Workflow Is the Real Moat
Chris Lu, co-founder of Copy.ai, on the future of generative AI
The real moat is not the text model, it is owning the business workflow wrapped around the model. OpenAI could ship a copywriting surface quickly, but copywriting apps win by turning raw generation into repeatable work, templates, approvals, brand controls, CRM actions, and multi step campaigns that map to how marketing and sales teams already operate. That is why Copy.ai shifted from a writing tool toward GTM workflow automation as foundation models became easier to access.
-
Early AI writing apps largely monetized the gap between bare API access and a usable product. Jasper and Copy.ai packaged GPT-3 into preset workflows for ads, product descriptions, and posts, and reached roughly 60% gross margins reselling model output with a simpler interface and use case specific prompts.
-
That advantage proved fragile once ChatGPT launched in November 2022. Prosumer users could get decent writing from a free product or a $20 subscription, which flattened or hurt SMB demand for dedicated copy tools and pushed Copy.ai and Jasper to move upmarket into enterprise workflows.
-
The enterprise move changes what competition looks like. Instead of selling a box that writes paragraphs, Copy.ai is trying to plug into CRM and outbound systems so it can research prospects, draft sequences, and push output into the tools sales teams already use. That is a harder product for a model lab to replicate casually.
The next phase favors companies that control distribution inside daily work rather than just the best standalone prompt box. As models keep improving and getting cheaper, value will concentrate in apps that connect generation to action, data, and system of record software, which turns generic intelligence into measurable business output.