Search

Moxo AI

Moxo’s AI Agents operate as embedded process intelligence inside flows - preparing work, validating submissions, and guiding participants through each step. By handling coordination, structure, and readiness at scale, they reduce manual effort and ensure consistency across complex, repeatable processes. Clear, actionable tasks are delivered with the right context, while humans remain responsible for judgment and decisions. Moxo AI is offered as a paid feature for teams looking to accelerate execution across processes. 

 

Key Benefits

  • Faster time to completion
    AI handles preparation, validation, and guidance steps, reducing manual effort and streamlining the path from intake to completion.

  • Consistent, high-quality outcomes
    Defined rules and checks ensure every step is completed accurately and in line with established process standards.

  • Lower operational overhead
    In-workflow guidance reduces back-and-forth questions, enabling participants to progress independently with greater clarity.

  • Scalable governance and control
    Admins manage organization-wide knowledge and restrictions, while flow owners define AI actions and escalation points, ensuring compliant, controlled automation across workflows.

 

Available AI Agents

AI Chat Assistant

Ideal for self-serve support and invoked by @mention, the AI Chat Assistant delivers contextual answers inside workflows by referencing connected knowledge and maintaining awareness of the last 15 messages in its mention thread. Admins and flow owners can configure prompts, response boundaries, and tone controls, ensuring accurate and consistent guidance while reducing support load.

 

AI Review Agent

The Review Agent evaluates File Request submissions against action metadata, configured criteria, and attached knowledge. It can complete review steps independently or operate alongside human reviewers, advancing or reopening actions with feedback based on defined rules. Escalation occurs whenever confidence thresholds are not met, ensuring consistent, governed review at scale while preserving human judgment where required.

 

AI Prepare Agent

The Prepare Agent readies actions for assignees by setting titles and descriptions, attaching or replacing files, and extracting data from workflow content to pre-fill form fields. Preparation logic, extraction sources, and escalation rules are configurable by flow owners, cutting manual coordination and improving consistency across processes.

 

Admin Configuration from the Manage Tab

Administrators manage organization-wide AI agent settings from the Manage tab, providing centralized governance, security, and visibility across all workflows. Key administrative controls include:

Enabling or Disabling AI Agents: Admins can turn on individual AI agents for the entire organization. When disabled in the Manage tab, agents cannot be used in any flow.

Managing Knowledge Sources: Admins can upload, update, or remove centralized knowledge assets used by AI agents across flows, including knowledge bases, policy documents, guidelines, instructions, and internal resources.

Base Prompt: The base prompt defines an AI agent’s core behavior, purpose, and approach to processing inputs. It applies across all flows and cannot be edited, ensuring consistent behavior and outputs.

Additional Prompts: Additional prompts allow admins to further shape agent behavior by defining tone, focus areas, or response constraints. Flow owners can extend these prompts at the template level to tailor behavior for specific processes.

Usage Reports for AI Agents: Admins can access aggregated AI usage reports from the Manage tab, providing visibility into adoption and frequency of agent use across flows. 

 

Step-by-Step Guide to Enabling AI Agents

  1. Go to Manage and click on AI Agents.


     
  2. Connect an AI Service by clicking on Connections. Select Moxo AI or enter an API key for an external LLM. 


     
  3. Toggle on the agents you want to activate: AI Chat Assistant, AI Review Agent, AI Prepare Agent.
  4. Hover over the button that denotes the AI Service per agent and click 'Edit' to Configure admin settings per agent. Click Save to confirm.
  5. Leverage the Usage section in the Manage tab to review templates and actions where agents are used.

 

Configuring AI Service Connections

  1. Navigate to the AI Agents section in the Manage tab.
  2. Click on Connections to manage all AI service integrations.
  3. Add a new AI Service by selecting Connect AI Service.


     
  4. Enter a Title for your AI service and choose your preferred AI service provider.
  5. Configure the AI Service:
    1. Enter the required API key for the selected provider.
    2. Test the connection to verify that the integration is working.
    3. Select the appropriate language model from the available options.
  6. Click Save to proceed.
  7. Once the AI service is successfully configured, you can toggle on the AI agents you want to activate and apply the desired service to an agent. 

 

Flow Template-Level Configuration

  1. Open or create any Flow Template to add an AI Agent.
  2. Locate the relevant AI Agent configuration area.
    • The Chat Assistant is configured under the Details tab.
    • The Prepare Agent is configured in actions upon setting the agent as a preparer role.
    • The Review Agent is set up specifically in File Request actions.
  3. Turn on agent-specific options on the target actions (e.g., require file review on File Request steps or set Prepare Agent as the action preparer).
  4. Update template-level prompts and settings to define rules, tailor behavior, or enforce response restrictions.
  5. Save the respective agent configurations.
  6. Save the template to apply the agents to future flows. 

 

Best Practices for Escalation and Governance 

  1. Apply conservative confidence thresholds during initial rollout, routing ambiguous cases to human review until agent behavior is validated.

  2. Limit agent-accessible knowledge to materials appropriate for participant-facing workflows.

  3. Keep Chat Assistant prompts and response boundaries narrow early on, expanding scope only after accuracy is confirmed.

  4. Review escalations and edge cases in active templates to iteratively refine prompts, rules, and confidence thresholds over time.

 

Troubleshooting & FAQs

How do I ensure responses follow my brand’s tone, language, and compliance guidelines?
Select a default tone for the AI Chat Assistant (formal, friendly, concise, or detailed), then use prompts to define brand language, compliance boundaries, and response constraints.

Can the Chat Assistant browse website links automatically?
The Chat Assistant uses the specific URLs or documents you attach and does not currently crawl linked pages.

Who can see AI Review feedback?
AI Review feedback for flagged submissions is currently shared directly with the assignee. Controls to limit visibility to internal teams are planned for a future release.