How to Automate WhatsApp Orders with AI: No Forms Needed

11/6/26

How to Automate WhatsApp Orders with AI: Audio, Photo and URL with No Forms

Updated: June 9, 2026

WhatsApp AI automating orders with audio photo and URL in a conversational flow with no forms

TL;DR: Automating WhatsApp orders with AI means you can receive the request as a voice note, photo or text, extract the order data without any form and confirm with the customer in under 30 seconds. Tubot runs on the WhatsApp Business API, manages that conversation end to end and follows up if the customer stops replying before confirming.

A distributor gets 40 messages before 10:00. Some are voice notes from customers who are driving, others are photos of handwritten lists, and one is free-form text with the usual order but no exact references. If someone on the team has to open each message, listen, transcribe and reply, the first order isn't confirmed until mid-morning. A WhatsApp AI resolves that flow in seconds, without asking the customer to switch channels or fill in any form.

Why the WhatsApp order is the real format, not the ideal one

Customers order the way they talk and wherever they are, not the way the company's system would prefer. For any business that takes recurring orders, WhatsApp is already the first point of contact, and the order arrives in whatever format the customer chooses: a voice note while driving, a photo of a delivery note or unstructured free text. If the process requires turning that into a form before it can be processed, the friction falls on the team, not the customer.

Spain's regulator CNMC reported that e-commerce in Spain invoiced €24.558 billion in the third quarter of 2024, up 12.6% year on year. That figure covers digital channels, but it excludes the volume of orders handled over WhatsApp without going through any e-commerce platform: distribution, hospitality, services and SMBs selling to regular customers outside a digital catalog.

According to the ONTSI report "La Sociedad Digital en España 2024", more than 90% of Internet users in Spain use WhatsApp daily. The channel where the customer already is, is the channel where the order arrives first. If there is no system to handle it on the spot, it gets processed by a person who has other things to do.

What can a WhatsApp AI automate in an order?

A WhatsApp AI can automate everything that doesn't require commercial judgment: receiving the request in any format, extracting the order data, confirming availability, applying each customer's terms and sending the confirmation. What escalates to a person is the exception, not the rule. For any business with recurring orders, that split reduces the daily operational load without changing the customer experience.

VOICE: the customer sends a voice note describing the order. Tubot transcribes it, detects the language and extracts product, quantity and terms. If the reply contains no URL, it can answer with a synthetic voice note.
VISION: the customer sends a photo of a handwritten list, a delivery note or a product label. Tubot analyzes the image within the conversation and builds the order.
WEB: the customer pastes the URL of a product page or an online catalog. Tubot reads the content and uses the exact reference without the customer having to copy it.
FOLLOW UP: if the customer sends the order but stops replying before confirming, Tubot sends a voice note after 4 hours with a specific observation about the case to close the conversation.

This multimodal pattern maps directly onto the order flows described in the guide to WhatsApp AI for ecommerce and the analysis of automating WhatsApp orders for SMBs, where the customer's first message defines the channel and format for the whole conversation.

Steps to automate WhatsApp orders with Tubot

Automating WhatsApp orders doesn't require changing the channel you use with customers or making them install a new app. It requires the AI to understand the order as it arrives, log it correctly and confirm with the customer in the same thread. That process has five operational steps that don't depend on the industry.

Connect your number to the WhatsApp Business API. Meta's official documentation on the WhatsApp Business Platform establishes that the API is the only channel that allows automating messages at scale with proper handling of the 24-hour window and approved templates for contacts outside that window.
Load your catalog, references and per-customer terms. Tubot needs to know products, prices, agreed terms and active references to build the order correctly, without inventing data that doesn't exist.
Define what the AI closes and what escalates to the team. Standard orders: Tubot handles them on its own. Orders with a special discount, an out-of-range quantity or an out-of-stock product: they escalate with the context already prepared for the team.
Activate VOICE, VISION and WEB based on the customer profile. If customers send voice notes, enable transcription. If they send photos, enable image analysis. If they paste external URLs, enable web content reading.
Set up follow-up. If the customer sends the order but doesn't confirm, Tubot steps in with a specific observation about the case within the 24-hour window, with no generic messages.

The order flow in practice

Illustrative example:

Lead (22 s voice note, 08:12): Hi, this is Carlos from Distribuciones Montero. I need last week's order, but add two more boxes of reference A-45 and drop the oil because we still have stock.

