Getting Started — Full Workflow
New to RFPilot? Follow these steps in order to go from zero to a fully generated RFP response.
The Page Guide panel in the top navigation bar shows real-time status and “What to do” instructions for every page you visit — use it at any point if you feel stuck.
Organization Dashboard
Your home screen — all organizations and projects at a glance.
Lists all organizations your account belongs to. Each card shows the organization name, your role (Owner / Admin / Member), and the LlamaCloud connection status.
Click an organization card to enter its dashboard where you can manage projects, team members, settings, and the Knowledge Base.
Use the New Project button to start a fresh RFP workspace inside the selected organization.
No organizations shown: Your account has no org membership. Use the “Create Organization” button or ask an Admin to invite you via the Team page.
⚠ Disconnected badge on org card: Go to Org Settings → Knowledge Hub Integration → connect to LlamaCloud. Without this, AI responses will be generic with no document citations.
No projects listed: Click New Project to create the first one. Projects are the main workspace for each RFP.
Only Owners and Admins can modify org settings. If you need access, contact your organization Owner.
Knowledge Base & Document Hub
Two complementary systems power the AI's knowledge. The Knowledge Base holds curated Q&A pairs. The Document Hub (LlamaCloud) holds uploaded source documents.
A Knowledge Base is a curated library of question-and-answer pairs that the AI can draw from when generating workflow responses.
Use it for pre-approved, company-standard answers — e.g. “Describe your cloud security posture” → your standard security narrative.
How to create a Knowledge Base:
The Document Hub connects RFPilot to your LlamaCloud account, enabling document-level semantic search during AI execution.
Supported file types:
Maximum file size: 50 MB per file.
Processing stages after upload:
Uploading → Parsing → Indexing → Ready
Large files may take 3–5 minutes. If stuck >10 min, delete and re-upload.
Connecting LlamaCloud (Admin only)
Document Hub must be connected before document-sourced AI responses are enabled. Until connected, the AI will generate answers from general knowledge only — no citations.
Organization Settings
Manage your org's name, integrations, workflow templates, and danger zone. Only Owners and Admins can access this page.
Update your organization's display name. This name appears across all projects and invitations.
Click Save Changes after editing the name. Changes are instant for all members.
Admins can upload JSON workflow template files here. These become available to all projects in the org via the Sync Templates button in the Workflows page.
Templates standardize the structure of AI-generated RFP responses across all projects.
Upload a template once, then every project team can sync it — no need to rebuild the workflow structure from scratch.
This page is only accessible to users with Owner or Admin role. Members see a locked view.
To change a member's role, go to Team → find the member → change their role in the dropdown.
The Danger Zone (delete organization) is irreversible. All projects, documents, and team members will be permanently removed.
Team Management
Invite colleagues, manage roles, and control who can do what within your organization.
If the email was not received, remove and re-invite. Check spam folders.
Find the member in the table → click the role badge → select a new role from the dropdown.
Role changes take effect immediately. The member will see different permissions on their next page load.
Owner role cannot be changed or transferred via the UI — contact system support for ownership transfers.
Always have at least one Admin beyond the Owner — so settings can be managed if the Owner is unavailable.
Give Member role to most users — they can fully use RFP workflows without access to sensitive settings.
Removing a member immediately revokes their access. Ensure their work is saved before removing.
Creating a New Project
A project is an isolated workspace for one RFP response — it holds the document, extracted questions, workflows, and generated answers.
Project names within the same organization must be unique. Plan a naming convention like [Client] — [Period] RFP to stay organized.
Once a project is created, it cannot be renamed via the UI. Plan your name carefully, or delete and recreate if needed.
Project Overview
A real-time summary of your project's progress — questions extracted, answers generated, and export readiness.
Questions answered / total: Tracks how many RFP requirements have AI-generated responses.
Completion % progress bar: Visual indicator of overall RFP response progress.
Sections overview: Breakdown of which document sections have been fully answered.
