Skip to main content
Coldread

Contact Intelligence

Contact Intelligence lets you ask natural language questions about any contact and get AI-powered answers based on their full conversation history.

What does it do?

Contact Intelligence lets you query a contact's full conversation history using natural language. Instead of manually reading through transcripts, ask questions and get instant, cited answers.

See it in action

Watch how Contact Intelligence queries work across your contact database.

How it works

1

Semantic search

Your query is embedded and compared against all conversation transcripts with that contact using pgvector.

2

Context retrieval

The most relevant conversation excerpts are retrieved and ranked by similarity.

3

AI synthesis

Gemini 2.5 Flash reads the retrieved excerpts and generates an answer with citations (call ID + timestamp).

Example queries

"What are their main pain points?"

Returns: Key frustrations mentioned across all calls

"When did they first mention pricing?"

Returns: Timestamp and context from first pricing discussion

"What objections have they raised?"

Returns: All objections with citations to specific calls

"Do they prefer email or phone follow-ups?"

Returns: Communication preferences mentioned in conversations

Usage limits

Contact Intelligence queries are usage-based and count against your plan limits:

  • Solo plan: 25 queries/month
  • Team plan: 100 queries/month
  • Business plan: 400 queries/month

Query limits vary by plan. Upgrade anytime if you need more.

Best practices

  • Be specific: "What pricing did they mention?" beats "pricing"
  • Ask one question at a time for clearer answers
  • Check citations to verify AI answers against source transcripts
  • Use for discovery, not precise facts (e.g., exact dates)