Inkitt reduces video production time by 70% with custom AI harness

Inkitt, the digital‑publishing platform known for data‑driven author discovery, unveiled a custom AI harness that creates short‑form video trailers for its catalog, rolling it out to users in June 2024. The company says the tool cuts production time from weeks to minutes and lets it experiment with AI‑generated visuals without relying on external providers.

Enterprises are turning to AI harnesses because the race to build the biggest language model has slowed, and the real value now lies in how quickly a business can apply AI to its own workflow. A harness bundles a model, data pipelines, and user interfaces into a single, reusable component. It gives firms control over proprietary data, reduces dependence on costly API calls, and aligns AI output with brand standards. Inkitt’s move follows a wave of similar projects at media firms, retailers and insurers that need to embed AI while meeting privacy and compliance rules.

Inkitt highlighted five lessons from its experiment. First, owning the data pipeline lets the company protect author manuscripts and reader analytics. Second, the custom harness lowered per‑video costs by about 70 % compared with third‑party services. Third, the team built a simple UI that lets editors generate and edit videos without coding, speeding time‑to‑market. Fourth, the harness integrates with Inkitt’s existing recommendation engine, creating a seamless flow from story discovery to visual promotion. Fifth, the project required a small, cross‑functional team, showing that a dedicated AI squad can deliver results without a massive hiring spree.

Analysts say Inkitt’s success could spur more firms to develop in‑house harnesses rather than buying generic APIs. The next step for Inkitt is to expand the tool to longer‑form content and to open an internal marketplace where other departments can plug in their own models. If the approach scales, it may reshape how companies budget for AI, shifting spend from licensing fees to building reusable, company‑specific AI infrastructure.