Project Timeline: Auto-generated milestones based on document upload, extraction, and execution dates.
If 0 questions: Go to Questions → upload RFP document → extract.
If 0 answers: Go to Workflows → open workflow → Execute All.
If 100% complete: Go to any workflow execution → click Export.
Resources (Document Indexes)
Select which document collections (indexes) from LlamaCloud this project can access during AI execution.
Each project can be restricted to specific document indexes from your organization's LlamaCloud project. This prevents irrelevant documents from influencing AI responses.
No indexes selected
All indexes from the org's connected LlamaCloud project are available — no restriction.
Specific indexes selected
AI only searches those specific document collections. Useful to scope answers to relevant domain documents.
If the org is not connected to LlamaCloud, the Resource selector is unavailable. Connect in Org Settings → Knowledge Hub Integration first.
Question Extraction
Upload your RFP document and let the AI identify all obligations, requirements, and deliverables automatically.
What gets extracted
- Direct questions (“Vendor must describe…”)
- Mandatory requirements (“Offeror shall…”)
- Deliverables and compliance items
- Technical specifications and standards
What is skipped
- Boilerplate / legal notice text
- Table of contents and formatting
- Evaluation scoring criteria
- Cover page / header information
If 0 questions are extracted, the document may be image-only (scanned PDF). Use a text-based PDF for best extraction quality. You can also add questions manually.
If the org is not connected to LlamaCloud, the index selector is grayed out and AI answers will have no document citations.
Workflows
A workflow is the blueprint that determines how the AI structures and generates the final RFP response document.
Templates are predefined workflow structures uploaded by your Admin. Click Sync Templates to pull them into your project. They ensure consistency across all RFP responses.
Custom Workflows are built from scratch. Use them for highly unique RFP formats that don't match existing templates.
Always try Sync Templates first — your Admin may have already prepared the perfect structure for your RFP type.
Each workflow contains sections, each with a blueprint (title, instructions, and format guidance for the AI).
During execution, each section maps to one or more extracted RFP requirements. The AI uses the blueprint to structure its response.
You can edit section blueprints directly in the execution interface before or after running the AI.
Running a Workflow
The AI processes each requirement using your Knowledge Base and documents, then drafts a response for each one.
Prerequisites before “Execute All”
At least one requirement added to the workflow (left panel)
At least one Knowledge Base index selected (recommended — AI will still run without it, but responses won't reference your documents)
If a row shows empty content after execution, retry that requirement. Empty AI responses are always flagged and never silently saved as complete.
Single AI call per requirement. Fast and cost-efficient. Best for standard, straightforward requirements.
Multiple AI calls with intermediate reasoning. Slower but more thorough. Use for complex technical requirements that need detailed answers.
Once all requirements have AI-generated responses, click Export in the top-right header.
The export downloads a formatted document containing all requirements and their AI responses, ready for review and submission.
Tip: Review each AI response inline before exporting — you can edit any response directly in the Live Output Document area.
Use the History button to view past exported versions of this workflow.
Requirements are the individual items the AI responds to. They live in the Design & Mapping panel on the left.
Actions available per requirement:
- ▶ Run — execute AI for this single requirement
- Edit the requirement text inline
- Delete — remove a requirement from the workflow
- Reorder — drag requirements up/down
Requirements are auto-imported from extracted RFP questions, or you can add them manually for custom sections.
Core AI Features Overview
How the automated pipeline works end-to-end.
Automated Document Parsing
Upload any RFP in PDF or DOCX format. The parsing engine preserves tables, headings, and document structure, feeding clean text to the AI for accurate question extraction.
Traceability Matrix Extraction
The AI scans parsed documents for obligations, deliverables, and interrogatives. These form an interactive compliance matrix with source-text snippets for auditing and compliance review.
Intelligent AI Drafting
The AI queries your Knowledge Base and LlamaCloud document indexes for relevant facts, then generates proposal answers automatically. Choose between Quick Mode and Multi-Step Reasoning per requirement.
