Now connects to Google Drive

Ask your Drive.Get the answer.

Connect Google Drive once. Cortex reads every document and answers your questions in plain language - with a link to the exact file and page each answer came from.

Connect Google Drive

See how it works

app.cortex.ai

Google Drive

Annual_Report_2024.pdf

Q3_Results.pdf

Strategy_2025.pdf

HR_Policy_v3.pdf

Synced 4 min ago

Ask your documents

What was our Q3 revenue and how does it compare to last year?

Based on Q3_Results.pdf, Q3 revenue reached $4.2M - up 23% YoY from $3.4M in Q3 2023, driven by enterprise contract growth in APAC.

Q3_Results.pdf · p.12

Annual_Report_2024.pdf · p.4

Ask a question about your documents…

Architecture

From your question to a cited answer

The actual pipeline behind every answer: retrieval, re-ranking, a grounding check, then a response cited to the source.

STEP 01

Sync & Chunk

Files from your connected Google Drive folder (and any direct uploads) are parsed and split into overlapping chunks.

document_chunks

STEP 02

Matryoshka Embeddings

Each chunk is embedded with gemini-embedding-001 and truncated to 768 dimensions, then stored in Supabase pgvector.

768-dim

STEP 03

Hybrid Retrieval

pgvector cosine similarity and Postgres full-text search run in parallel, fused with Reciprocal Rank Fusion.

RRF · k=60

STEP 04

AI Re-ranking

The top 10 candidate chunks are scored for relevance and cut down to the 3 that actually answer the question.

gemini-3.1-flash-lite

STEP 05

Agentic Decision

The agent checks whether the retrieved context is sufficient, or reaches out to live web search when it isn't.

tool-calling · Tavily

STEP 06

Streamed Response

The final answer streams back token-by-token over Server-Sent Events, cited to the exact source chunk.

SSE · gemini-3.1-flash-lite

Why it's different

Answers you can actually trust

Every answer is grounded in your own files, checked for relevance before it's written, and cited so you can verify it in one click.

Retrieval

Finds what you mean, not just what you type

Cortex matches on meaning and on exact wording at the same time, then merges both - so it surfaces the right passage whether or not you used the words in the document.

vector cosine+BM25 full-textRRF · k=60

Privacy

Your files stay yours

Read-only access to the folder you choose. Everything is scoped to your workspace with row-level isolation, so documents never cross between accounts.

Re-ranking

Reads the whole doc, quotes the part that matters

Candidate passages are scored for relevance and narrowed to the few that actually answer your question before anything is written.

Streaming

Answers stream as they're written

Q3 revenue reached $4.2M - up 23% YoY from enterprise contract growth in APAC.

Grounding

Tells you when your files don't have the answer

If the documents don't cover your question, Cortex says so instead of guessing - and can reach out to live web search when you want it to.

Cited, always

Every claim traces back to the exact file and page it came from - click a citation to open the source.

Q3_Results.pdf · p.12Annual_Report_2024.pdf · p.4

Your Drive already has the answers. Start asking.

Connect Google Drive and ask your first question in under a minute. Cortex handles the rest.

Connect Google Drive

Cortex

Now connects to Google Drive

Ask your Drive.Get the answer.

Connect Google Drive once. Cortex reads every document and answers your questions in plain language - with a link to the exact file and page each answer came from.

Connect Google Drive

See how it works

app.cortex.ai

Google Drive

Annual_Report_2024.pdf

Q3_Results.pdf

Strategy_2025.pdf

HR_Policy_v3.pdf

Synced 4 min ago

Ask your documents

What was our Q3 revenue and how does it compare to last year?

Based on Q3_Results.pdf, Q3 revenue reached $4.2M - up 23% YoY from $3.4M in Q3 2023, driven by enterprise contract growth in APAC.

Q3_Results.pdf · p.12

Annual_Report_2024.pdf · p.4

Ask a question about your documents…

Architecture

From your question to a cited answer

The actual pipeline behind every answer: retrieval, re-ranking, a grounding check, then a response cited to the source.

STEP 01

Sync & Chunk

Files from your connected Google Drive folder (and any direct uploads) are parsed and split into overlapping chunks.

document_chunks

STEP 02

Matryoshka Embeddings

Each chunk is embedded with gemini-embedding-001 and truncated to 768 dimensions, then stored in Supabase pgvector.

768-dim

STEP 03

Hybrid Retrieval

pgvector cosine similarity and Postgres full-text search run in parallel, fused with Reciprocal Rank Fusion.

RRF · k=60

STEP 04

AI Re-ranking

The top 10 candidate chunks are scored for relevance and cut down to the 3 that actually answer the question.

gemini-3.1-flash-lite

STEP 05

Agentic Decision

The agent checks whether the retrieved context is sufficient, or reaches out to live web search when it isn't.

tool-calling · Tavily

STEP 06

Streamed Response

The final answer streams back token-by-token over Server-Sent Events, cited to the exact source chunk.

SSE · gemini-3.1-flash-lite

Why it's different

Answers you can actually trust

Every answer is grounded in your own files, checked for relevance before it's written, and cited so you can verify it in one click.

Retrieval

Finds what you mean, not just what you type

Cortex matches on meaning and on exact wording at the same time, then merges both - so it surfaces the right passage whether or not you used the words in the document.

vector cosine+BM25 full-textRRF · k=60

Privacy

Your files stay yours

Read-only access to the folder you choose. Everything is scoped to your workspace with row-level isolation, so documents never cross between accounts.

Re-ranking

Reads the whole doc, quotes the part that matters

Candidate passages are scored for relevance and narrowed to the few that actually answer your question before anything is written.

Streaming

Answers stream as they're written

Q3 revenue reached $4.2M - up 23% YoY from enterprise contract growth in APAC.

Grounding

Tells you when your files don't have the answer

If the documents don't cover your question, Cortex says so instead of guessing - and can reach out to live web search when you want it to.

Cited, always

Every claim traces back to the exact file and page it came from - click a citation to open the source.

Q3_Results.pdf · p.12Annual_Report_2024.pdf · p.4

Your Drive already has the answers. Start asking.

Connect Google Drive and ask your first question in under a minute. Cortex handles the rest.

Connect Google Drive