Modernization Simulator / Proposal Orchestration

How the Simulator Creates Proposals

Proposal creation is not a single generation prompt. In the Simulator, it is a staged orchestration path: signals and source materials are normalized, Scanner intelligence can deepen the client context, uploaded evidence is checked and synthesized, portfolio and experience data constrain the offer, and the final proposal remains a draft for human validation before it is sent.

3 Commercial outputs: proactive proposal, RFP creation, reactive RFP-based proposal
2 Primary motions: shape demand upstream or respond to explicit demand
<1h Target path from prepared inputs to proposal artifact before human validation
10 Proposal sections generated as structured evidence-backed content

The thesis

The Simulator creates proposals by moving from evidence to commercial argument. The RFP Orchestrator does not merely write polished text; it decides which source path is being used, what customer intelligence can be trusted, which services are actually available, what materials have gaps or contradictions, and where a human must validate the final judgment.

Input Surface
Scanner signal / Simulator archive / raw RFP / material package / additional customer documents
Classify the commercial motion before writing

01. Signal and Demand Motion

Scanner / Lead / Direct RFP

Proposal creation begins by asking what kind of demand exists. A proactive proposal begins before the customer has written an RFP: Scanner data, vulnerability signals, future pressure, narrative-reality gaps, or customer-status analysis can trigger the advisory flow. A reactive proposal begins after the customer has already named the need through a formal RFP, direct request, or material package.

This distinction matters because proactive work shapes the brief, while reactive work responds to a brief. The same Simulator can support both, but the commercial posture is different.

Proactive demand shaping Reactive request response Human validation later
Proactive triggerScanner intelligence and diagnostic engines identify why a customer should engage before a formal request exists.
Reactive triggerA direct RFP, lead, or uploaded material package defines the response boundary.
Normalize every path into typed source data

02. Source Normalization

RFPSourceData / ProposalGenerateRequest

The RFP module uses a bridge type, RFPSourceData, to merge different input paths: standard simulation output, forensic audit output, raw RFP content, and multi-file material packages. It carries organization context, phases, key results, metrics, risks, selected portfolio entries, forensic enrichments, raw RFP classification, material verification, deep analysis, and optional Scanner company data.

The Proposal module uses ProposalGenerateRequest. It can consume a structured RFP document, raw RFP content, audit/source data, supplementary customer materials, metadata, tone, focus areas, and vendor strengths.

standard forensic raw-rfp material-package
Archive bridgeSimulator and RFP HTML archives carry embedded JSON capsules that can be reused as source material.
Raw document pathPDF, HTML, TXT, and archive content can be extracted and treated as proposal source material.
Condense large context into writer-ready intelligence

03. Intelligence Synthesis

Scanner / Materials / Knowledge Bases

Before the final RFP or proposal is written, the Simulator tries to enrich the source data. If a client organization can be matched in the Scanner database, the raw company intelligence is condensed into a concise brief: profile, financial context, pain points, competitive landscape, win themes, and suggested proposal tone.

Additional customer materials are also analyzed. PDFs and HTML documents are extracted, classified, summarized, checked for contradictions and gaps, and synthesized into a material intelligence brief. Portfolio data constrains what services can be offered. The broader knowledge-base layer can add evidence about pricing, capability, legislation, delivery terms, risk, references, capacity, CRM context, and real consultant experience where relevant.

Scanner brief Material brief Knowledge bases Vendor portfolio
Why this layer existsLarge, uneven context is turned into bounded intelligence before long-form document generation begins.
Failure behaviorScanner and experience enrichments are useful but non-fatal; the proposal path can continue if optional enrichment fails.
Create the RFP artifact when the buyer-side brief must be structured

04. RFP Creation

/api/rfp-generate

RFP creation is the buyer-side artifact path. It can restructure raw source material into a professional RFP, or turn Simulator audit and roadmap data into an outcome-oriented RFP. The generator is instructed to transform activity language into outcome language, define measurable acceptance criteria, categorize service requirements, and specify client/vendor/shared risk ownership.

This path produces a structured RFPDocument with sections for executive summary, outcome specifications, service scope, required capabilities, evaluation criteria, timeline constraints, success criteria, risk allocation, compliance requirements, and submission requirements.