Knowledge Base Q&A
Pre-build a library of standard answer pairs for common RFP topics. The AI will cite and incorporate these during execution, ensuring consistency with your company's approved messaging.
Role-Based Access Control
Owner, Admin, and Member roles ensure the right people can access the right features. Only Admins can modify org settings and upload workflow templates.
FAQ & Troubleshooting
Answers to the most common questions and issues.
Why is the AI response empty after execution?
The Document Hub may be disconnected, or no indexed documents matched the requirement. Check Org Settings → Document Hub connection, ensure indexes are selected in the workflow, then retry the requirement.
Why can't I click 'Execute All'?
You need at least one requirement added to the workflow. Add requirements from the Design & Mapping panel on the left, or import from extracted RFP questions.
Some rows show 'error' after execution — why?
The AI returned empty content for those requirements. Click the ▶ retry button on each failed row. If it keeps failing, verify your Knowledge Base has relevant documents and that indexes are selected.
The response looks generic and has no citations.
No Knowledge Base indexes are selected for this workflow. Open the index selector (Knowledge Base button in the header) and choose the relevant document collections.
Why is the Page Guide showing warnings?
The Page Guide validates your setup for the current page and flags missing prerequisites. Each warning includes a 'What to do' instruction. Follow the steps and the warning will resolve automatically.
Why is document processing stuck?
Large files (>20 MB) can take 3–5 minutes to process. If stuck for more than 10 minutes, delete the document and re-upload it. Contact admin if the issue persists.
Why can't I see any Knowledge Bases?
Knowledge Bases belong to organizations. Ensure you are viewing the correct organization. If none exist, an Admin needs to create one via the Knowledge Base tab.
The Document Hub upload button is disabled.
Your organization is not connected to LlamaCloud. Go to Org Settings → Knowledge Hub Integration and connect your LlamaCloud credentials.
Why are 0 questions extracted from my RFP?
The document may be a scanned (image-only) PDF with no text layer. Switch to a text-based PDF for best results. You can also add requirements manually in the workflow execution panel.
I invited a team member but they can't log in.
They need to accept the invitation email first. If the email was not received, remove the invitation from the Team page and re-invite. Ask them to check their spam folder.
Can I edit AI-generated responses?
Yes. All responses are editable inline in the Live Output Document on the right side of the Workflow Execution page. Changes are auto-saved every 30 seconds.
How do I export the final RFP response?
Once all requirements have responses, click the 'Export' button in the top-right of the Workflow Execution page. A formatted document will be downloaded to your device.
What AI models does RFPilot use?
RFPilot supports multiple LLM providers configured by your Administrator (e.g., Gemini, GPT-4, Claude). The specific model used depends on your organization's settings. Contact your Admin for details.
Can I use RFPilot without a LlamaCloud connection?
Yes. The AI will still generate responses using its general knowledge and your Knowledge Base Q&A pairs. However, responses will not cite or reference uploaded documents, which may reduce quality for technical RFPs.
How do workflow templates work?
Admins upload JSON workflow template files in Org Settings → Workflow Templates. Project teams then click 'Sync Templates' in the Workflows page to import them. Templates standardize the structure of AI responses across all projects.
Support & Contact Admin
What to do when you encounter a problem or need administrative action.
Reporting a Bug
If you encounter a platform error, unexpected behavior, AI generating harmful content, or data loss — report it with:
- Steps to reproduce the issue
- The page/feature where it occurred
- Screenshot if possible
- Your browser and OS version
When to Contact Admin
Reach out to your organization Admin when you need:
- Access to restricted Projects or KBs
- Role upgrade (Member → Admin)
- LlamaCloud connection setup
- Workflow template uploads
- Org-wide AI model configuration
- Billing or quota increases
- Removing or transferring ownership
Quick links from the Page Guide
The validation panel in the header links directly to these sections based on what needs attention on your current page. Every “Learn more” link lands on the relevant anchor above.