See SaaS Trial Flow in Action
Follow Sarah's journey from trial signup to paid customer and discover every feature that made it happen
Scroll to explore the complete journey and features
Sarah's 14-Day Journey to Becoming a Customer
Watch how AI agents, powered by your product knowledge, nurture a trial user from signup to conversion
The Result: $12,000 ARR Customer
Sarah went from trial signup to paid customer in 14 days with zero manual intervention. The AI agents handled 12 touchpoints, responded to 3 questions with accurate, source-backed answers, and escalated at the perfect moment. Total agent cost: $0.42 (14 days × $0.03/day AI processing).
Platform Capabilities
Deep dive into every feature that powers intelligent trial nurturing
Tracking & Data Collection
Intelligent Event Tracking
Know exactly what every trial user is doing, in real-time
Technical Details
RESTful API endpoint (/api/track) with API key authentication. Event Router automatically processes events, creates trial associations, and triggers agent notifications. All events stored in PostgreSQL with JSONB payloads for flexibility. Real-time notifications via Event Notification Service trigger immediate agent responses.
Key Features
- RESTful API with API key authentication
- Auto-collection of browser/device data (device, browser, location, session)
- UTM parameter tracking for attribution
- Offline event queue with automatic retry
- Flexible JSONB event payloads
- Event-to-trial auto-association
- Processing status tracking (processed/pending/error)
- Real-time agent notifications
- Event count requirements (min_count per event)
- Time window constraints
Event Flow: SDK → API → Database → Agent Notification
Competitive Advantage
Most platforms require complex event schema setup. Our system auto-collects essential data and accepts any custom properties via JSONB, making integration trivial.
JavaScript SDK Integration
8-minute setup with automatic data collection, no backend work required
Technical Details
TypeScript-based npm package (saastrialflow-sdk) under 25KB gzipped. Full type definitions included. Offline support with event queue and exponential backoff retry. Session tracking with automatic continuation.
Key Features
- Simple 3-line integration
- Offline support with event queue
- Automatic retry with exponential backoff
- Session tracking and continuation
- UTM parameter auto-capture
- Custom event properties
- Debug mode for development
- Error callbacks for monitoring
- Manual flush control
- Full TypeScript types
import { initSaasTrialFlow } from 'saastrialflow-sdk';
// Initialize once in your app
const analytics = initSaasTrialFlow({
apiKey: process.env.NEXT_PUBLIC_SAASTRIALFLOW_API_KEY
});
// Identify user
await analytics.identify("user@example.com", {
name: "Sarah Chen",
company: "Acme Corp"
});
// Track events with custom properties
await analytics.track("workspace_created", {
workspace_name: "Marketing Team",
plan_tier: "pro"
});
// SDK handles:
// ✓ Offline queue if user loses connection
// ✓ Auto-retry with exponential backoff
// ✓ Session tracking across page loads
// ✓ UTM parameter capture
// ✓ Browser/device data collectionCompetitive Advantage
Unlike analytics tools that require backend integration, our SDK is 100% client-side. No server setup, no API keys in backend code, no security risks.
Trial Journey Management
Stage-Based Trial Funnel
Customize your trial journey, track progress in real-time, and guide users through every step
Technical Details
4 predefined stages: Registration → Activation → Adoption → Commitment. Each stage has customizable event requirements (required vs optional), time limits, and progression rules. Database tables: pipeline_stages, stage_event_requirements. Sequential progression with automatic agent handoffs.
Key Features
- 4 customizable trial stages
- Stage-specific event requirements
- Required vs optional event tracking
- Time-based stage limits (optional)
- Real-time progress tracking (percentage completion)
- Automatic stage progression on completion
- Visual pipeline health indicators
- Multi-tenant isolation
- Export/import configuration
- Event count minimums (e.g., 'used feature 3 times')
4-Stage Progression with Automatic Handoffs
Competitive Advantage
Fixed email sequences break when users don't follow the 'happy path'. Stage-based progression adapts to how users actually behave, ensuring relevant messaging at every step.
