Arhitecture · Proposal Modeling Engine
The Closmore "Win-Pattern" Engine:
Architecting Proposal Generation with AI RAG
Most sales teams treat a "winning proposal" as a static document to be copied. That is a mistake. A winning deal is a sequence of validated decisions.
Closmore treats a "winning proposal" as a Structured Data Schema. We use a Retrieval-Augmented Generation (RAG) pipeline to map historical "Win-Patterns" onto new account context.
Step 1: The "Win-Pattern" Extraction (The Schema)
We don't store "proposals." We store Winning Logic.
Sales Logic: Closmore parses your winning proposals to extract core logic, ignoring the "fluff." It identifies the Constraint Set (Must-Haves), Risk Mitigation (Safety Labels), and the Milestone Sequence (the exact order of "easy tasks" that led to the signature).
Technical Logic (util 3 / util 2): When a deal is marked Closed-Won, the Chucking Function triggers. It breaks the proposal into semantic chunks and tags them with metadata (JSON schema):
milestone_sequence— Array of objects{step_id, effort_score, client_action_required}risk_mitigation_labels— Array of strings (e.g.,"Security-Compliance-Assurance")stakeholder_alignment_path— Array of roles (e.g.,"CTO","Database Admin")value_alignment_logic— Vector embedding of the "Why" (the core business outcome)
Step 2: The "Target Account" Context (The Input)
Before the AI drafts a single word, Closmore pulls the live "Account Background" from your Vector Database to ensure the proposal is hyper-relevant.
Contextual Input (fx 2): The Function - Tokenise Input module normalizes raw data from the CRM and transcripts:
- CRM Data:
current_pain_points(extracted via NER — Named Entity Recognition) - Stakeholder Data:
current_stakeholders(mapped by role/influence) - Risk Profile:
current_risk_profile(derived from the "Fear to Start" analysis)
Step 3: The "Contextual Alignment" (The Mapping)
Sales Logic: When a rep starts a new deal, Closmore compares the prospect's "Fear to Start" profile against your "Win-Pattern" database. It looks for Structural Similarity. If the new prospect has a "Security/Compliance" fear, the system retrieves the "Risk Mitigation" schema from the "Role Model" deal.
Technical Logic (fx 6): The Intelligence Consolidate Function performs Schema Alignment:
- Step A (Retrieval): The
RAG Orchestratefunction queries the Vector DB for the "Winning Schema" with the highest Cosine Similarity to the "Target Account Context." - Step B (Injection): The
Intelligence Consolidate Functionperforms a JSON-merge, injectingcurrent_pain_pointsandcurrent_risk_profileinto the schema fields. - Step C (Constraint Enforcement): The prompt passed to the LLM is Schema-Locked. It is instructed to only output in a specific JSON format that maps to your proposal sections.
Step 4: The "Dynamic Roadmap" (The Orchestration)
Sales Logic: Closmore generates a Dynamic Roadmap — not a static Word doc. It pre-fills milestones, safety labels, and stakeholder roles to de-risk the start. This turns the client from a "passive observer" into an active approver.
Technical Logic (fx 7): The Output Function transforms the validated JSON schema into final execution assets.
- Milestone Generation: The system iterates through the
milestone_sequencearray, auto-generating "easy tasks" (e.g., test plan sign-offs) tailored to the current account. - Risk Mitigation: The system pulls specific "Safety Labels" from the schema to address the prospect's unique "Fear to Start."
Step 5: The Systemic Advantage (The Learning Loop)
Sales Logic: Every time a rep uses this "Model" and gets a sign-off, Closmore logs the outcome. The system learns: "When we propose this specific milestone sequence, the 'Fear to Start' stall drops by X%."
Technical Logic (util 4): When the rep logs the outcome in the CRM, the CRM Update Function triggers a re-indexing of the "Winning Schema." If a specific milestone_sequence resulted in a stall, the system lowers its success_weight in the Vector DB. The system self-optimizes its proposal logic based on real-world deal velocity.
Ready to Architect Your Pipeline?
I do not sell "sales tools." I architect GTM systems that turn stalled enterprise deals into predictable revenue.
If you are a Founder or GTM Leader tired of "firefighting" and ready to build a system that actually scales, let's talk.
If you have a high-stakes pipeline with recurring "Silence Traps," or would like audit your existing sales pipeline system with Closmore SalesOS framework. I would like to show you how we can re-engineer your deal flow.