Agent use cases

Not sure where to start with AI agents? Here are real use cases from app businesses like yours.

Each agent can work in two ways:

• Chat — Ask questions and get answers on demand. Perfect for exploration and ad-hoc analysis.

• Flows — Automate tasks that run in the background based on triggers. Perfect for repetitive work and event-driven automation.

How agents work in flows

When you add an AI agent to a flow, you give it instructions—just like you would in chat. The agent runs those instructions automatically when the flow triggers, then passes its response to the next action (like sending a Slack message or tagging a customer).

Think of it as writing a prompt that runs on autopilot.

Start with templates, then build your own

Mantle includes pre-built template agents that are ready to use out of the box. These give you a head start—just enable the tools you need and start chatting. You can customize them by adding rules to match your workflow.

Want something more specific? You can create your own custom agents from scratch, tailored exactly to your needs.

Customer insights agent

Turn your customer data into actionable insights without building reports or writing queries.

Tools to enable:

• Get apps

• Search customers

• Get customer

• Get subscriptions

• Get plans

• Customer analytics (metrics)

For advanced flows (usage analysis):

• Usage events (metrics)

• Usage metrics

• Search tickets

• Get ticket

Chat example

Ask questions about your customers and get instant answers:

• "Who are my top 5 customers by lifetime value?"

• "Show me customers who upgraded in the last 30 days"

• "Find customers on trial that have been active in the last 7 days"

Simple flow

Trial activity check — Find active trial customers who haven't converted and notify your team.

• Trigger: Scheduled run (9:00 AM daily)

• AI instructions: "Find customers currently on trial who have been active in the last 7 days but haven't converted yet. Summarize their activity and recommend which ones to reach out to."

• Then: Post the agent's response to Slack (#sales)

Flow configuration:

The scheduled trigger runs the AI agent daily at 9am. The second step uses "Action completed" to fire after the agent finishes, then posts the response to Slack using {{ previousActions.runAgent.outputs.response }}.

Example output:

Advanced flow

Trial conversion analysis — Analyze what's working in your trial experience and identify patterns in conversions.

• Trigger: Scheduled run (weekly, Monday)

• AI instructions: "Analyze trial conversions from the past week. Compare customers who converted vs those who didn't—look at plan chosen, time to convert, and any patterns. Identify what's working and suggest improvements to the trial experience."

• Then: Post insights to Slack (#growth)

Flow configuration:

A weekly schedule gives the AI enough data to identify meaningful patterns. The agent compares converted vs churned trials using subscription and customer data—no usage events required.

Example output:

Support sidekick agent

Help your support team resolve tickets faster by automatically gathering context, searching documentation, and suggesting responses.

Tools to enable:

• Get apps

• Search customers

• Get customer

• Get subscriptions

• Search tickets

• Get ticket

• Search saved replies

• Get app documentation repositories

• Search app documentation

• Get app documentation page

Optional (highly recommended):

• Connect your codebase — lets the agent understand how your app actually works

Chat example

Get customer context and find answers without switching tabs:

• "What plan is customer [email] on and when did they last login?"

• "Show me this customer's subscription history"

• "Search our docs for articles about webhooks"

Simple flow

Auto-tag tickets — Categorize incoming tickets automatically based on content.

• Trigger: Ticket message is received

• AI instructions: "Read this support ticket and categorize it. Return a single category: billing, technical, feature-request, or general."

• Then: Tag the ticket with the category

Flow configuration:

When a ticket comes in, the AI categorizes it. The response schema ensures the agent returns a structured category field. The "Action completed" trigger then applies that category as a tag using {{ previousActions.runAgent.outputs.category }}.

Example output:

Advanced flow

Smart response suggestion — Analyze tickets, gather context, search docs, and create response suggestions for your team.

• Trigger: Ticket message is received

• AI instructions: "Analyze this ticket. Look up the customer's account, search our docs for relevant articles, and draft a helpful response. Also assess priority (high/medium/low) based on urgency."

• Then: Create response suggestion → Set ticket priority → Post to #support-queue

Flow configuration:

This flow chains multiple actions after the AI runs. Each action can reference different parts of the agent's response—the draft goes to the suggestion, the priority sets the ticket field, and a summary posts to Slack.

Example output:

Content writer agent

Keep your documentation up to date. The agent can write new docs, update existing ones, and identify gaps based on what your customers are actually asking about.

Tools to enable:

• Get app documentation repositories

• Get app documentation tree

• Get app documentation page

• Search app documentation

• Create app documentation page

• Update app documentation page

• Create app documentation group

• Create FAQ

• Update FAQ

Chat example

Write and update documentation through conversation:

• "Write a help article explaining how to set up webhooks"

• "Update the billing FAQ to include our new annual discount"

• "What questions are customers asking that we don't have docs for?"