Outcome specifications Evaluation criteria Risk allocation Submission requirements
Model providersAnthropic, OpenAI, and Kimi can be used for long structured JSON generation, with provider fallback where available.
Document statusThe generated RFP starts as a draft, carrying sourceData for traceability.
Respond with a vendor-side proposal

05. Proposal Generation

/api/proposal-generate + polling

Proposal generation is the vendor-side response path. The browser submits a generation job and receives a job ID immediately, avoiding proxy timeouts while the long-running generation continues in the background. The client polls status until the proposal is complete.

The final prompt combines the RFP summary, Scanner intelligence, material intelligence, service portfolio, experience database, vendor strengths, tone, and formatting rules. It generates a ProposalDocument with ten sections: executive summary, proposed approach, team composition, deliverables, timeline, pricing, quality assurance, risk management, relevant experience, and terms and conditions.

Async job 10 proposal sections Portfolio constrained Experience enriched
Generation targetWith prepared source materials, the intended commercial path is a ready proposal artifact in <1 hour.
Human-in-the-loopThe human does not draft each section manually; the human validates claims, judgment, tone, and send-readiness.
Review before sending

06. Validation, Refinement, and Export

Review / Refine / Archive

The generated proposal is not treated as automatically final. It enters review as a draft with source provenance: raw materials, databases used, portfolio context, generation provider (Anthropic / OpenAI / Kimi), and timestamp. Users can refine individual sections through targeted feedback instead of regenerating the entire document.

Export keeps the artifact portable. Proposal and RFP documents can be downloaded as JSON or HTML, and the surrounding Simulator archive model keeps source context recoverable for later advisory or commercial workflows.

Draft Section refinement Source provenance HTML / JSON export
Validation pointHuman review moves to the end of the pipeline, where it has maximum context and minimum manual drafting burden.
Quality boundaryThe system can generate and structure; the accountable human decides whether the artifact is commercially and ethically ready.

Three Proposal Outputs

The Simulator's proposal architecture is best understood as three related outputs, not one generic document.

Upstream

Proactive Proposal

Begins with a signal, not a tender. Scanner intelligence and diagnostic engines identify a reason to engage, the Simulator forms the modernization thesis, and the proposal becomes a consultation-ready artifact for shaping the customer's demand before procurement language exists.

Buyer-Side Artifact

RFP Creation

Turns roadmap, audit, raw source, or material-package evidence into a structured RFP. This is useful when the customer or advisor needs to convert strategic intent into a professional procurement document with outcomes, requirements, criteria, and risk ownership.

Vendor-Side Response

Reactive RFP-Based Proposal

Starts from a direct RFP, lead, or source package. The Simulator extracts requirements, enriches context, maps portfolio services, uses experience data, and writes a structured response designed for review and refinement.

Knowledge Bases and Databases Used by Proposal Creation

Proposal orchestration is only useful if the document is grounded in the right operating facts. These data sources act as the proposal intelligence layer: some trigger demand, some constrain the offer, some validate feasibility, and some make the final proposal commercially realistic.

Market and Client Intelligence

Scanner Database

Provides company-level signals, vulnerability, financial context, future pressure, and win themes for proactive proposals and client-specific RFP responses.

Market and Client Intelligence

Passive Intelligence Database

Adds background market signals, observed changes, weak signals, and non-request-based context that can help explain why outreach or proposal timing is credible.

Commercial Memory

Win / Loss Analysis Database

Feeds lessons from previous pursuits into win themes, risk language, positioning, objection handling, and proposal emphasis.

Delivery Feasibility

Capability / Competence Database

Checks whether the proposed work matches actual organizational skills, methods, certifications, technologies, and delivery capabilities.

Delivery Feasibility

Capacity Planning

Helps ensure proposed timelines, staffing, allocation, and delivery promises are realistic rather than merely persuasive.

People and Evidence

Interactive CV / Diary Database

Supplies consultant profiles, relevant experience, delivery history, and success-story evidence for team composition and credibility sections.