Trial CRM
Complete trial visibility and manual control when you need it
Technical Details
Database table: trials. Full CRUD operations via API. Custom fields stored as JSONB for flexibility. UTM attribution tracking. Unsubscribe status management. Multi-tenant isolation via RLS policies.
Key Features
- Contact details (name, email, company)
- Trial period tracking (start/end dates)
- Current stage and progress percentage
- Last activity timestamp
- Status (active, paused, converted, churned)
- UTM attribution tracking
- Custom field storage (JSONB)
- Pause/resume trials
- Manual status changes
- Export data (JSON/CSV)
Complete Trial Timeline
See every event, email, and agent decision in chronological order
Technical Details
API endpoint: /api/trials/[id]/timeline. Merges emails, agent actions, and events chronologically. Includes full decision reasoning, tool usage, and engagement metrics. Real-time updates.
Key Features
- Chronological activity feed
- Email sent/received tracking
- Engagement visualization (positive/neutral/negative)
- Agent decision logs with reasoning
- Event completion status
- Stage transitions
- Status changes
- Error tracking and recovery
- Expandable details for each item
- Action reasoning display
Use Cases
Debugging Low Engagement
See exactly when a user went silent, what emails were sent, and how agents adapted
Understanding Conversions
Identify the exact touchpoint that led to conversion and replicate across trials
AI Agent System
AI-Powered Decision Engine (LangGraph)
Contextual decisions based on real behavior, not scheduled sequences
Technical Details
Built on LangGraph StateGraph with OpenAI GPT-4. Dual-phase decision making: (1) Analysis Phase gathers context, (2) Action Phase executes decisions. 8 decision nodes: Load Context, Classify Reply, Analyze Conversation, Decision Engine, Generate Response, Schedule Wait, Update Memory, Handoff Workflow. Full reasoning logged in agent_actions table.
Key Features
- LangGraph StateGraph architecture
- 8 specialized decision nodes
- Dual-phase decision making (analysis + action)
- Context gathering from multiple sources
- Behavioral pattern analysis
- Intent signal detection
- Confidence scoring (0-1 scale)
- Full decision reasoning logs
- Error recovery and retry logic
- Graceful degradation on failures
// Agent analyzes context
const context = {
daysSinceSignup: 2,
eventsCompleted: 1,
requiredEvents: 3,
lastEmailSent: "2 days ago",
emailOpened: true,
lastActivity: "1 day ago"
};
// LLM-powered decision with reasoning
const decision = await decisionEngine.decide(context);
// Returns:
// {
// action: "SEND_EMAIL",
// reasoning: "User opened last email but hasn't completed next event. Send guidance.",
// confidence: 0.87,
// emailType: "technical_support",
// waitTime: null
// }Competitive Advantage
Fixed sequences send 'Day 3' emails on Day 3, regardless of user behavior. Our Decision Engine adapts in real-time: if a user logs in at 2 AM after 10 days of silence, agents respond immediately.
Autonomous Email Agents
24/7 intelligent nurturing without any manual work
Technical Details
Each trial gets a dedicated agent instance. 4 agent types (Registration, Activation, Adoption, Commitment) with different goals. Automatic handoffs with context transfer. Agents run on cron schedule (every 2 hours) plus real-time event triggers.
Key Features
- 4 stage-specific agent types
- Dedicated agent per trial
- Automatic stage handoffs
- Context transfer between agents
- Goal-oriented behavior (conversion, engagement, education)
- Cron-based + event-driven execution
- State management (active, waiting, completed, paused)
- Error recovery and retry
- Manual takeover capability
- Human escalation triggers
Intent-Based Email Replies
AI detects intent and responds contextually in seconds
Technical Details
Advanced pattern matching and LLM-based classification. 8 intent types: technical_support, use_case_discovery, value_proposition, feature_education, acknowledgment, conversation_ending, unsubscribe_handling, human_escalation. Sentiment analysis (positive/neutral/negative). CAN-SPAM/GDPR compliant unsubscribe handling.
