About Conversations with AIDP Deep Insights Agents

A conversation with an AIDP Deep Insights agent lets you investigate a question, review the agent’s findings, and ask follow-up questions that build on the results.

AIDP Deep Insights agents can inspect available data sources, issue queries, and perform calculations before presenting their findings. Responses can include explanations, tables, visualizations, and recommendations.

For example, an agent with access to expense and opportunity data can investigate a question such as, “Does travel and meal spending correlate with higher win rates? Recommend where we should adjust spending.”

Verify important results before using them to make decisions.

Chat Window

The chat window provides these components:

Component Description
New chat

Starts a new conversation.

Chat

Lists previous conversations that you can reopen.

First message

Explains the agent’s purpose and can provide sample questions.

Message input area

Lets you enter a question or review a featured prompt before submitting it.

Mode and model selector

Lets you choose the response mode, model, and available reasoning level.

Featured Prompts

Displays prompts prepared by the agent author. You must have permission to see and use featured prompts.

Each conversation has its own context. Ask follow-up questions in the same conversation when you want to build on earlier messages and findings. Start a new conversation when you want to investigate a separate question. Previous conversations remain available under Chats.

Response Modes

Choose the response mode based on the analysis you need:
  • Instant: Use for routine data lookups, basic summaries, straightforward calculations, clearly scoped questions, and follow-up requests. Instant mode prioritizes response speed and uses a relatively short processing workflow.
  • Research: Use for multi-step analysis, trend investigation, root-cause exploration, performance comparisons, pattern or anomaly analysis, and structured analytical reports. Research mode can require additional processing time.

The response time may increase depending on the complexity and ambiguity of your questions.

Models and Reasoning Levels

You can select a model and a reasoning level for your conversation.

The reasoning level controls the amount of model effort applied within the selected response mode:
  • Use a lower reasoning level for simple, clearly scoped requests.
  • Use Medium for a balance between analytical depth and processing time.
  • Use a higher reasoning level for complex requests that require more detailed evaluation.

Response mode and reasoning level are separate settings. For example, you can select Research mode with a Medium reasoning level.