RAG Development Project – Request for Indicative Quotation
Budget / Salary£5,000–10,000
TypeFreelance project
LocationRemote
Posted55 minutes ago
We're looking for an experienced AI/RAG developer or white-label development agency to help deliver a knowledge retrieval and AI-assisted response generation solution for a UK B2B organisation.
The system will support an internal technical team by retrieving relevant information from historical customer support records and technical documentation, then generating evidence-based draft responses for human review.
We will manage the client relationship and Salesforce integration. We're primarily looking for a specialist to design, develop and validate the **RAG and AI components**.
### Expected scope
- Ingest and index historical Salesforce Cases and email correspondence, PDFs, Word documents, technical diagrams, images and potentially videos.
- Develop automated ingestion and synchronisation processes for new, updated and deleted content.
- Configure semantic/keyword retrieval, ranking and source referencing.
- Integrate a commercially available LLM to generate technically accurate draft responses with citations and appropriate handling of uncertainty.
- Provide APIs for integration with Salesforce and support ongoing maintenance.
### Proposed milestones and deliverables
1. **Architecture and evaluation framework:** Agree technology, representative test data, benchmark enquiries and measurable acceptance criteria.
2. **Ingestion proof of concept:** Demonstrate accurate extraction, chunking, metadata and indexing across representative document types.
3. **Retrieval validation:** Demonstrate that relevant, authoritative evidence is consistently retrieved and selected.
4. **Scaled testing:** Validate retrieval performance against a substantially larger knowledge base.
5. **Response generation:** Deliver accurate, evidence-based draft responses with traceable citations.
6. **Production readiness:** Deliver documented APIs, automated synchronisation, monitoring, deployment and technical handover.
These milestones are suggestions, not fixed requirements. **We welcome challenges to the proposed approach and recommendations for a more effective or cost-efficient delivery model.**
### Information requested
At this stage, we're looking for:
- An indicative cost range and estimated development effort.
- An outline of your recommended architecture and technologies.
- Relevant experience delivering production RAG systems, particularly with technical documentation.
- Whether you would use managed services or reusable components to reduce development time.
- An estimated delivery timeframe and earliest availability.
- Expected ongoing infrastructure, AI usage and maintenance costs.
- **What additional information you would need from us to provide a comprehensive technical proposal and quotation.**
An initial working MVP is ideally required by **mid-January 2027**, although we recognise that scope may need to be prioritised to achieve this.
Please identify any major technical risks or assumptions in your initial response.
The system will support an internal technical team by retrieving relevant information from historical customer support records and technical documentation, then generating evidence-based draft responses for human review.
We will manage the client relationship and Salesforce integration. We're primarily looking for a specialist to design, develop and validate the **RAG and AI components**.
### Expected scope
- Ingest and index historical Salesforce Cases and email correspondence, PDFs, Word documents, technical diagrams, images and potentially videos.
- Develop automated ingestion and synchronisation processes for new, updated and deleted content.
- Configure semantic/keyword retrieval, ranking and source referencing.
- Integrate a commercially available LLM to generate technically accurate draft responses with citations and appropriate handling of uncertainty.
- Provide APIs for integration with Salesforce and support ongoing maintenance.
### Proposed milestones and deliverables
1. **Architecture and evaluation framework:** Agree technology, representative test data, benchmark enquiries and measurable acceptance criteria.
2. **Ingestion proof of concept:** Demonstrate accurate extraction, chunking, metadata and indexing across representative document types.
3. **Retrieval validation:** Demonstrate that relevant, authoritative evidence is consistently retrieved and selected.
4. **Scaled testing:** Validate retrieval performance against a substantially larger knowledge base.
5. **Response generation:** Deliver accurate, evidence-based draft responses with traceable citations.
6. **Production readiness:** Deliver documented APIs, automated synchronisation, monitoring, deployment and technical handover.
These milestones are suggestions, not fixed requirements. **We welcome challenges to the proposed approach and recommendations for a more effective or cost-efficient delivery model.**
### Information requested
At this stage, we're looking for:
- An indicative cost range and estimated development effort.
- An outline of your recommended architecture and technologies.
- Relevant experience delivering production RAG systems, particularly with technical documentation.
- Whether you would use managed services or reusable components to reduce development time.
- An estimated delivery timeframe and earliest availability.
- Expected ongoing infrastructure, AI usage and maintenance costs.
- **What additional information you would need from us to provide a comprehensive technical proposal and quotation.**
An initial working MVP is ideally required by **mid-January 2027**, although we recognise that scope may need to be prioritised to achieve this.
Please identify any major technical risks or assumptions in your initial response.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.