Product · Property, SME, Insurance · 2 min read
Your Customers Send Voice Notes. Your AI Should Understand Them.
16 June 2026 · IQPilot Team

In Malaysia a good share of WhatsApp enquiries arrive as voice notes and screenshots. With AI Response to Voice and File Analysis switched on, the assistant transcribes, reads and answers them — and can reply in a voice of its own.
The messages most bots ignore
Most chatbots handle typed text and nothing else. A customer who sends a 40-second voice note asking about the corner unit, or a screenshot of a competitor's price list, gets silence — or worse, a canned 'I did not understand that'. Those are often your most engaged prospects, and they get the poorest experience.
Switch on voice
In the AI flow settings, enable AI Response to Voice. Incoming voice notes are transcribed and answered in text like any other message. Enable Voice Reply and choose a Voice Type, and the assistant can answer with a voice note of its own — useful for customers who are driving or simply prefer listening. Keep voice replies short; a 15-second answer beats a 60-second monologue.
Let it read images and documents
With File Analysis on, the AI reads photos, PDFs and screenshots. A customer can send a floor plan and ask which unit it is, a bank slip to confirm a booking fee, or an invoice for the AI to check. For agents, the practical win is that 'is this the layout you have?' gets an accurate answer at 11pm.
- Sticker replies keep the tone friendly when customers send them
- QR and barcode analysis handles scanned collateral
- Everything is logged in the shared inbox for the agent to review
Set expectations honestly
Transcription is very good, not perfect — background noise and heavy dialect can trip it. Tell the AI to confirm key details it heard ('Just to confirm, you are looking at the 3-bedroom in Tower B?') before acting on them, and route anything financial to a human. Done that way, voice and file understanding is the difference between an assistant that handles 60% of enquiries and one that handles 90%.



