Free cookie consent management tool by TermsFeed

Best Practices for Designing AI Agents

On This Page

Prompting for AI Agents

Learn how to write effective prompts that get the best results from Appian AI. In this course, you'll explore what prompts are, how they're used in Appian, and how to design prompts for AI agents by applying best practices.

Overview

This page provides best practices for designing effective AI agents. Use these guidelines to help your agents perform reliably and efficiently.

For foundational concepts, see About AI Agents. For step-by-step creation guidance, see Create and Configure an AI Agent.

Write effective prompts

Prompts guide your AI agent's reasoning and actions. The quality of the prompt directly affects how well the AI agent performs.

  • Be concise but complete. Provide only the information the AI agent needs to complete its task while including all essential context.
  • Use clear, direct language. Avoid vague or abstract phrases. Instead, describe what you want the AI agent to do in plain terms.
  • Focus on intent and outcome. Make sure your prompt explains the goal and any specific conditions or rules to follow.
  • Include structure where needed. Use bullet lists or numbered steps in longer prompts to improve clarity.

Example

1
2
3
4
5

Review the attached intake form.

1. Identify the request type (support, access, or update).
2. Determine whether the form is complete.
3. Output the decision as "ready for review" or "missing information."

Design simple AI agents first

Start by creating small, focused AI agents that handle a limited part of your process. Once these AI agents perform well, you can combine or expand them to support broader workflows.

  • Begin with use cases such as intake, triage, or data review.
  • Avoid building AI agents that handle every possible step or exception from the start.
  • Validate early versions through testing to refine prompts, tools, and logic before adding complexity.

Define tools with precision

Each tool defines what your AI agent can access or perform. Clear tool definitions help your AI agent make better decisions.

  • Keep tools task-specific. Each tool should do one thing well, such as starting a process or retrieving a record.
  • Avoid overlapping tools. Duplicate or similar tools can cause confusion or reduce accuracy.
  • Review tool names and descriptions. Ensure they clearly describe the tool's purpose and scope.

Use clear and meaningful input names

When defining your AI agent's inputs, choose names that make their purpose obvious. This improves readability and helps the AI agent use data correctly.

  • Use descriptive names, such as customerFormDoc or casePriorityLevel.
  • Avoid generic names like id, data, or info.
  • Match input names to their function in the process model or data fabric.
  • Include brief descriptions in your prompt for each input to clarify intent.

Design modular AI agents for collaboration

When building complex workflows, create small, specialized AI agents that each handle a specific task. You can then use the AI agent tool to have these AI agents collaborate and complete multi-step processes.

  • Start by creating focused AI agents for individual tasks, such as one for triage, one for data extraction, or one for approval routing.
  • Use a main AI agent to orchestrate and coordinate the specialized AI agents.
  • Keep each AI agent's tools task-specific to avoid overlapping functionality between collaborating AI agents.
  • This modular approach improves maintainability and allows you to update individual AI agents without affecting the entire workflow.

Caution:  Do not design multi-agent collaborations deeper than two levels. Limit multi-agent collaboration to two levels maximum. Do not create patterns where

  • a parent AI agent calls a child AI agent that calls another child AI agent,
  • where a child AI agent calls its parent AI agent, or
  • where an AI agent calls itself.

    These patterns can cause infinite loops, excessive concurrent executions, and difficult troubleshooting.

Use lists for bulk processing

When your AI agent needs to process multiple items, use list inputs and outputs to handle them in a single transaction instead of building complex looping logic.

  • Pass multiple documents, records, or text strings directly to your AI agent as a list input.
  • Configure outputs as lists when your AI agent needs to return multiple structured items.
  • This approach reduces orchestration complexity and minimizes the number of tool calls required.
  • Keep array limits in mind: 50 documents/folders or 50 records/primitives per variable.

Monitor and control AI agent runs during testing

While testing your AI agents, actively monitor their execution to catch issues early and conserve resources.

  • Use the Stop button to immediately halt agents that are heading off-track or consuming unnecessary AI actions.
  • Review the step-by-step execution history to identify where agents make incorrect decisions or use wrong tools.
  • Stop and adjust rather than waiting for a flawed run to complete—this saves time and resources during development.
  • Keep the maximum tool call limit in mind (default: 30) when designing your agent's workflow.

For more information about testing, monitoring, and troubleshooting AI agents:

Feedback