Tubot: Got it, Carlos. I'm checking your last order. To confirm: same order as last week, adding two A-45 boxes and removing the 5 L olive oil. That comes to 18 lines in total. Want me to confirm stock before closing it?

Lead: Yes, tell me if there's A-45.

Tubot: Stock confirmed: 12 boxes available in the warehouse. With the two extra ones the order closes with no shortage. Shall I confirm for delivery on Thursday?

Lead: Yes, Thursday works.

Tubot: Order logged for delivery on Thursday. The logistics team is receiving it now with the full summary. If you need to change anything before Wednesday, reply here.

Lead: Perfect. Hey, do you have the new summer catalog? I saw something on your website. [pasted URL]

Tubot: I checked the link. The summer catalog has 34 new references, twelve of them compatible with the lines you already order regularly. Want me to put together a selection of the ones that fit your order profile?

Lead: Yes, send it over.

Tubot: Selection ready: here are the 12 references with price and availability. If you want to add any of them to your next order, tell me here and I'll note it down.

Total time from the first voice note to the logged order: four minutes. The team walks in at 09:00 with the orders already confirmed and a summary of each one, without having listened to a single audio.

Manual orders versus orders with Tubot

This table compares manual handling with the automated flow for the most common situations in businesses with recurring WhatsApp orders. Use it to identify where the operational load concentrates and what changes when the AI handles the first contact.

Situation	Without AI	With Tubot
Voice note with an order at 08:00	Waits for someone to listen and transcribe it	Transcription and confirmation in minutes
Photo of a handwritten list	Someone deciphers it and types it in line by line	VISION extracts the lines and builds the order
Recurring standard order	The customer repeats it, someone copies the previous order	The AI retrieves the last order and confirms changes
Customer doesn't confirm after the quote	Manual follow-up, or none at all	Voice note after 4 hours with a specific order detail
URL of a catalog or external product page	Someone opens the link and looks up the reference	WEB reads the page and uses the exact reference
Order outside the team's hours	Waits until the next day	Processed before the team opens

Which metrics should you track in the first month?

The useful metrics in an automated order flow aren't the number of messages sent or the open rate. They are the time between the customer's first message and the order confirmation, the percentage of orders processed without human intervention, and the volume of orders that arrive outside the team's hours and get processed correctly. Without those three figures, you can't tell whether the flow works or where to improve it.

Confirmation time: from the customer's first message to the order confirmation. With manual handling it can take hours; with Tubot, minutes.
Orders without escalation: percentage of orders Tubot closes without team intervention. It shows how much of your volume is automatable with the catalog and terms you've loaded.
After-hours orders: conversations that arrive when the team is off and are processed correctly before they open. It's the most direct indicator of the automation's value for businesses whose customers order before 09:00 or after 18:00.

For data processing, it's worth adjusting the privacy notice on your WhatsApp channel as indicated by the AEPD on the right to information. Order flows with customers who already have a commercial relationship usually rely on contract performance as the legal basis. To review your specific case, you can start at Tubot or write via Tubot's contact page.

Frequently asked questions

Can a WhatsApp AI handle orders with different terms per customer?

Yes. Tubot loads each customer's terms: agreed prices, discounts, usual order frequency and preferred references. The order is built with those terms, not a generic price list. If the order exceeds an amount that requires approval, it escalates to the team with the context already prepared.

What happens if the customer's voice note describes several products at once?

Transcription processes the full audio, extracts every order line and builds a summary before asking for confirmation. If the audio is ambiguous on any reference, the AI asks only about that specific point, without making the customer repeat the whole request from the start.

How does Tubot handle Meta's 24-hour window for recurring orders?

As long as the customer writes at least once every 24 hours, the service window stays open and Tubot replies freely. If the customer doesn't reply within that window, Tubot can re-engage with a Meta-approved template to follow up on the pending order. That logic is handled by the product, with no violations and no risk of the number being blocked.

Does it work for B2B orders with multiple lines and references?

Yes. The multimodal flow is especially useful in B2B, where orders are recurring, high-volume and arrive in mixed formats: a voice note, a photo of a delivery note and free text in the same thread. Tubot keeps context across turns, so the customer can add lines without repeating what's already confirmed.

Can a confirmed order be integrated with an external management system?

Tubot runs on the WhatsApp Business API and can integrate with management systems, ERPs or CRMs via a product connection. The confirmation is logged in your system without the team having to copy data manually from the chat history.