AI Agents Overview

Spark! includes six specialized AI agents that automate the most time-consuming aspects of affiliate program management. Powered by Chargezen's Zen AI platform, these agents work 24/7 to optimize your program while you focus on strategy and relationships.

Each agent specializes in a specific area of affiliate management. They work independently but share insights to create a cohesive, intelligent system.

All six agents follow the same operational model, designed to balance automation with human oversight:

Agents run in the background 24/7, analyzing data streams in real-time. They don't wait for you to check dashboards—they proactively surface insights and opportunities.

• Discovery Agent — Scans creator databases daily for new matches
• Listening Agent — Monitors social platforms every 15 minutes
• Performance Agent — Analyzes metrics hourly, generates weekly reports
• Fraud Agent — Evaluates every conversion in real-time
• Payout Agent — Processes commissions daily, runs payouts on schedule
• Outreach Agent — Manages sequences and follows up automatically

Agents take action based on configurable triggers. You define the rules, agents execute:

• Event-based — New affiliate signup, order placed, mention detected
• Threshold-based — Affiliate hits revenue milestone, risk score exceeds limit
• Schedule-based — Weekly performance reviews, monthly payout processing
• Pattern-based — Engagement drop detected, unusual activity flagged

Agents present recommendations through a unified Agent Insights Dashboard. Depending on your autonomy settings, you can review and approve actions before they're taken.

• Pending Actions — Queue of recommendations awaiting your approval
• Recent Actions — Log of what agents have done automatically
• Action Details — Full context and reasoning for each recommendation
• Override Controls — Reject, modify, or manually execute any action

Agents improve based on your feedback. Every approve, reject, or modification teaches the system your preferences:

• Approval patterns — Which recommendations you consistently approve
• Rejection reasons — Why certain suggestions don't fit your program
• Manual overrides — Actions you take instead of agent suggestions
• Outcome data — Results of agent-driven vs. manual decisions

Configure how autonomous each agent operates. You can set different autonomy levels for different agents based on your comfort level and the risk profile of each action type.

Agent acts without waiting for approval. Best for low-risk, high-volume tasks where speed matters more than individual review.

• Example actions: Save UGC to library, send welcome email, log data
• Best for: Listening Agent content collection, Payout Agent calculations
• You see: Summary reports of actions taken

Agent recommends actions and waits for your approval. Balances automation efficiency with human judgment for important decisions.

• Example actions: Approve affiliate, send outreach, process payout
• Best for: Discovery Agent approvals, Outreach Agent messages, Fraud Agent holds
• You see: Pending action queue with one-click approve/reject

Agent provides recommendations but never takes action. You execute manually based on insights. Good for learning how agents think before trusting automation.

• Example actions: Suggest tier promotion, flag suspicious activity, recommend outreach
• Best for: New users, high-stakes decisions, compliance-sensitive programs
• You see: Recommendations with full reasoning, you take action manually

Agent is turned off completely. Data collection continues but no analysis or recommendations are generated. Use this temporarily during program restructuring.

The central hub for monitoring and interacting with all six agents. Access it from the main navigation under AI → Agent Insights.

• Agent Status Cards — Health status, last run time, actions today
• Pending Actions Count — How many recommendations await review
• Impact Summary — Revenue driven, fraud prevented, time saved
• Quick Actions — Jump to pending queue, view settings, run manual scan

• Unified Queue — All pending recommendations across agents
• Filter by Agent — Focus on specific agent recommendations
• Bulk Actions — Approve or reject multiple items at once
• Detail Expansion — Click to see full reasoning and context

• Chronological Log — Every agent action with timestamps
• Action Types — Filter by action type (approve, reject, alert, etc.)
• Outcome Tracking — See results of past agent decisions
• Export — Download activity log for compliance or analysis

• Per-Agent Autonomy — Configure autonomy level for each agent
• Notification Preferences — How and when to be notified
• Trigger Configuration — Customize thresholds and rules
• Training Feedback — Review and adjust agent learning

The six agents share data and insights to create an intelligent, coordinated system. Here's how they collaborate:

When Discovery Agent finds high-fit creators, it can automatically trigger Outreach Agent to begin recruitment sequences. Configure this in Discovery Agent settings:

• Set minimum fit score to trigger outreach (e.g., 80+)
• Choose outreach sequence template
• Set daily outreach limits to avoid overwhelming creators

When Listening Agent detects someone mentioning your brand who isn't an affiliate, it flags them for Discovery Agent to evaluate as a potential recruit.

Fraud Agent assigns risk scores to every conversion. Payout Agent uses these scores to automatically hold suspicious commissions for review before processing payments.

When Performance Agent detects declining engagement from a previously active affiliate, it can trigger Outreach Agent to send re-engagement messages automatically.

Track the ROI of your AI agents with built-in impact metrics:

Based on aggregated data from Spark! customers:

• 38% higher response rate — Outreach Agent vs. generic templates
• 67% of revenue — From top 10 creators (identified by Discovery + Performance)
• $4,200+ prevented quarterly — Fraudulent commissions caught by Fraud Agent
• 142+ monthly mentions — Untagged brand mentions detected by Listening Agent
• 15+ hours saved weekly — Time saved on manual affiliate management tasks

Agents are enabled by default when you set up Spark!. To configure them:

1. Go to AI → Agent Insights → Settings
2. Review each agent's current autonomy level
3. Configure notification preferences
4. Set any custom thresholds or rules
5. Monitor the Pending Actions queue daily for the first week