Commercial Control

Price Database / Pricing Guardrails

Keeps pricing, cost assumptions, payment schedules, and commercial terms inside approved boundaries and helps avoid unsupported pricing promises.

Commercial Control

Risk Management Plan

Provides risk categories, mitigation patterns, contingency logic, and ownership framing for risk registers and delivery governance.

Proof and Positioning

Reference Cases

Grounds relevant experience, case examples, proof points, and comparable outcomes in reusable reference material instead of invented credibility.

Client Adaptation

Client-specific Customization Rules

Captures client-specific language, governance expectations, tone, exclusions, preferred formats, decision logic, and relationship constraints.

Legal and Terms

Legislation / Terms of Deliveries

Informs compliance requirements, delivery terms, procurement constraints, contract assumptions, and the boundaries of what can be promised.

Relationship System

CRM

Adds account ownership, relationship history, opportunity stage, prior interactions, and follow-up logic so the proposal fits the customer relationship.

Proactive Versus Reactive Is Not a Cosmetic Difference

The same document machinery can support both motions, but the strategic posture changes.

Dimension Proactive Proposal Reactive RFP-Based Proposal
Starting point Scanner trigger, future pressure, customer vulnerability, or diagnostic hypothesis. Formal RFP, direct lead, uploaded source document, or material package.
Commercial posture Shape the customer's thinking before the request is fully named. Respond faster and better inside an already visible competitive process.
Primary question Why should this customer act now, and what modernization move is becoming necessary? How do we meet the stated requirements with the strongest evidence, team, and service fit?
Human role Validate whether the outreach thesis is credible, timely, and relationship-safe. Validate compliance, claims, pricing, fit, tone, and send-readiness.

Quality Gates and Constraints

Proposal quality comes from the constraints around generation. The Simulator makes the Anthropic / OpenAI / Kimi model answer inside a bounded evidence system rather than asking it to improvise a commercial document from memory.

Classification

Outcome or standardized path

Raw RFP materials are classified as outcome-driven or standardized, with extracted requirements, complexity, domain signals, suggested services, and metadata.

Material Readiness

Completeness before deep analysis

Multi-file packages are checked for relevance, detected document type, cross-document themes, completeness, missing areas, and readiness before deeper synthesis.

Grounding

Use real services and experience

Proposal generation must map requirements to the actual service portfolio. It can also use real CVs and success stories where they fit, instead of inventing the entire delivery story.

Formatting

Structured sections, not prose soup

The proposal prompt requires tables for risks, timeline, pricing, team, deliverables, and quality assurance. This makes the output reviewable.

Fallbacks

Provider resilience

Long JSON generation can use Anthropic, OpenAI, or Kimi, with provider fallback where available. Optional enrichment failures do not automatically block the whole path.

Traceability

Sources used are recorded

Proposal documents keep raw materials, external sources, generation provider (Anthropic / OpenAI / Kimi), and generated timestamp so review can ask where the content came from.

Public Interfaces Behind the Narrative

The page is descriptive, but it is grounded in concrete interfaces in the Simulator codebase.

RFP MODULE POST /api/rfp-classify POST /api/rfp-verify-materials POST /api/rfp-deep-analysis POST /api/rfp-generate POST /api/rfp-refine PROPOSAL MODULE POST /api/proposal-generate -> returns { jobId } GET /api/proposal-status/:jobId POST /api/proposal-refine CORE TYPES RFPSourceData RFPDocument ProposalGenerateRequest ProposalDocument ProposalSources

What This Does That a Single Prompt Cannot

Generic Prompt Simulator Proposal Orchestration
Mixes raw RFP, client context, services, and sales language into one instruction. Separates intake, classification, intelligence synthesis, RFP creation, proposal generation, validation, and refinement.
May invent services, team credentials, risk logic, or customer context. Constrains generation with portfolio entries, experience data, Scanner intelligence, uploaded materials, and source provenance.
Produces a document-shaped answer, often with weak review handles. Produces typed RFP and Proposal documents with named sections, source records, export paths, and section-level refinement.
Human effort is spent drafting and repairing. Human effort moves to validation: checking judgment, claims, compliance, fit, and commercial readiness.