Vector File Database
Upload your documents once. Query them by meaning, not keywords. When you ask a question, the AI finds and references the exact sections that matter – even in 100-page documents.
This is semantic search: the system understands what you’re asking, not just the words you use. Ask about “early termination” and it finds the “cancellation provisions” clause. Ask about “market growth” and it locates the projections, wherever they’re buried.
AI without your documents is half-informed AI
You have contracts, research reports, technical specs, competitive analyses. The AI has never seen them. So every question requires you to paste in “relevant context” – and hope you guessed which context was relevant.
Worse: long documents don’t fit in the paste window. You’re summarizing 100-page reports into 2-page excerpts, losing detail and hoping you kept the right parts.
Vector File Database changes this. Upload your documents to a project. The AI can now search and reference any section, any time, without you manually extracting context.
Automatic indexing for intelligent retrieval
Upload once. The system handles everything else.
1. Chunking
Intelligent splitting
Your document is split into meaningful sections – paragraphs, chapters, logical units – preserving context within each chunk.
2. Embedding
Meaning capture
Each section is converted to a vector representation that captures its semantic meaning, not just keywords.
3. Indexing
Fast lookup
Vectors are stored in a database optimized for similarity search. Finding related content is nearly instant.
4. Retrieval
On-demand context
When you ask a question, the system finds the most relevant sections and includes them in the AI’s context window.
Search by meaning. Not by keyword.
Traditional search finds documents containing your exact words. Semantic search finds documents about what you mean.
Keyword Search
You search “termination clause” → Finds documents with exactly “termination clause” → Misses documents saying “cancellation provisions,” “ending the agreement,” or “contract expiry.”
Semantic Search
You search “termination clause” → Finds sections about ending contracts → Includes “cancellation provisions,” “early exit terms,” “contract termination” – all semantically related content.
Questions that work with uploaded files
Specific Fact Retrieval
“What was the revenue figure in the Q3 report?”
“Who is listed as the primary contact in the partnership agreement?”
“What’s the deadline mentioned in the SOW?”
Document-Based Analysis
“Based on the uploaded spec, what are the biggest technical risks?”
“Does our contract allow us to sublicense the software?”
“What assumptions is this financial model making?”
Cross-Document Questions
“How does the pricing in our proposal compare to the competitor analysis?”
“Are there any conflicts between the tech spec and the requirements doc?”
Works when both documents are in the same project.
Summarization
“Summarize the key findings from the research PDF.”
“What are the main recommendations in the consultant’s report?”
“Give me the executive summary of this 80-page document.”
Upload what you have
Reports, contracts, research papers
Word
.docx documents, proposals, specs
Text
.txt, .md, plain text files
Code
Source files for technical analysis
Best results: PDFs with actual text (not scanned images). Well-structured documents with headings. Remove cover pages and appendices that aren’t relevant.
When file context matters
Contract Analysis
Upload the contract. Ask “What are our obligations if we miss the deadline?” or “Can we terminate early?” The AI finds and interprets the relevant clauses without you hunting through pages.
Research Synthesis
Upload multiple research reports. Ask “What do these sources say about market growth in Asia?” The AI searches across all documents and synthesizes findings.
Technical Documentation
Upload specs, architecture docs, API references. Ask “How does the authentication system work?” or “What are the rate limits?” The AI becomes an expert on your technical stack.
Competitive Intelligence
Upload competitor materials, analyst reports, market research. Build a project-level intelligence base that all five AIs can reference when analyzing your market position.
Two systems, complementary intelligence
Vector File Database handles your uploaded documents – contracts, reports, specs. Semantic search finds relevant sections when you ask questions.
Knowledge Graph handles conversation-derived intelligence – entities, decisions, relationships extracted from your chats.
They work together. When you discuss a document in conversation, Knowledge Graph captures the key entities and decisions. The original document remains searchable in Vector File Database. Cross-reference both when you need the full picture.
Frequently Asked
How big can my files be?
Up to 50MB per file. Very large files (hundreds of pages) work fine – the chunking system handles them. For massive documents, you may get better results with focused questions about specific sections.
Do I need to tell the AI which file to look at?
Not usually. The system searches all files in your project. But you can be explicit (“According to the Q3 report…”) if you want to anchor to a specific document.
What if the AI doesn’t find what I’m looking for?
Try being more specific, or use terms from the document itself. “Check the section about liability” might work better than a general question. You can also ask follow-up: “Is there anything else in the document about this?”
Are my files private?
Files are project-scoped and user-isolated. They’re encrypted at rest and in transit. Your files are not used to train models. Enterprise plans add additional controls.
Can I search across multiple projects?
Files are project-scoped by default. Master Projects can access files across connected projects when you need cross-project intelligence.
Your documents. Your AI’s context.
Stop pasting excerpts and hoping you got the right parts. Upload once, query forever.