clic para entrar
← back to blog
Tech tips for business

Multimodal use cases for your operation

Multimodal use cases for your operation

For years, talking to a computer meant typing. If you wanted a system to understand a photo, an audio clip, or a scanned document, you needed a separate development effort for each format. With the arrival of GPT-4o in 2024, that boundary shifted: a single model began to process image, voice, and text natively, in one conversation. What used to be three projects can now be one.

For a small business, that change isn’t a distant technicality: it’s the difference between a customer sending you a photo of the product they want and getting a useful reply in seconds, versus your team manually transcribing the voice notes that come in over WhatsApp. “Multimodal” sounds abstract, but in day-to-day life it translates into concrete tasks that can already be solved. In this article we show you where to apply it first, no fluff.

What “multimodal” means for your business

A multimodal model is one that combines different types of information—text, images, audio—within a single task, without you having to separate them. In practice, it stops mattering what format the data arrives in: what matters is what you want to do with it.

Think about the ways your operation already receives information today:

  • Photos: receipts, IDs, photos of broken equipment, screenshots of an error.
  • Voice: customer voice notes, support calls, instructions dictated by field salespeople.
  • Documents: invoices, receipts, scanned contracts, forms filled out by hand.
  • Text: emails, WhatsApp messages, comments on social media.

All of that already comes into your business. The question isn’t whether you have the data, but how much time your team loses moving it from one place to another.

Diagram of image, voice, and text converging into a single workflow
Image, voice, and text stop being silos: a single flow understands them and turns them into action.

Use cases you can apply right now

You don’t need to reinvent your company to take advantage of this. These are realistic scenarios for a small business:

  • WhatsApp support with photos: a customer sends an image of the part they’re looking for and the system identifies it, cross-references it with your catalog, and suggests the right product.
  • Voice notes that become records: the audio from a field salesperson is transcribed, summarized, and saved as an order or report, without anyone typing it out.
  • Capturing receipts: a photo of a receipt or invoice is converted into structured data ready for your accounting, with human review where it matters.
  • Faster support: a screenshot of the problem, plus the customer’s description, helps diagnose without a back-and-forth of emails.

Multimodal doesn’t change what you do; it shortens the path between the data you already receive and the decision you need to make.

How to put it in place without overinvesting

The temptation is to want to automate everything at once. Our recommendation is the opposite: start focused and measurable. Choose a single flow, one where time is obviously lost today, and measure it before and after. That way the result shows up in numbers, not promises.

Two principles matter here. The first is that technology adapts to your operation, and not your operation to the technology: custom software preserves your data, processes, and history instead of forcing you to redo them. The second is that the person doesn’t disappear; the model prepares the work and your team validates what has consequences—an amount, a shipment, a tax detail. That’s how you make decisions with trustworthy information.

Your first multimodal flow

If you want to take the first step without risking too much, here’s a sensible path:

  • Identify a bottleneck with mixed formats: where photos, audio, or documents come in and someone processes them by hand.
  • Define a simple metric: minutes per task, capture errors, or customer response time.
  • Start with a pilot of a few weeks: a single case, with human review, to see real results.
  • Protect your data: work with providers who respect your information and explain to you where and how it’s processed.
  • Scale what works: once a flow proves its value, replicate the idea for other similar ones.

Multimodal is no longer a lab demo: it’s a tool ready to solve everyday tasks in your business. At Normandia Web we design custom software so image, voice, and text become work done, without you having to redo your operation. If you want to see where it fits in your company, let’s pick a first case together and measure what changes.

Ready to put it to work in your company?

Tell us what’s costing you time, money or control. We’ll help you figure out where to start.

Start your consultation →