The Anatomy of a Genesis App: Memory, Agents, Automation
Understand the internal structure of every Genesis app β how Projects, Agents, and Automations work together to create living software.
What's inside a Genesis app?
When you describe something and Genesis builds it, what actually gets created? Understanding the anatomy helps you build better systems β and debug them when things go wrong.
Let's dissect a real app.
π¬ Why This Matters:
Understanding anatomy = building better apps + debugging faster + customizing effectively.
π― The Three-Layer Architecture
Every Genesis app has the same structure:
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β GENESIS APP β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β β
β LAYER 1: π§ MEMORY (Projects & Databases) β
β βββββββββββββββββββββββββββββββββββββββββββββ β
β Schema, tables, fields, relationships, views β
β β
β LAYER 2: π€ INTELLIGENCE (AI Agents) β
β βββββββββββββββββββββββββββββββββββββββββββββ β
β Personality, knowledge, capabilities, reasoning β
β β
β LAYER 3: β‘ MOTION (Automations) β
β βββββββββββββββββββββββββββββββββββββββββββββ β
β Triggers, actions, integrations, sequences β
β β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
The Subject: Neon CRM Dashboard
We'll examine Neon CRM Dashboard β a sales pipeline management system.
From the outside, it's a CRM. Leads, deals, contacts, pipeline stages. But inside, it's three systems working as one.
Layer 1: Memory (Projects & Databases)
The data foundation. Everything the app knows.
Schema
PROJECTS (Tables)
βββ Contacts
β βββ name (text)
β βββ email (text)
β βββ company (relation β Companies)
β βββ stage (select: lead, qualified, opportunity, customer)
β βββ created_at (date)
β
βββ Companies
β βββ name (text)
β βββ size (select: startup, smb, enterprise)
β βββ industry (text)
β βββ annual_value (number)
β
βββ Deals
β βββ name (text)
β βββ contact (relation β Contacts)
β βββ value (number)
β βββ stage (select: discovery, proposal, negotiation, closed)
β βββ probability (number)
β βββ close_date (date)
β
βββ Interactions
βββ contact (relation β Contacts)
βββ type (select: email, call, meeting)
βββ notes (text)
βββ date (date)
Views
Same data, different perspectives:
- Pipeline view β Deals as kanban by stage
- Table view β All records as spreadsheet
- Calendar view β Interactions and close dates
- Board view β Contacts by stage
Relationships
Connections that create context:
- Contact β Company (who works where)
- Deal β Contact (who owns the opportunity)
- Interaction β Contact (history of engagement)
This is the Memory Pillar.
Layer 2: Intelligence (AI Agents)
The thinking layer. Agents that reason about your data.
CRM Agent
The primary agent, configured with:
Personality:
You are a helpful sales assistant. You help track deals,
suggest next actions, and provide pipeline insights. Be
concise and action-oriented.
Knowledge:
- Access to all CRM tables
- Understanding of sales stages
- Company and industry context
Capabilities:
- Answer questions about pipeline
- Suggest follow-up actions
- Draft outreach emails
- Summarize account status
Agent Interactions
The agent can:
- Query data: "What deals are closing this month?"
- Analyze patterns: "Which stage has the most stalled deals?"
- Suggest actions: "You haven't contacted Acme Corp in 2 weeks"
- Execute tasks: Create interaction record, update deal stage
This is the Intelligence Pillar.
Layer 3: Motion (Automations)
The action layer. Workflows that run automatically.