Simple flow

Daily docs gap report — Analyze recent support tickets and send a summary of documentation gaps.

• Trigger: Scheduled run (daily)

• AI instructions: "Review support tickets from the last 24 hours. Identify questions that our documentation doesn't answer well. List the top 3 gaps with suggestions for new articles."

• Then: Post the report to Slack (#docs-team)

Flow configuration:

The AI agent analyzes yesterday's tickets and identifies documentation gaps. The Slack action posts the full response so your docs team can review and prioritize.

Example output:

Advanced flow

Auto-draft from support patterns — Review weekly tickets, identify top issues, and draft help articles for review.

• Trigger: Scheduled run (weekly, Friday)

• AI instructions: "Analyze this week's support tickets. Find the top 3 recurring issues that need documentation. For each, draft a help article with clear steps to resolve the issue."

• Then: Post summary to Slack (#docs-team) → Send notification email with weekly draft summary

Flow configuration:

The agent has documentation tools enabled, so it creates draft pages directly during execution. After the agent finishes, the follow-up actions notify your team via Slack and email so they can review what was drafted.

Example output:

Marketing specialist agent

Connect your customer data to your marketing tools. Find audiences, draft campaigns, and analyze what's working.

Tools to enable:

• Get apps

• Search customers

• Get customer

• Traffic sources (metrics)

• Funnel metrics

• Email performance (metrics)

• Get flows

• Create flow

Chat example

Get marketing insights and draft campaigns:

• "Draft a win-back email for customers who churned in the last 60 days"

• "Which traffic sources have the best trial-to-paid conversion?"

• "Find customers who would be good candidates for a case study"

Simple flow

Review request after upgrade — Send a personalized review request to happy customers.

• Trigger: Subscription is upgraded

• AI instructions: "This customer just upgraded. Look at their usage and draft a short, personalized review request email. Mention specific features they've been using."

• Then: Send a notification email with the review request

Flow configuration:

The "Subscription is upgraded" trigger fires automatically when a customer upgrades. The AI drafts a personalized email, then the "Action completed" trigger sends it as a notification email using {{ previousActions.runAgent.outputs.response }}.

Example output:

Advanced flow

Smart win-back campaign — Analyze why customers churned and create personalized outreach based on the reason.

• Trigger: Subscription is cancelled

• AI instructions: "Analyze why this customer churned. Check their usage history, support tickets, and subscription details. Categorize the reason (price, feature gap, support issue, competitor) and draft a personalized win-back email with an appropriate offer."

• Then: Send notification email with personalized content → Tag customer with churn reason

Flow configuration:

When a subscription is cancelled, the AI analyzes the customer's history to determine why they left. The response schema returns both a churnReason and emailContent—the notification email uses the content, and the customer is tagged with the reason for tracking.

Example output:

Product manager agent

Understand how your product is being used, identify growth opportunities, and spot trends before they become problems.

Tools to enable:

• Get apps

• Search customers

• Get customer

• Activity metrics

• Funnel metrics

• Trial metrics

• Usage events (metrics)

• Usage metrics

• Churn metrics

• Retention metrics

Chat example

Get instant insights into feature usage and customer behavior:

• "Which features are most correlated with customer retention?"

• "Show me usage trends for our API over the last 3 months"

• "What's driving our trial-to-paid conversion rate changes?"

Simple flow

Weekly metrics digest — Compile key product metrics and send a Monday morning summary.

• Trigger: Scheduled run (weekly, Monday 9am)

• AI instructions: "Generate a weekly product metrics digest. Include: active users, trial conversions, top features by usage, and any concerning trends. Keep it brief and actionable."

• Then: Post to Slack (#product-team)

Flow configuration:

A weekly schedule triggers the AI to pull and summarize your key metrics. The agent has access to all your analytics tools, so it can compare week-over-week and flag anything unusual.

Example output:

Advanced flow

Feature launch tracker — Track new feature adoption over 30 days with weekly reports and trend alerts.

• Trigger: Custom event (feature launched)

• AI instructions: "Track adoption of this new feature. Compare usage to our baseline for similar features. Flag if adoption is below 20% after week 1, or trending down. Generate a weekly adoption report."

• Then: Post adoption report to Slack (#product-team) → Send notification email

Flow configuration:

Custom events let you trigger flows from your own app. When you launch a feature, fire a feature_launched event and the AI analyzes adoption immediately—posting results to Slack and sending an email report to the team.

Example output:

Revenue operations agent

Analyze revenue performance, track subscription health, and identify risks and opportunities.

Tools to enable:

• Get apps

• Search customers

• Get customer

• Get subscriptions

• Get transactions

• MRR metrics

• Subscription metrics

• Revenue metrics

• Churn metrics

• Retention metrics

Chat example

Get instant revenue visibility:

• "What's our MRR growth rate this month vs last month?"

