Bland AI users setting up phone agents for their business.
What goes in
FAQs, policies, product details, and other company information.
Why it matters
Agents use this knowledge on live calls. The easier it is to build, the better agents can answer.
Problem
Disconnected workflows made knowledge hard to build
Creating a knowledge base is not intuitive, with disconnected workflows that make it hard for users to build, organize, and understand their information effectively.
Presentation
Case study walkthrough
Interactions
From concept to micro-interaction
Human In The Loop
The agent prompts users for input and helps complete tasks.
Message Interaction
Animate the response into view so the exchange feels conversational, not abrupt.
Feedback state
Confirm actions with subtle motion so users know their input was received.
Agent Thinking
It shows that the agent is thinking and processing information.
Check confirmation
Reinforce successful completion with a clear, satisfying check interaction.
Identifying as link
Shows that the agent recognizes links in the chat.
Results
Usability testing validated a clearer, faster path from upload to live agent knowledge
5 → 2
Steps to launch
Consolidated disconnected setup into one guided flow
90%
Task completion
Users created a knowledge base without assistance in testing
35%
Faster setup
Reduced time from upload to ready-to-use agent knowledge
Reflection
What I learned
One guided flow worked better than five disconnected steps.
Thinking states and feedback made the agent feel trustworthy.
Early product and engineering input kept the design practical.