Key Features
- 8 intent types with pattern matching
- LLM-based intent classification
- Sentiment analysis (positive/neutral/negative)
- Question extraction from replies
- Unsubscribe detection and handling
- Human escalation signals
- Priority scoring (0-100)
- Context-aware responses
- Conversation history integration
- Graceful error handling
Use Cases
Technical Question
User asks 'How do I import data?' → Agent detects technical_support intent → Searches RAG → Responds with help article
Pricing Inquiry
User mentions 'pricing' or 'cost' → Agent detects value_proposition intent → Pulls pricing from RAG → Escalates to sales if high intent
Feature Request
User asks 'Do you have X?' → Agent searches knowledge base → Either confirms with docs or escalates to human
Unsubscribe
User says 'unsubscribe' or 'stop' → Auto-marks as unsubscribed → Sends confirmation → Never emails again
RAG Knowledge Base
AI that knows YOUR product: Accurate answers grounded in your actual content
The Problem with Generic AI Email Tools
- ×Generic AI tools hallucinate or give vague answers about YOUR product
- ×Manual prompts can't cover every product detail, pricing tier, or integration
- ×Customers ask specific questions: "Do you integrate with X?" "How does Y feature work?"
- ×Wrong answers destroy trust and tank conversion rates
How RAG Solves This
RAG (Retrieval-Augmented Generation) automatically imports your website and help center content, stores it in a vector database, and retrieves the most relevant snippets when generating emails. Your AI agents cite real sources with URLs, building trust and ensuring accuracy.
How RAG Works
Behind the Scenes: Vector Search
// User question triggers RAG search
const question = "How does pricing work?";
// Generate embedding
const embeddings = new OpenAIEmbeddings({
modelName: 'text-embedding-3-small'
});
const queryEmbedding = await embeddings.embedQuery(question);
// Vector similarity search (pgvector + HNSW)
const { data } = await supabase.rpc('match_agent_knowledge', {
query_embedding: queryEmbedding,
filter_company_id: companyId,
match_threshold: 0.65,
match_count: 5
});
// Returns: Top 3-5 most relevant snippets
// [
// {
// content: "Our pricing has 3 tiers...",
// similarity: 0.89,
// sourceUrl: "https://yoursite.com/pricing",
// sourceTitle: "Pricing Plans"
// }
// ]Technical Architecture
agent_knowledge_sources (original docs) + agent_knowledge_chunks (embedded vectors)Key Features
Real-World Use Cases
Product Questions
Feature Inquiries
Objection Handling
Technical Support
Massive Competitive Advantage
Most AI Email Tools:
- ×Use generic LLM knowledge (prone to hallucinations)
- ×Give vague, unhelpful answers to product questions
- ×No source citations or verification
- ×Break trust when customers fact-check
SaaS Trial Flow with RAG:
- ✓Company-specific, accurate, verifiable responses
- ✓Every answer includes source URLs for transparency
- ✓Builds trust through citation-backed information
- ✓Critical for complex B2B SaaS products
This isn't just a feature. It's what makes AI trial nurturing actually work for B2B SaaS.
Email & Communication
Custom LLM Prompts
Fully customizable AI personality and messaging for your brand
Technical Details
Two-table architecture: email_prompts (system defaults, no company_id) + company_email_prompts (company-specific instances). 21+ default prompts across 7 rotation groups (welcome, onboarding, value, discovery, education, relationship, technical). Full editing of system + user templates. Version tracking and rollback support.