Trigger β Action Flows
AUTOMATION 1: New Lead Alert
βββ Trigger: New contact created (stage = lead)
βββ Action: Send Slack notification to #sales
βββ Action: Create welcome task for sales rep
AUTOMATION 2: Deal Stale Alert
βββ Trigger: Deal stage unchanged for 14 days
βββ Action: Send email reminder to deal owner
βββ Action: Update deal flag to "at risk"
AUTOMATION 3: Won Deal Sequence
βββ Trigger: Deal stage changed to "closed won"
βββ Action: Update contact stage to "customer"
βββ Action: Send celebration to #wins channel
βββ Action: Create onboarding project
βββ Action: Update revenue forecast
AUTOMATION 4: Weekly Pipeline Report
βββ Trigger: Every Monday 9am
βββ Action: Query all open deals
βββ Action: Calculate weighted pipeline
βββ Action: Send summary to #sales
Integration Points
Automations connect to external services:
- Slack β Notifications and alerts
- Email β Outreach and follow-ups
- Calendar β Meeting scheduling
- Webhooks β External systems
This is the Execution Pillar.
How the Layers Interact
The power is in the connections:
ββββββββββββββββββββββββββββββββββββββββββββββββββββ
β β
β User asks: "Show me stalled deals" β
β β β
β βΌ β
β βββββββββββββββββββββββββββββββββββββββββββ β
β β π€ AGENT (Intelligence) β β
β β - Interprets question β β
β β - Queries memory for deals β β
β β - Analyzes stage durations β β
β βββββββββββββββββββββββββββββββββββββββββββ β
β β β
β βΌ β
β βββββββββββββββββββββββββββββββββββββββββββ β
β β π§ MEMORY (Projects) β β
β β - Returns deals with stage > 14 days β β
β β - Includes contact and value info β β
β βββββββββββββββββββββββββββββββββββββββββββ β
β β β
β βΌ β
β Agent presents: "3 deals stalled: β
β - Acme ($50k, 21 days in proposal) β
β - BigCo ($100k, 18 days in negotiation) β
β - StartupX ($25k, 30 days in discovery) β
β Suggest follow-up actions?" β
β β β
β βΌ β
β User: "Yes, create follow-up tasks" β
β β β
β βΌ β
β βββββββββββββββββββββββββββββββββββββββββββ β
β β β‘ MOTION (Automations) β β
β β - Creates tasks for each stalled deal β β
β β - Sends reminders to deal owners β β
β β - Updates CRM flags β β
β βββββββββββββββββββββββββββββββββββββββββββ β
β β β
β βΌ β
β Memory updated, user notified β
β β
ββββββββββββββββββββββββββββββββββββββββββββββββββββ
This is Workspace DNA in action.
Building Better Apps
Understanding anatomy helps you build:
For Better Memory
- Define clear relationships between entities
- Use appropriate field types
- Create views for different perspectives
For Better Intelligence
- Write detailed agent instructions
- Give agents relevant context
- Define capabilities clearly
For Better Motion
- Map out trigger conditions
- Chain actions logically
- Connect to relevant services
Inspect Your Apps
Every Genesis app has this structure. To improve any app:
- Examine Memory β Is the schema right? Missing fields?
- Review Intelligence β Is the agent helpful? Accurate?
- Check Motion β Are automations firing? Missing triggers?
The anatomy is your debugging guide.
Resources
Learn more:
- Projects & Databases β Memory Pillar
- Custom AI Agents β Intelligence Pillar
- Automations & Workflows β Execution Pillar
- Workspace DNA β Full architecture
Genesis Deep Dives:
- The Origin of Living Software β Where it all began
- How Workspace DNA Works β The architecture explained
- Build Your First AI App β 12 beginner examples
- Build Without Permission β The Genesis Manifesto
Genesis Showcase:
- 10 AI Dashboards β Finance, CRM, ops
- 10 AI Client Portals β Deploy instantly
- 10 AI Booking Systems β Clone for business
- 20 AI App Ideas β Inspiration gallery
Explore Taskade AI:
- AI App Builder β Build complete applications
- AI Agent Builder β Create intelligent assistants
- AI Workflow Automation β Automate any process
Build with Genesis:
- Browse All Generator Templates β Apps, dashboards, websites, and more
- Browse Agent Templates β AI agents for every use case
- Explore Community Apps β Clone and customize
