No Concept Visuals With Three Days Until the Pitch: How Agencies Build an AI Mood Board Fast

Three days until the pitch. The client wants to see three different creative directions, but all you've got is a brief and a folder of competitors' old campaigns. The art director is about to burn the night on a mood board, the copywriter is drafting concept lines, but the visual side — real, believable, brand-specific imagery — doesn't exist yet. And not a single lira has been paid for this work; if the pitch isn't won, every hour spent on it is gone. Building concept visuals for a pitch is, for that reason, the most expensive 'free' work an agency does. In an industry where every agency enters dozens of pitches a year and wins only a handful, lowering the cost of this prep work directly affects the agency's margins, independent of the win rate.

##Why Is Building Concept Visuals for a Pitch So Expensive?

The pitch stage is an odd corner of agency economics: the work isn't sold yet, there's no budget, but the client still wants to 'see' three creative directions — not as words, as images. The classic route means scouring three separate style references, a designer spending days editing stock photography, or an expensive test shoot. In an AI-assisted production line, this step flips: the brief, the brand guide (if one exists) and real product imagery go in; what comes back is three to five concept directions, each backed by ten to fifteen image variants, adapted by language and composition for different markets if needed. What used to take one to two weeks now lands on the table in one to two days — a timeline that actually fits a pitch calendar. And even if the pitch is lost, the work doesn't fully go to waste — the reference palette and brand-guide mapping built for it mean the next pitch doesn't start from zero.

##Where Art Direction Sits, and Where the Model Work Takes Over

This doesn't replace the art director's job — it pulls their decision earlier. Art direction still decides which two or three directions are worth testing, sets the reference palette and the compositional logic; the model work fills that decision in with concrete images. The first set that comes back is usually raw: lighting is inconsistent, one variant's composition doesn't fit the brand, product detail isn't sharp in another. Revision kicks in from there — the art director decides which direction to keep and which to drop, and the second round only deepens the directions that survived; this also cuts down the number of revision rounds, because the reference gets defined more precisely so the same mistake doesn't repeat in the second set. Approval always stays with a human; the model just compresses hours of manual editing into minutes.

##How It Runs, Briefly

Roughly, the line runs like this for a pitch — and because each step is tracked separately, an error in one step doesn't stall the whole process, only that step gets rerun:

• ▸The brief, brand guide, and any existing product imagery go into one package
• ▸The art director picks two or three possible directions and sets the reference palette
• ▸Each direction gets ten to fifteen image variants, with market-specific adaptations added where needed
• ▸The art director filters the variants, drops what doesn't work, carries what does into a second round
• ▸The chosen direction(s) get turned into the presentation format — slides, a mood board, a short preview video
• ▸If the pitch is won, the same direction moves straight into the production calendar — including TVN's shoot schedule if real footage is needed

##A Real Scenario: Three Days to Prep a Pitch

Say a beverage brand is looking for a new agency, the pitch is three days out, and the client wants to see two directions — one 'natural/raw', one 'energetic/colorful'. Day one: the brief and the brand's existing imagery (product shots, frames from past campaigns) go into the system; the art director sets up two reference palettes. Day two morning, the first sets land — twelve images per direction, in different compositions and settings. On the first pass, the lighting in the 'natural' direction doesn't match the brand's tone; the art director tightens the reference and asks for a second set that same afternoon. The 'energetic' direction holds up on the first pass — only three variants move to round two, the rest get dropped. Day three morning, there are eighteen strong images on the table — two directions, presentation-ready. These get turned into a slide deck and a short mood board video; that afternoon, a short video preview of both directions also gets put together before the client meeting, so the static images are joined by a sense of motion too. The classic route would likely have needed a test shoot, a studio day, and one to two weeks of design work — it simply wouldn't have fit the pitch calendar, which is why most agencies walk into pitches with a moodboard collage instead of real imagery. Here, after three days, there's real, brand-specific imagery on the table — and if the pitch is won, none of it gets thrown away; it moves straight into the production calendar.

##The Honest Limit: Where AI Stops, and Where TVN and Orni Take Over

This line doesn't solve everything, and saying so is actually healthier for the pitch itself. Once a pitch is won, the campaign's hero shot — the frame where fabric texture, fit, a real location, or a real actor has to be exactly right — still needs a real shoot; that's where TVN comes in, with traditional video production and channel consulting. Likewise, tracking which of the dozens of variants produced during a pitch went to which client, at which approval stage — especially on large corporate accounts — starts getting lost in email and WhatsApp past a certain point; Orni's (ornisoftware.com) software side solves that operational bottleneck by connecting brief-to-approval into one flow. So the AI production line solves the pitch's visual speed problem, TVN takes on the part of the won job that needs a real shoot, and Orni wires the operations of the whole process into one system — together, that's a different promise than a tool that only generates images: the promise of actually finishing the job. (Luvi Creator itself is part of this same three-house structure; on the agency side, this line runs together with TVN and Orni as a single operation.) When that split gets talked through early, at the pitch stage, nobody argues about where a given frame is supposed to come from once the won job hits the delivery calendar.

##FAQ

>Can the concept visuals made for a pitch be used directly in the won campaign?

Some of them — especially for social or digital formats where product detail isn't critical — can be used directly. Frames that need a brand-critical hero shot or print use are usually re-shot for real; the pitch imagery stays as the direction-setting reference.

>How many concept directions can be tested at once?

In practice, two or three is the sweet spot; more than that strains the art director's revision capacity and creates decision fatigue for the client in the room. Ten to fifteen variants per direction is usually enough for a solid first round.

>Does this only make sense for large agencies?

No — if anything, this is what changes whether a small team dares to pitch at all. A large agency can budget for a test shoot; a small team usually can't. The AI line closes that gap and lets a small team walk into a pitch with visuals just as strong.

>Is there a brand-safety risk in the visuals produced for a pitch?

Every variant goes through art-direction and brand-side approval before it reaches the presentation or goes live; model output never gets added to a deck directly. That step doesn't get skipped even at the pitch stage, because a prospective client's first impression of the agency is formed by exactly these images.

It doesn't matter if your pitch calendar is three days or two weeks — bring us the brief. Luvi Agency pairs the AI line that puts concept visuals on the table in days with TVN's production power and Orni's operations software; together we work out which direction needs a real shoot and which the production line can carry on its own.