• "Show me our revenue churn by plan tier"

• "What's our net revenue retention for the last quarter?"

Simple flow

Weekly revenue report — Send a summary of key metrics every Monday morning.

• Trigger: Scheduled run (weekly, Monday 9am)

• AI instructions: "Generate a weekly revenue report. Include: MRR, MRR growth, churn rate, new subscriptions, and upgrades/downgrades. Highlight any significant changes from last week."

• Then: Post to Slack (#revenue)

Flow configuration:

Same pattern as the metrics digest—scheduled trigger, AI analysis, Slack output. The agent pulls MRR, churn, and subscription data automatically and highlights week-over-week changes.

Example output:

Advanced flow

Churn prevention pipeline — Analyze why customers cancelled and automatically create save offers for recoverable situations.

• Trigger: Subscription is cancelled

• AI instructions: "Analyze this cancellation. Review their usage history, support interactions, and subscription details. Determine if this is recoverable (yes/no) and why. If recoverable, suggest a specific save offer."

• Then: Create task with the suggested save offer → Notify the team on Slack (#revenue)

Flow configuration:

The AI assesses each cancellation and returns a saveOffer suggestion and churnReason. The next step creates a task with the save offer for your team to act on, and posts a summary to Slack so everyone stays informed.

Example output:

Affiliate manager agent

Manage your affiliate program without jumping between screens.

Tools to enable:

• Get apps

• Search affiliates

• Get affiliate details

• Get affiliate programs

• Get affiliate program details

• Search referrals

• List affiliate payouts

• Affiliate metrics

• Affiliate performance (metrics)

Chat example

Get instant affiliate program visibility:

• "Who are my top performing affiliates this month?"

• "Show me pending commission payouts"

• "What's the average LTV of customers referred by affiliates vs organic?"

Simple flow

Monthly performance report — Compile last month's affiliate stats and send to your team.

• Trigger: Scheduled run (monthly, 1st of month)

• AI instructions: "Generate a monthly affiliate report. Include: total referrals, conversions, commissions earned, top 5 affiliates, and any affiliates trending up or down."

• Then: Post to Slack (#partnerships)

Flow configuration:

Monthly scheduled trigger runs on the 1st of each month. The AI compiles affiliate performance data and posts a summary to keep your partnerships team informed.

Example output:

Advanced flow

Affiliate engagement program — Automatically engage affiliates based on their performance tier.

• Trigger: Scheduled run (weekly)

• AI instructions: "Review affiliate performance this week. Identify: top performers (5+ conversions), inactive affiliates (no activity in 30 days), and near-milestone affiliates (close to a bonus tier). Draft appropriate messages for each group."

• Then: Post performance summary to Slack (#partnerships) → Send notification email with engagement report

Flow configuration:

The AI segments affiliates into performance groups and summarizes recommendations for each. The team gets a Slack summary and an email report so they can follow up with the right affiliates.

Example output:

Sales pipeline agent

Manage your sales pipeline, track deals, and stay on top of follow-ups.

Tools to enable:

• Get apps

• Get deal flows

• Get deal activities

• Get deals

• Create deal

• Update deal

• Update deal stage

• Log deal activity

• Get tasks

• Create task

Chat example

Manage your pipeline without leaving your workflow:

• "Show me all deals in the demo scheduled stage"

• "What deals are expected to close this month?"

• "Log a demo completed activity for the Acme Corp deal"

Simple flow

Stale deal alerts — Find deals with no recent activity and create follow-up tasks.

• Trigger: Scheduled run (daily)

• AI instructions: "Find deals with no activity in the last 7 days that aren't closed. For each, suggest a specific follow-up action based on the deal stage."

• Then: Create follow-up tasks → Post summary to Slack (#sales)

Flow configuration:

The AI finds stale deals and suggests follow-up actions for each. The response includes deal IDs and recommended tasks, which the next action uses to create tasks automatically and notify the team.

Example output:

Advanced flow

Deal progression automation — Automatically set up next steps when deals move stages.

• Trigger: Deal stage is updated

• AI instructions: "A deal just moved to {{new_stage}}. Based on this stage, create the appropriate follow-up tasks. Demo scheduled: create prep task with research on the prospect. Proposal sent: create follow-up reminders. Won: trigger onboarding checklist."

• Then: Create stage-specific tasks

Flow configuration:

The "Deal stage is updated" trigger fires whenever a deal moves. The AI receives the new stage via {{ deal.stage }} and decides what tasks to create. The response schema defines taskTitle, taskDescription, and dueInDays fields that map directly to the task creation action.

Example output:

These are starting points—your agents can do much more. Start with the use case closest to your daily work and expand from there.

• Create your own agent — Build a custom agent tailored to your needs

• Agent flows and triggers — Learn how to set up automated flows