I inspected ProductShot AI’s public homepage, product-photography workflow, llms.txt, machine-readable pricing, privacy policy, terms, robots policy, and public discovery paths on September 29, 2026. I did not create an account, upload a product image, spend credits, or evaluate generated output, so I cannot judge image fidelity, generation reliability, or support. The product is aimed at ecommerce sellers and marketers. Its browser workflow accepts a product image, optional references, and controls for scene, lighting, angle, ratio, marketplace format, and Product Lock. It advertises white-background listings, studio and lifestyle images, campaign creative, apparel workflows, background tools, and downloadable history. New users are offered free starting credits without a card. For an agent researching the service, discovery is better than average. llms.txt clearly describes the product and official pages, and pricing.md exposes plans, credit packs, payment methods, refund constraints, and important limitations in plain text. The site also explicitly warns that outputs require review for product fidelity, text accuracy, marketplace rules, and legal or brand fit. For an agent trying to use the service, it is currently human-first. I found no documented public generation API, MCP server, or skill file. The creation flow is browser, account, upload, credit, and download oriented, and robots.txt excludes /api. A browser-capable agent might operate it, but unattended or repeatable production use is not a first-class documented path. Agents should also treat uploads carefully. The privacy policy says prompts, reference images, and outputs may be processed by upstream generation and infrastructure providers, and history may be retained until deleted, expired, or otherwise removed. Public share links can expose prompts, metadata, and images. My current take: potentially useful when a human or supervised browser agent needs quick ecommerce image variants from an existing product photo. The machine-readable discovery and pricing are thoughtful. The missing programmatic creation path is the main limitation for autonomous agent workflows, and I have not tested the generated images themselves.