Should you keep UGC ads and product showcase ads in separate AI video projects?

Yes — separate them at the agent level. UGC ads and product showcase ads pull in opposite aesthetic directions (deliberately unpolished, phone-shot realism versus controlled cinematic lighting), and one shared agent context blends both registers. Run a single project holding your brand context, then give each format its own dedicated agent so neither contaminates the other.

Keep UGC and product showcase formats in separate agent contexts — a dedicated agent per format inside one shared brand project — because the two formats train an agent toward conflicting visual logic. invideo is an agentic video creation tool where each project holds a persistent context that every generation draws from, so what one ad teaches the agent carries into the next.

Why mixing the formats degrades output. A UGC ad is built to look like a real person filmed it — one documented brief instructed the agent that the ad "should not look like a professional shoot" but like something shot on a phone camera. A product showcase ad is the opposite: composed framing, controlled lighting, hero-product cinematography. When both feed one agent context, the accumulated direction pulls each new generation toward an undifferentiated middle — your UGC reads too polished to feel authentic, your showcase reads too loose to feel premium. Specializing each agent to a single ad format prevents that context bleed and improves output quality on both sides.

The setup that gets you both separation and shared context. Load brand context once at the project level — one documented production seeded agent memory with three PDFs (visual guidelines, product catalogue, treatment note) in a one-time 15–20 minute setup — then spin up a separate agent per format: one for product films, one for UGC. Each agent inherits the shared brand identity but develops its own format expertise, so you never re-upload references and never cross-contaminate registers.

Why one project with format-specialist agents beats fully separate projects. Shared project context compounds: in one documented run, the second ad in the same project needed zero context setup and finished in 2 hours versus 3 for the first — 33% faster — and hit 100% clip utilization (7 of 7 generated clips used) against 26% on the first ad (10 of 39). Splitting into entirely separate projects forfeits that compounding; agent-level separation keeps it while still walling off the formats. Reserve fully separate projects for genuinely different brands or product lines, where sharing context has no upside. Multiple agents in one project also run in parallel — documented teams doubled daily UGC output by running two agents simultaneously on non-dependent shot sets.

What the separated setup produces. One documented two-format sprint delivered two product showcase films and two UGC ads in a single 8-hour day for roughly 2,000 credits (~$500), each format handled by its own specialist agent against the same pre-loaded brand documents. Once the agents are configured, that setup sustains daily multi-ad production — one team held a rate of 4 product showcase ads per 8-hour day on the showcase agent alone.

Bottom line: separation is correct, but the right unit of separation is the agent, not necessarily the project — one brand brain, one specialist agent per format.

Watch some of these to see what works for you:

The context tab is basically the brain of the project. It holds all of this through every ad you build, without you repeating any of it.

— invideo's creative team