What is Konvo AI?
Think of your Konvo AI as a new team member. Right now, it has zero knowledge about your business or customers. Through configuration, you'll teach it what to know, how to communicate, and what actions to take, turning it into an autonomous support agent that represents your brand.
This article explains the building blocks of your AI and how they fit together. Each component has its own detailed guide in the articles that follow.
The building blocks
Your AI is made up of three core components. Each one answers a different question.
1. Knowledge Hub: what your AI knows
The Knowledge Hub is your AI's memory. It stores all the information your AI needs to answer customer questions: website content, product pages, FAQs, policies, documents, and custom answers you've written.
When a customer asks a question, your AI searches the Knowledge Hub for relevant information and responds based on what it finds.
Example: A customer asks "What's your return policy?" Your AI searches the Knowledge Hub, finds the return policy, and explains it.
📖 Full guide: Knowledge Hub (2.1)
2. AI Persona: how your AI communicates
Your AI Persona controls tone, personality, and brand voice. It shapes every message your AI sends, whether it's answering a question, executing a process, or recommending a product.
The same information sounds different depending on the persona.
Casual: "Hey! Our return window is 30 days. Just keep your receipt and you're good to go 👍"
Professional: "Our return policy allows returns within 30 days of purchase with proof of receipt."
📖 Full guide: AI Persona (2.2)
3. Skills, Processes & Product Recommendation: what your AI can do
This is what transforms your AI from a chatbot into an agent.
Skills are the individual actions your AI can perform: checking order status, cancelling orders, updating shipping addresses, processing refunds, managing subscriptions, escalating to a human agent.
Processes are the step-by-step instructions that tell your AI how to use those skills. They define the workflow: what information to gather, what conditions to check, what to do in different scenarios, and when to escalate.
The distinction matters. A skill gives your AI the ability to cancel an order. A process tells it how: ask for the order number, verify the customer's identity, check whether the order has shipped, then either cancel or escalate.
Product Recommendations are a special capability that lets your AI suggest relevant products from your catalog based on what the customer is looking for.
📖 Full guides: Skills (2.3), Processes (2.4), Product Recommendations (2.5)
How the components work together
Your AI doesn't use these components in isolation. Every customer interaction involves multiple building blocks working together.
Example 1: Simple Question
Customer asks: "What's your return policy?"
What happens:
AI recognizes this is a question (not an action)
AI searches Knowledge Hub for return policy information
AI formats response using Persona tone
AI sends answer to customer
Components used: Knowledge Hub + Persona
Example 2: Action Request
Customer says: "I want to cancel my order"
What happens:
AI recognizes this requires the Cancel Order skill
AI follows the Process for order cancellation:
Asks for order number and email
Checks order status
Sends OTP for verification
Cancels the order
AI confirms cancellation using Persona tone
Components used: Skills + Processes + Persona
Example 3: Product Recommendation
Customer asks: "What products are good for sensitive skin?"
What happens:
AI recognizes product recommendation request
Product Recommendations searches product catalog for "sensitive skin" matches
AI explains why these products fit using info from Knowledge Hub
AI presents 4-5 products with links and descriptions using Persona tone
Components used: Product Recommendations + Knowledge Hub + Persona
Example 4: Complex Multi-Step Flow
Customer says: "I need to change my shipping address"
What happens:
AI triggers Process: Edit Shipping Address
Process Step 1: Ask for order number and email
Process Step 2: Use Skill to check order status in Shopify
Process Step 3 (Branch):
If order not shipped → Use Skill to update address → Confirm with customer
If order already shipped → Check with 3PL via side conversation → Update based on response
Throughout: AI communicates using Persona tone
Components used: Processes + Skills + Persona
Priority order
When your AI receives a message, it evaluates what to do in this order:
Processes (highest priority). If a process trigger matches the customer's message, that process takes control. This means processes override Knowledge Hub content on the same topic.
Skills. If no process triggers, the AI checks whether the customer needs an action (like checking order status).
Knowledge Hub. If it's a question rather than an action request, the AI searches for information to answer it.
Persona (always active). No matter which component handles the interaction, the persona shapes how the AI communicates.
This priority order is important to understand. If you create a process for order cancellations, it will take priority over any cancellation-related content in the Knowledge Hub.
Recommended setup sequence
If you're configuring your AI for the first time, follow this order:
Step 1: Knowledge Hub (2.1). Give your AI information to work with. Connect your website, add FAQs and policies, create custom answers for important topics. Goal: your AI can answer common questions accurately.
💡 Also do: AI Persona (2.2). Configure your AI's tone and brand voice. You can do this at any point during setup, but many teams pair it with the Knowledge Hub so they can test both together.
Step 2: Skills (2.3). Enable skills that work immediately, like "Get Order Status." This is your quickest win since "where is my order?" is typically the most common question.
Step 3: Processes (2.4). Build processes for actions that need step-by-step instructions: order cancellation, address changes, refund requests. Goal: your AI can take actions, not just answer questions.
Step 4: Test (2.6). Run test conversations covering your most common customer scenarios. Rate responses, identify gaps, fix them, and re-test.
Step 5: Deploy (2.7). Activate your AI on your communication channels (WhatsApp, Live Chat, Email) and configure escalation behavior.
Step 6: Monitor and improve (2.8). Use the Dashboard to track performance, identify knowledge gaps, and continuously refine.
Key principles
Your AI is consistent across channels. The same AI (same knowledge, same skills, same processes) works on every channel. The AI adapts its format (conversational for chat, structured for email), but the intelligence is identical.
Processes override Knowledge Hub. Use this strategically. Put general information in the Knowledge Hub. Use processes when you need precise control over how something is handled.
Build comprehensive processes. You cannot trigger one process from another. If a workflow has multiple branches, build them all within a single process.
Start simple, then expand. Begin with the Knowledge Hub and one or two skills. Add processes gradually based on real customer demand. Over-complicating early leads to more errors than value.
AI configuration is iterative. Your AI won't be perfect on day one. Launch with a solid foundation, monitor performance weekly, and improve based on what you see in the Dashboard.
What to Explore Next
→ 2.1 Knowledge Hub — start here
Common mistakes
Overlapping processes. Having multiple processes on the same topic causes unpredictable behavior. One process per topic.
Skipping testing. Always test your most common customer scenarios before going live.
Forgetting to re-sync. If your website content changes, re-sync in Konvo so your AI has the latest information.
Empty Knowledge Hub. An AI without information can't help customers. Populate it before enabling your AI on any channel.