I Tested AI POP Campaign Design in Bellto: Floor Display, Wobbler and Shelf Strip

I wanted to see one thing end to end. Could a brand team sit down with nothing but a product, a store count and a rough idea, and come out the other side with concepts for every piece of a launch and a list of manufacturers to ask for a price?

So I ran a launch through Bellto, an AI agent for POP campaigns that takes a campaign from the brief to shop drawings, and recorded it. The launch was a simulation. I imagined Nutella bringing its Nutella Peanut flavor to Spain and needing point-of-purchase materials for it. Nutella and Ferrero had nothing to do with this and aren't clients.

I've spent close to four years in point-of-purchase, and Bellto is what we're building now. The video is in Spanish. This is the written record of what happened in it.

What did I give Bellto to start?

I created a campaign and called it the Nutella Peanut launch in Spain. The screen splits in two. On the left is a chat with the agent. On the right is the campaign view, with a summary, the briefing and other tabs.

My first message was the idea in a few lines. Take point-of-purchase materials to 500 stores. A floor display, semi-permanent or temporary, plus materials for the shelf.

I knew it would ask a lot of questions, because that's how it gets to a good proposal, so I got ahead of some of them. I attached a photo of the product, the standard Nutella jar, and an image with some of the graphics. I said the campaign was mainly for the 400 g jar and gave its size, 160 mm tall and about 105 mm across. Those measurements matter, because the facings get worked out from them.

Bellto filled part of the briefing sheet straight away from that first message and came back with questions about the rest.

What did it ask before proposing anything?

The delivery date first. I told it the date didn't matter for this test. Then which shelf materials I had in mind, and I said a wobbler and a shelf strip. Then whether the floor display would arrive assembled and loaded with product. I said no, it gets set up in store, without product.

I could tell what was coming next and answered before it asked. One unit of each material per store. Sure enough, its next question was an estimate per store for each material. It also asked about material to narrow the scope, and I said folded cardboard. There are cardboard displays that are semi or fully self-assembling, but that's a decision for later, once the design goes out to quote.

Last round was style, brand guidelines and number of shelves. For the floor display I asked for four shelves and four facings, and for the brand look to follow the product's design and the image I'd attached.

Then it asked whether I wanted to close the brief. I accepted the proposal, saved it, and told it to go ahead with the pieces.

The thing I'd do differently is bring the brief more complete from the start and let the agent ask only about what's left. I built it up in the chat on purpose so the questions would show on camera. If you have a launch with a product, a store count and a rough mix, the Bellto waitlist is where that brief goes.

What came back, piece by piece?

Bellto proposed three pieces: the floor display, a Nutella wobbler and the shelf strip. Creating them put three buttons at the top of the view, one per piece, and generation started on its own, two versions per piece. It began with the wobbler, worked two images at a time, then moved on to the floor display and the shelf strip. All six took about five minutes.

The floor display. The first version was the simpler of the two. Its graphics were curious, and that one was partly on me. It had taken the graphic image I uploaded and used it literally, because I'd told it to use that image as a reference. The second version I liked more. Its header is shaped like the jar, a dummy of the product on top of the unit. Four shelves, four facings, as asked. You can see a bit more of the cardboard inside it, but it's a good option.

Looking at it raised a point I hadn't put in the brief, which is the total product load. A 400 g jar adds up fast. Four facings three deep is twelve jars, close to 5 kg on each shelf, so the shelves have to be built for that. I never told Bellto how deep the shelves were loaded, and it's the kind of figure that belongs in the brief.

The wobbler. A regular wobbler, and the design was good. The second version followed the same style with a shape that was a little stranger.

The shelf strip. This was the weakest of the three. Both versions came out as a price-tag strip rather than a shelf strip. The design itself had a nice idea in it, a classic Spanish checked tablecloth that brings to mind an afternoon snack of bread and Nutella, but the format was wrong.

What did I have to correct?

The floor display and the wobbler were good enough to go forward as they were. The shelf strip needed a new version from scratch.

I asked for it in the chat, in plain words. Neither version works. Those are price-tag strips, not a shelf strip. I want a shelf strip that sits on the shelf, with the Nutella logo standing out above it.

The new version came back built from zero. It did what I asked with the logo, which stands out. The overall design was still a bit unusual.

Two lessons from this part. A reference image gets used the way you describe it, so if you upload graphics for inspiration, say that, or they may end up printed on the piece. And the piece types close to each other need a clear description, because a price strip and a shelf strip both live on the shelf edge. When a version misses, the fix is a sentence in the chat about what the piece should be. Before Bellto takes over, the concept image for a single display is what AI POP Displays does, and the same habit of naming the format pays off there too.

If your next launch includes shelf pieces like these, put it on the Bellto list and we'll run it with you.

How did the manufacturer step work?

Once I had the versions I wanted, the next step was asking for a price. I opened the quotes tab and let Bellto recommend manufacturers.

The recommendation works from the units and the materials of the campaign. Bellto has a database of manufacturers, Spanish ones in this case, and it suggested three: one in Valencia, one in Barcelona and one in Madrid. For each one it showed details such as their certifications and how they work this material.

From there there's a button to request a quote from those manufacturers. The quote comes from them. The same concept goes to several manufacturers from one place, instead of being re-briefed to each.

For more on why one shared concept matters at the quoting stage, I went through it in from AI render to quotable display.

What do I take from this run?

The order of the conversation is the useful part. Bellto doesn't draw anything until the brief is closed, and every question it asked was one a manufacturer would have asked me later, by email, after the first concept was already wrong. The mix, the per-store quantities and the material were decided before any image existed. How that planning step works in more depth is in planning a POP campaign with an AI agent.

Two versions per piece was enough to choose a floor display and a wobbler, and to see that the shelf strip had the wrong format in both.

What the run asks of the person driving it is precision about the piece and about the load. Say what each reference image is for, name the shelf piece clearly, and put how many products go on each shelf in the brief.

Where is Bellto now?

Bellto is an AI agent for brands and agencies that takes a POP campaign from brief to shop drawings. Beyond what this video shows, it designs every piece with you at real scale in parametric 3D, with proportions derived from the product and the facings, and a dimension change recomputes the whole piece and re-runs the manufacturing checks in seconds. Every approved piece comes out as a 3D model, a part-by-part cutlist, dimensioned drawings, a STEP of the assembly and a cut DXF per part, a package a workshop can quote without redrawing. The fuller reference is in what is Bellto.

We're building it now. The waitlist is open to any brand, and we're contacting the first ones soon to run the first real campaigns end to end. It opens by invitation, in small groups. Pricing goes first to the people on the list, and there's no card. POP manufacturers can sign up too.

If you have a launch like this one coming, with a product, a store count and a rough idea of the pieces, join the Bellto waitlist and tell us about it.