Talent CRM with skills/rates
○Bench: Match people to Blueprint roles
○Retro: Async retrospective flow
○Retro: Estimation model v1
Milestone

Closed feedback loop: scope → staff → deliver → learn → improve estimates. Learning loop starts turning.

2

Client & Sales Layer

Phase 2
Weeks 11–18
◈ Relay◻ Proof
○Relay: Client communication drafting
○Relay: Scope drift detection vs. SOW
○Relay: Change order generation
○Proof: Dynamic case study generator
○Proof: Prospect-tuned briefs
○Cross-product data flows
Milestone

Full client lifecycle managed. Communication and sales collateral auto-generated from project data.

3

Strategy Layer

Phase 3
Weeks 19–26
◉ Scout◆ Compass
○Scout: Market monitoring engine
○Scout: Competitive intelligence briefs
○Compass: Decision framework with scoring
○Compass: Scout → Decision → Project pipeline
○Cross-product workspace view
○Platform-level analytics dashboard
Milestone

Full pipeline: detect opportunity → decide → scope → staff → manage → learn → sell.

Validation Gates

Gate 1: Core Loop PMF

Week 10

Pass: 3+ firms (50-100 employees) use Blueprint + Bench + Retro on real projects. At least 2 report that their next estimate was more accurate because of Retro data.

If not: The feedback loop isn't clicking. Interview users to find where the chain breaks. May need to narrow to a single vertical (e.g., web dev agencies only).

Gate 2: Client Layer Value

Week 18

Pass: Relay's drift detection catches real scope creep. Proof generates a case study that gets used in a real sales conversation.

If not: These may be 'nice to have' not 'need to have'. Double down on the core loop and revisit client/sales features after more core loop usage data.

Gate 3: Full Pipeline

Week 26

Pass: The full Scout → Proof pipeline has been used end-to-end by at least one power user. Platform narrative resonates in sales conversations.

If not: Full pipeline may be aspirational. Focus on the 2-3 product combinations users naturally adopt together. Let the platform emerge from usage patterns.

Build Priority Logic

Blueprint + Bench + Retro first because together they close the feedback loop. This is the core value prop: scope → staff → deliver → learn → improve. The learning loop starts turning immediately. You can't sell “projects that get smarter” without the learning loop.

Relay + Proof second because they extend the value of existing data. Relay needs a SOW to detect drift against (from Blueprint). Proof needs project history to generate case studies (from Retro). They're high-value but dependent on upstream data.

Scout + Compass last because they're the highest-ambiguity products. They have the broadest use cases and the hardest PMF to prove. By the time you build them, you'll have months of user data telling you exactly what “market intelligence” and “strategic decisions” actually mean to your users.