Skip to main content
PRO-02 · Professional Services

Proposal and Pursuit Studio

Outcome​

Compress the pursuit team's response cycle by decomposing an RFP into scored requirements, assembling compliant answers from vetted credentials, case studies and CVs, and flagging every requirement the firm cannot evidence before the bid or no-bid decision is made.

Portfolio attributeValue
Primary personaPursuit team
Primary aggregatePursuitResponse
Cosmos DB containers3
Grounding dataRFPs, credentials, case studies, CVs, brand content
Development estimateUSD 60-220 monthly dev estimate
Runtime optionsPython FastAPI and C# ASP.NET Core

Productivity hypothesis​

These measures reflect the workflow, evidence, decision boundary, and industry risk for this use case.

Impact typeMeasureConservative internal assumptionCustomer measurement
ProductivityPursuit draft cycle time12-20% reduction(baseline median pursuit-draft minutes - assisted median minutes) / baseline median minutes * 100
Quality or riskRework rate5-12% reduction(baseline corrected or repeated pursuit-task rate - assisted rate) / baseline rate * 100

Important: Conservative internal pilot assumptions only; not public benchmarks, commitments, or expected customer results. Validate against a comparable baseline or matched control cohort with mandatory human review.

Rationale: RFPs, credentials, case studies, CVs, and brand content support faster pursuit-draft assembly.

Pilot guardrail: Prohibit unsupported claims and require approval for credentials, pricing, commitments, and proposal language.

Customer validation plan​

  1. Baseline the current workflow with representative users and tasks.
  2. Run the accelerator on the same task mix with human review retained.
  3. Measure pursuit draft cycle time and rework rate with the documented formulas.
  4. Replace internal assumptions with customer-observed results before a business case.

Why Azure Cosmos DB​

This AI solution needs an operational data plane, not only a vector index:

  • Hybrid vector and full-text retrieval combines semantic recall with exact identifiers, codes, names, and domain language.
  • Hierarchical partition keys and metadata predicates keep tenant and domain boundaries in the normal query path.
  • Operational JSON, source chunks, workflow and agent state, feedback, and audit records remain together for low-latency reads and governed writes.
  • Change feed reacts to source and operational updates without polling.
  • Autoscale throughput, configurable consistency, and multi-region distribution support variable demand and enterprise availability.

Domain workflow​

Domain workflow

StageActorInput → outputApproval
1. Solicitation decompositionProposal and pursuit studiosolicitationDocument → RfpRequirementPolicy controlled
2. Credential and personnel matchingProposal and pursuit studioRfpRequirement → credentialMatchSetPolicy controlled
3. Gap identification and bid or no-bid framingPursuit leadGapFinding → bidDecisionHuman required
4. Response section draftingProposal and pursuit studio with writerscredentialMatchSet → ResponseSectionPolicy controlled
5. Compliance review and submissionProposal manager and partnerResponseSection → submittedProposalHuman required

Technical architecture​

Technical architecture

Cosmos DB data model​

ContainerHierarchical partition keyDocumentsSearch paths
pursuitResponses/tenantId, /pursuitIdPursuit, RfpRequirement, ResponseSection, GapFinding/requirementVectorEmbedding, /sectionVectorEmbedding, /requirementVector, /sectionVector
firmCredentialLibrary/tenantId, /credentialIdCredentialAsset/assetVectorEmbedding, /assetVector
personnelMasterProfiles/tenantId, /personnelIdPersonnelProfile/cvVectorEmbedding, /cvVector

Query contract​

QueryContainerScopePurpose
UncoveredMandatoryRequirementspursuitResponsessingle-partitionList mandatory requirements with no drafted, compliant answer as the submission deadline approaches.
ConsentedCredentialRetrievalfirmCredentialLibrarycross-partition-boundedMatch a requirement to case studies and references that are approved, consented and unexpired. Intentional cross-partition fan-out across the firmCredentialLibrary reference container, bounded by the tenant prefix, approval and client consent state and the submission deadline, returning only the top-ranked credential assets.
QualifiedPersonnelForRolepersonnelMasterProfilescross-partition-boundedFind proposal-ready personnel meeting the clearance, certification and availability the solicitation demands. Intentional cross-partition fan-out across the personnelMasterProfiles reference container, bounded by the tenant prefix, the required clearance rank, the required certification and availability, returning a shortlist of candidate profiles.
OpenGapsBySeveritypursuitResponsessingle-partitionSurface unresolved evidence gaps that should shape the bid or no-bid and teaming decision.

Python and C# services route each document type to its declared container and hierarchical partition key. ETag-based writes protect mutable aggregates, while append-only records and bounded cross-partition queries preserve scale.

Azure deployment​

Azure deployment

Clone the source repository, authenticate with Azure, and run:

Set-Location accelerators/professional-services/proposal-and-pursuit-studio
az login
./Deploy.ps1 -Runtime python -Location eastus2

Change python to csharp to deploy the .NET implementation.

Explore the implementation​

Return to all accelerators