Introducing Chat: Search and Interact With Internal Data in Real Time

New Features: Sources and Citations

The biggest upgrade in Chat 2.0 is citations and source traceability. Every answer Chat gives is now backed by references to the original files it pulled from; think call transcripts, risk memos, or claims PDFs.

Hover over a highlighted section in Chat’s response, and you’ll see the exact document and location it came from.

Such traceability doesn’t just make people trust it more. It also helps stop mistakes and shows auditors and regulators exactly where the information came from. That way, it’s easy to check if the AI’s answers are correct, all in one place.

Business impact:

• Speeds up due diligence and approvals
• Builds trust with audit teams and regulators
• Reduces internal back-and-forth over data quality
  

Using Chat in Action

Chat is built to serve multiple roles across finance and insurance. Here are a few examples:

• Finance: Search internal knowledge bases such as financial accounting data or compliance and regulatory knowledge, search external knowledge bases for data from sources such as Crunchbase, PitchBook, and others, extract information for onboarding.
• Insurance: Search internal knoweldge bases such as policy and product information, claims data and processing, risk assessments, locate documentation linked to regulatory guidelines and verify compliance, generate quick summaries for support tickets.

Let’s walk through how to use Chat to its full potential.

1. Open Chat

Navigate to the left-hand panel in AgentFlow and select Chat.

From this view, you can select between two modes:

• Chat: Optimized for searching across large datasets, Spaces, and high document volumes
• Instant Chat: Ideal for ad-hoc chats across 3–5 documents

2. Create or Select a Space

Spaces are how AgentFlow organizes your internal data. Each Space contains related documents, think of them like AI-ready data rooms.

You can create a new Space by uploading relevant documents, or use an existing one. For this walkthrough, let’s use an existing Space: td-stt-tts.

To create a new space, hit the “Create new space” button and name your Space where your data will be stored. After creation, you will see it in the list, available for selection any time you use Chat.

Business value of Spaces:

• Keep data structured by use case, client, or knowledge sources
• Organize internal knowledge bases such as contracts, historical legal records, and research documents
• Enable secure, permission-based access while maintaining clear separation of data

3. Add Your Prompt

Now, type your question. For example:

"What’s the TFSA limit this year, and how do I safely check my personal room?"

Or use voice-to-text by clicking the microphone icon.

Chat returns a response, with citations embedded and clickable.

4. Enable Web Search

Toggle Web Search to expand Chat’s knowledge to real-time internet data. This is especially helpful for:

• Pulling in market updates
• Finding regulatory changes
• Supplementing internal documents with public info

Chat clearly separates web results from internal sources, so there’s no confusion about where each piece of data comes from.

Faster Search for Faster Decisions

Chat 2.0 is built for speed and certainty:

• Faster search means your team can answer underwriting questions in seconds, not hours
• Fewer mistakes means your compliance and audit teams spend less time reviewing errors
• Clear traceability means everyone from analysts to execs can trust what they’re seeing

This isn’t just about productivity. It’s about improving employee experience and unlocking real-time decision-making without cutting corners on risk.

‍

Book a demo to see Chat live and learn how it can help you implement AI across your workflows.