Hybrid AI Interface Rule
Blend conversational guidance with precise traditional controls.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 96%
Choose interface modes task by task rather than assuming every AI product should become a chat box. Conversational interaction is useful when users need to express an open-ended goal, provide ambiguous context, or ask the system to propose a plan. Traditional clicks, forms, and visual controls are often better for exact selections, configuration, confirmation, and understanding current state. Blend the two so a user can describe intent in natural language and then inspect or adjust the resulting workflow through precise controls. The mechanism reduces prompting burden, avoids brittle interpretation, and preserves the speed and predictability of established interface patterns.
Origin
Kieran Flanagan articulated this rule after discussing a failed conversational onboarding experiment and recommended Gummy as an example of a blended interface.
Core principles
- 01Conversational UI is not universally superior.
- 02Natural language works well for intent and ambiguity.
- 03Traditional controls work well for precision and visible state.
- 04The best interface can change by task stage.
- 05Users should not be forced to prompt for every action.
How to run it
- 1
Decompose the workflow
List each user action and identify whether it is exploratory, ambiguous, exact, repetitive, or consequential.
Pro tip Observe where users naturally describe versus select.
Watch out Do not classify the entire product with one interface assumption.
- 2
Assign conversational tasks
Use natural language for goals, questions, broad instructions, and iterative clarification.
Pro tip Provide examples that teach users what they can ask.
Watch out An empty chat box can recreate the blank-page problem.
- 3
Assign visual controls
Use buttons, forms, menus, and canvases for exact configuration, comparison, and confirmation.
Pro tip Keep important state continuously visible.
Watch out Do not hide consequential settings inside conversation history.
- 4
Connect both modes
Let conversational actions update visible controls and let control changes inform the conversation.
Pro tip Use one shared underlying state model.
Watch out Conflicting state between modes destroys confidence.
- 5
Test task completion
Measure whether ordinary users can complete the full workflow faster and with fewer errors than in either single-mode design.
Pro tip Test novices and experienced users separately.
Watch out Novelty preferences during a demo may not predict repeated use.
In the wild
A user tells an AI agent, “Send qualified website leads to sales.” The system proposes a workflow conversationally, then displays trigger, qualification, destination, and notification settings as editable visual blocks.
→ The user gains fast natural-language setup while retaining precise control and visible system state.
Common mistakes
Chat for every adjustment
Prompting is inefficient when users need to make small exact changes or compare current settings.
Disconnected interaction modes
Conversation and visual controls must update the same state or users cannot know which configuration is authoritative.
Is it for you?
Best for
It is best for AI workflow builders, onboarding systems, and applications mixing open-ended goals with precise configuration.
Not ideal for
It is not ideal for simple products where one interaction mode already handles every task efficiently.
From the transcript
“I actually don't believe conversational UI is the best uh UX experience for all AI apps.”
“I think in some cases it's easier just to do traditional click click click UX.”
“I think he's done an incredible job of blending both a traditional UI interface together and like a conversational UI.”
From the episode
The AI Workflow That Lets 50 People Do the Work of 500 ($2B Founder Reveals)