Reforged Labs

Reforged Labs worked as a text-to-ad generator: it turned plain text prompts plus uploaded game footage into finished video ads. A studio described the spot it wanted, and the system pulled the relevant visual assets using a mix of large language models, machine perception, and computer vision. The result was a usable creative, not a storyboard.

Speed was the headline selling point. The company promised campaign-ready results within 24 hours, compressing a process that normally stretched across days or weeks. That's the gap. For studios racing a launch window, turnaround mattered more than polish.

Each ad network wants different dimensions and formats. Reforged Labs resized finished creatives to fit those specs and exported them as campaign-ready units. That removed a tedious manual step from the workflow, since teams didn't have to reformat every ad by hand.

After a campaign went live, the models analyzed which attributes seemed to make certain ads more effective. The company framed this as filling a gap in standard reporting tools, which show what an ad did but rarely explain why. That's the missing piece. If the analysis was right, teams could act on it instead of guessing.

Ad creative automation was the point of the whole system. It fed results back into production. It adjusted visual elements toward performance-lifting traits, generated new original ads from the same asset pool, and pushed improvements out to live campaigns. Low performers were retired automatically. It's a genuinely ambitious loop, and also the hardest part to trust without your own data.

Rather than juggling separate tools per network, the platform aimed to run creation, testing, and replacement in one place, which is the kind of consolidation that lean marketing teams appreciate when they don't want to stitch together five separate services just to ship a single ad campaign. That appeal was real.