Follow me on LinkedIn - AI, GA4, BigQuery
Most Voice AI systems start to fall apart the moment you interrupt them. They may lose track of context, restart their script, or ignore what you just said.

Humans, despite being imperfect, rarely do that. They pause, acknowledge the interruption, and return to where they left off.

What if your AI agent could do the same, or even better?

Meet Kate (a Voice AI Agent).

I built a Voice AI agent named Kate, designed around one simple principle: never lose the conversational thread.
Her primary task is to collect structured lead information from incoming callers:
  • Full name.
  • Email address.
  • Phone number.
  • Website URL.
  • Country.
  • Annual online revenue.
  • Consultation expectations.

Unlike a human operator, Kate can handle up to 20 simultaneous calls, meaning no customer is ever placed on hold.

She offers instant, parallel engagement, something physically impossible for human call agents.

What Makes Kate Different?

Kate isn’t just multitasking; she’s flow-aware.

That means she can recognise what stage of the conversation she’s in, remember what’s already been discussed, and gracefully recover after any detour.


Suppose a caller interrupts her mid-question, for example, asking about pricing or company hours.

In that case, she’ll briefly answer, acknowledge the query, and then seamlessly resume the original flow from the exact step she paused on.


You can interrupt her once, twice, or ten times; she’ll stay composed, responsive, and on track.

The only time she deviates from this behaviour is in two edge cases:
  • When someone tries to sell her something.
  • When she detects a robocall.

In those situations, she’s been explicitly trained to exit flow awareness and end the call gracefully.

What Is Flow Awareness Prompting?

Flow Awareness Prompting is a conversational prompting technique that teaches an AI agent to:

  1. Track context and task state, knowing which step it’s on and what information has been gathered.
  2. Handle interruptions naturally, responding briefly without derailing the main objective.
  3. Resume the flow precisely from where it was interrupted.

In other words, it provides the AI with a working memory of the current conversation, enabling it to act less like a script and more like a skilled listener who knows how to continue a thought after being interrupted.

Why does Flow Awareness matter in business?

In traditional call centres or IVR systems, once a user deviates from the script or interrupts, the voice AI agent often forgets the previous context.

The user may have to start over, repeating information they already provided, which increases drop-offs and reduces trust.


Flow awareness solves that problem by allowing the AI to remember what has been said, what remains to be done, and which step comes next, even after side questions or clarifications.

Flow awareness prompting allows:

Higher Completion Rates: Users can deviate without interrupting the process, resulting in more calls finishing successfully.

Faster Resolution: Short side answers, quick resumption, no redundant repetition.


Consistency and Compliance: The AI doesn’t skip required verification or disclaimers after interruptions.

Better UX and Trust: Users feel heard and respected; the system behaves predictably and intelligently.

Sales Efficiency: In lead capture or sales calls, the AI preserves momentum across stages (configuration → pricing → checkout) instead of resetting the flow.

How to build Flow Awareness into your Voice AI Agent?

Step 1. Design the Flow Map.