Key Features
- 21+ default email templates
- 7 rotation groups for variety
- Company-specific customization
- Full prompt editing (system + user sections)
- Tone selection (professional, friendly, enthusiastic)
- Word count limits
- Creativity level control (0-1)
- Agent type assignment
- Intent type mapping
- Priority scoring
- Version tracking
- Revert to defaults
Competitive Advantage
Generic AI tools use one-size-fits-all prompts. Our dual-table system lets you customize every email while maintaining upgradability from system defaults.
Conversation Intelligence (Multi-Inbox)
Unified inbox for all trial communications with smart filtering
Technical Details
Database tables: emails, conversation_threads, conversation_views. RFC-compliant message threading (In-Reply-To headers). Pre-computed view counts for performance. Real-time read/unread tracking (only for inbound replies). Multi-tenant isolated.
Key Features
- 8 inbox categories (All, Trial Replies, Campaign Replies, Bounced, Unsubscribed, Active, Manual, Paused, Completed)
- Full email threading (In-Reply-To headers)
- Read/unread indicators (inbound only)
- Agent type filtering
- Status filtering
- Search by name/email/company
- View modes (compact/detailed)
- Manual takeover from AI
- Send manual replies
- Pause AI responses
- Conversation handoff
White-Label Email Sending
Send from your own domain, maintain your brand identity
Technical Details
Database table: smtp_configurations. Supported providers: Mailgun, SendGrid, AWS SES, Postmark. SMTP fallback support. DNS record generation for SPF, DKIM, DMARC. Domain verification workflow. Test email capability.
Key Features
- Custom from address (you@yourdomain.com)
- Custom reply-to address
- SPF/DKIM/DMARC setup wizard
- DNS record auto-generation
- Domain verification
- Multiple provider support (Mailgun, SendGrid, AWS SES, Postmark)
- SMTP fallback
- Test email sending
- Branded email headers
- Company-specific templates
# Add these DNS records to your domain:
# SPF Record (TXT)
Host: @
Value: v=spf1 include:mailgun.org ~all
# DKIM Record (TXT)
Host: smtp._domainkey
Value: k=rsa; p=MIGfMA0GCSqGSIb3DQEBAQUAA...
# DMARC Record (TXT)
Host: _dmarc
Value: v=DMARC1; p=none; rua=mailto:dmarc@yourdomain.com
# Verification complete ✓
# Emails now send from: agents@yourdomain.comEmail Event Tracking
Track opens, clicks, bounces, and trigger real-time agent responses
Technical Details
Database table: email_events. Webhook integration with all major providers. 7 event types: delivered, opened, clicked, bounced, complained, unsubscribed, failed. Metadata captured: client info (OS, browser, device), geolocation (country, region, city), clicked URLs, bounce reasons, complaint feedback, raw webhook payload.
Key Features
- 7 event types (delivered, opened, clicked, bounced, complained, unsubscribed, failed)
- Open tracking (pixel-based)
- Click tracking (link rewriting)
- Bounce detection (hard/soft)
- Spam complaint tracking
- Unsubscribe request handling
- Client detection (OS, browser, device)
- Geolocation tracking (country, region, city)
- Clicked URL capture
- Bounce reason classification
- Real-time webhook processing
- Agent notification triggers
Webhook Flow: Provider → Handler → Database → Analytics
Analytics & Configuration
Trial Analytics & Insights
Understand what's working and optimize your conversion funnel
Technical Details
Pre-computed database views for performance. Real-time metrics via dashboard endpoints. Email analytics with delivery, open, click, bounce tracking by provider.
Key Features
- Trial metrics by stage (count, conversion rate, avg time)
- Drop-off analysis per stage
- Engagement scoring
- Agent performance metrics (email send rate, response rate, goal achievement)
- Email analytics (delivery rate, open rate, click rate, bounce rate)
- Provider performance comparison
- Prompt effectiveness tracking
- Rotation usage analytics
- Custom date range filtering
- Export capabilities
Agent Configuration & Settings
Fine-tune AI behavior for your specific product and users
Technical Details
Database table: agent_configurations. JSONB configuration storage for flexibility. Stage-specific settings (Registration, Activation, Adoption, Commitment). Model selection, temperature control, max tokens, follow-up limits, wait time schedules.
