Unlike standard chat, which provides direct responses, Agents use advanced reasoning to understand what steps and tools are needed, execute those steps in sequence, and synthesize information from various data sources to answer your question.
How do Agents work?
Agents follow a structured reasoning process:
Breaking Down Your Request – The Agent analyzes your question and determines what steps are needed to answer it completely
Selecting the Right Tools – For each step, the Agent chooses the most appropriate data sources or tools (like your uploaded documents, Financial Modeling Prep API, or FRED economic data)
Learning as They Go – Each step builds on previous steps. For example, an Agent might first look up a company's stock symbol, then use that symbol to retrieve financial data
Retrieving Data – Agents pull information from your knowledge base documents or fetch real-time data from external APIs based on what you select.
Generating Your Answer – After completing all necessary steps, the Agent combines everything it learned to provide a complete, well-sourced response
When should I use Agents?
Use Agents when you need:
Multi-step analysis – Questions that require combining information from multiple sources
Real-time financial data – Current market data, company financials, or economic indicators from FMP or FRED APIs
Complex research tasks – Requests that involve looking up information, then using that information to gather more data
Web research – When you enable web search, Agents can find current information beyond your uploaded documents
A few examples where Agents can be used:
"Compare the revenue growth of Apple and Microsoft over the last 3 years"
"What are the latest economic indicators from FRED for inflation trends?"
"Analyze Sonos financial performance using FMP data"
How do I enable Agents?
Agents are available through the selector under the chat interface. When you enable Agents, you activate a different processing pathway that gives the AI access to advanced reasoning and tool selection capabilities.
Important: Agents work in conjunction with your Source Selector settings. Whatever sources you enable (like News, Web Search, or specific data APIs) become available as tools the Agent can use.
What data sources can Agents access?
Agents can pull information from:
Your Knowledge Base – All documents you've uploaded to ToltIQ
Financial Modeling Prep (FMP) API – Real-time and historical market data, company financials, and economic indicators
Federal Reserve Economic Data (FRED) API – Macroeconomic data and economic research
Web Search (when enabled) – Current information from the web using Anthropic's search infrastructure
Note: Currently you need to specify which source you want to use in agent mode. Having the agent select automatically is on our roadmap.
Do I need Agents enabled to use FMP?
Yes. External API integrations like FMP require Agents to be enabled.
Can I see what the Agent is doing?
Absolutely. ToltIQ provides complete transparency into Agent activity:
Real-Time Visibility – Watch which tools the Agent calls, what data it requests, and the reasoning behind its decisions
Human-Readable Logs – All thinking processes and tool executions are displayed in accessible language in your chat interface
This transparency ensures you understand exactly how the Agent arrived at its answer and can verify the sources used.
What's the difference between using chat with and without Agents?
The advantages of using agents in it’s current release is to enable showing model reasoning and use FMP. This is just the beginning of a suite of enhanced agentic tools and processes!
Need Help?
If you have questions about using Agents or need assistance configuring your data sources, contact the ToltIQ support team through the in-app chat bot (?) button on the bottom right of your screen or reach out to your ToltIQ account manager.