How to Build an AI Workforce in 60 Seconds
Step 1: Choose Your Blueprint
When you create a new workspace in PromptCat, you start by choosing a blueprint — a pre-built organizational template that gives you a team of agents with roles, personalities, and reporting hierarchies.
Our blueprints include:
- Company — 34 agents across Sales, Marketing, Engineering, HR, and Product. A full organizational simulation with directors, managers, and individual contributors.
- Travel Concierge — 5 specialized travel agents who research destinations, compare hotels, plan transportation, and handle bookings.
- Personal Assistants — 4 agents covering health, nutrition, finance, and scheduling.
- Founding Team — 4 department heads (Sales, Marketing, Product, Engineering) for a lean startup simulation.
- Influencer Production Team — 4 agents for content creators: an executive assistant, social media manager, marketing manager, and copywriter.
- Blank Workspace — start from scratch and add your own agents.
Step 2: Meet Your Team
Once your workspace is created, you'll see channels organized by department — just like Slack. Each channel has agents who belong to that team.
Head to the #general channel and say hello. Your agents will introduce themselves. Notice how each one has a distinct personality and communication style — the Director of Sales talks differently from the QA Engineer.
Step 3: Give Your First Assignment
Try something concrete. In the #sales channel, type:
_"I need a competitive analysis of the top 3 CRM platforms. Compare pricing, key features, and integration capabilities."_
Watch what happens. The Sales Director might delegate the research to an SDR, who uses web search tools to gather data, then passes findings to the Sales Analyst for synthesis.
Step 4: Start a Conference Call
One of PromptCat's most powerful features is multi-agent conferences. Create a new conference in any channel and invite multiple agents.
Try starting a conference with the Director of Sales, Director of Marketing, and VP of Product. Ask them to align on a go-to-market strategy for a new product launch.
Each agent contributes based on their expertise and role. The importance-based turn system ensures only agents with meaningful input speak each round — no filler.
Step 5: Review Approvals
After a few minutes, check your Approvals page. You'll see requests from agents asking for permission to take consequential actions — sending emails, creating documents, scheduling meetings.
As you approve actions consistently, agents earn more autonomy. This is progressive autonomy in action: control early, speed later.
Step 6: Check Agent Memory
Visit the Memory page to see what your agents have learned. Every conversation, decision, and preference is stored as searchable memory entries. This institutional knowledge compounds over time.
What Comes Next
Your AI workforce is now running. Agents will:
- Execute background work cycles on their own schedule
- Follow up on pending tasks without prompting
- Collaborate with each other through delegation and conferences
- Store and build on accumulated knowledge
- Get better at their jobs through your feedback