Key Features
- AI model selection (GPT-4, GPT-3.5, etc.)
- Temperature control (creativity vs consistency)
- Max tokens per response
- Follow-up email limits
- Wait time schedules (min/max hours between emails)
- Decision thresholds
- Engagement trigger rules
- Escalation rules
- Goal definitions per stage
- Context window settings
- Stage-specific configuration
- Company-wide defaults
Lead Management & Engagement Tools
AI-Powered Lead Forms
Capture leads from any source with intelligent AI follow-up
Technical Details
Database table: lead_forms. Each form has dedicated AI agent configuration with custom tone, goal, and context. AI-powered field mapping automatically matches form fields to contact schema. Webhook submissions with token authentication. Auto-start agents with configurable delay.
Key Features
- Custom form builder with drag-and-drop interface
- AI field mapping (auto-match form fields to contacts)
- Webhook submission endpoint (/api/forms/[slug]/submit)
- Per-form AI agent configuration (tone, goal, context)
- Auto-start agents with delayed follow-up (hours)
- Default lifecycle stage assignment
- Auto-tagging on submission
- Custom unsubscribe text per form
- Webhook token authentication
- Submission tracking and analytics
- Field mapping approval workflow
- Integration with contact CRM
// Submit to your form endpoint
const response = await fetch(
'https://yourapp.com/api/forms/contact-form/submit?token=your_webhook_token',
{
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
name: "Sarah Chen",
email: "sarah@acmecorp.com",
company: "Acme Corp",
message: "Interested in enterprise plan"
})
}
);
// Behind the scenes:
// 1. AI maps fields → contact schema
// 2. Creates/updates contact in CRM
// 3. Applies default tags and lifecycle stage
// 4. Starts AI agent after configured delay
// 5. Agent begins personalized nurture sequenceUse Cases
Website Contact Form
Create form → Embed on website → AI maps submissions → Agent follows up based on inquiry type
Event Registration
Conference signup form → Custom agent goal (event prep) → Nurture attendees with event details and pre-event engagement
Demo Request
Demo form with company size/use case → AI prioritizes by fit → High-priority leads get immediate agent response
Competitive Advantage
Most form builders require manual field mapping and separate marketing automation setup. Our AI automatically maps any form structure and starts intelligent follow-up within hours.
Unified Contact CRM
Manage all contacts (trial users, leads, customers) in one centralized database
Technical Details
Database table: contacts. Complete contact lifecycle management with engagement scoring, custom fields (JSONB), tagging system, and subscription management. Bulk operations support. Full-text search across email, names, company. Multi-tenant isolated via RLS policies.
Key Features
- Complete contact profiles (name, email, company, phone)
- Email engagement scoring (0-100)
- Lifecycle stage tracking (lead → trial → customer)
- Flexible tagging system (arrays)
- Custom data fields (JSONB)
- Activity tracking (last contacted, last active)
- Subscription management (opt-in/out)
- Unsubscribe reason tracking
- Source attribution tracking
- CSV/Excel bulk import
- Advanced search and filtering
- Bulk operations (tag, update, delete)
- Integration with trials, forms, campaigns
Use Cases
Complete Customer Journey
Lead (form submission) → Trial user (product signup) → Customer (conversion) - all tracked in one contact record
Multi-Source Lead Management
Import existing customer list + capture new form leads + track trial signups = unified database for segmentation and campaigns
Engagement-Based Workflows
Track email opens/clicks → Calculate engagement score → Segment high-engagement contacts → Target with campaigns or AI agents
Competitive Advantage
Unlike trial-only systems, our CRM handles your entire customer database. Trial users, form submissions, manual entries, imports - all in one place with unified AI agent access.
Dynamic Contact Segments
Create auto-updating contact lists for precision targeting
Technical Details
Database table: segments. JSON-based filter structure with real-time size estimation. Dynamic segments auto-update as contacts change. Supports multiple filter types with AND/OR logic. Cached estimated_count for performance. Used for campaign targeting and analytics.
Key Features
- Dynamic segments (auto-update as contacts change)
- Static segments (fixed point-in-time)
- Multiple filter types:
- • Tags (contains/not contains)
- • Lifecycle stage (registration, activation, etc.)
- • Engagement score (ranges)
- • Creation date (date ranges)
- • Last activity (relative/absolute)
- • Subscription status (subscribed/unsubscribed)
- Real-time recipient count estimation
- Combine filters with AND/OR logic
- Save and reuse for campaigns
- Segment-based analytics
- Export segment contacts
Use Cases
Re-Engagement Campaign
Create segment: Last activity > 30 days ago + Engagement score < 30 + Subscribed = True → Send re-engagement campaign
High-Value Prospects
Filter: Tags contains 'enterprise' + Lifecycle stage = 'trial' + Engagement score > 70 → Priority follow-up campaign
Onboarding Cohorts
Dynamic segment: Created within last 7 days + Lifecycle stage = 'registration' → Welcome campaign series
Competitive Advantage
Most platforms require manual list updates. Our dynamic segments auto-refresh, ensuring your campaigns always target the right people based on real-time data.
Broadcast Email Campaigns
Send beautiful, targeted emails to segments with built-in analytics
Technical Details
Database table: campaigns. Unlayer WYSIWYG email editor integration. Template storage (HTML + JSON). Segment-based targeting with inclusion/exclusion filters. Real-time sending status. Complete email event tracking (delivered, opened, clicked). Campaign overview analytics view.
Key Features
- WYSIWYG visual email editor (Unlayer)
- Drag-and-drop email design
- Segment-based targeting
- Tag inclusion/exclusion filters
- Multi-segment targeting
- Custom from email/name
- Custom reply-to address
- Per-campaign unsubscribe text
- Recipient count estimation
- Send now or save as draft
- Campaign status tracking (draft/sending/sent/cancelled)
- Email analytics (delivered, opened, clicked)
- Real-time sending progress
- Campaign performance metrics
// 1. Create campaign
const campaign = await createCampaign({
name: "Product Launch Announcement",
subject: "Introducing AI-Powered Analytics",
from_email: "team@yourapp.com",
from_name: "Your Product Team",
target_segments: ["high-engagement", "trial-users"],
exclude_tags: ["unsubscribed", "bounced"]
});
// 2. Design email with visual editor
// User builds email in Unlayer WYSIWYG editor
// Template saved as HTML + JSON
// 3. Estimate recipients
const { count } = await estimateRecipients(campaign.id);
// Returns: 1,247 recipients
// 4. Send campaign
await sendCampaign(campaign.id);
// Result:
// - Queues emails to 1,247 contacts
// - Tracks delivery, opens, clicks
// - Updates campaign analytics in real-timeUse Cases
Product Launch
Target all active contacts + high-engagement trials → Beautiful announcement email → Track opens/clicks → AI agents follow up with engaged recipients
Feature Adoption
Segment: Has NOT used feature X + Trial user + Engaged → Educational campaign about feature → Track who clicks → Agent nurtures interested users
Win-Back Campaign
Segment: Last activity > 60 days + Previously engaged → Re-engagement offer → Monitor responses → Reactivate responsive contacts
Competitive Advantage
AI agents handle 1-to-1 nurturing. Campaigns handle 1-to-many broadcasts. Combined, you get personalized follow-up for engaged users and efficient mass communication for announcements, updates, and promotions.
Ready to Deploy Intelligent Trial Nurturing?
AI agents powered by YOUR product knowledge. See RAG in action with your own website during the 8-minute setup.
Join 10 founding companies transforming their trial conversion with AI agents. No credit card required.