The archive · Product Ideas · Product decision · 2023
Pixian.AI takes on subscription background removal by publishing its benchmark losses
An API-first rival to remove.bg publishes head-to-head quality scores — including its losses — and lets customers benchmark their own images free.
Pixian.AI
What it had to solve
Online background removal was dominated by an incumbent such as remove.bg that customers reached through subscriptions or credits, while Pixian.AI argued that AI image processing was commoditizing fast and chose to accelerate that reality — positioning itself as infrastructure, 'more S3 and less Adobe', rather than another software vendor.
How it works
Pixian.AI launched an AI image-background-removal service at a moment when the category was consolidating around an established leader, and it chose to treat AI commoditization as the business model rather than the threat. Its stated philosophy: see itself less as a tech startup and more as an outsourced MLOps extension, with a goal of being 'more S3 and less Adobe' for state-of-the-art image processing.
The offer is built on radical verifiability. The site publishes head-to-head benchmark results against 'the best known competing service' across five image classes — people in studio shots, people in other backgrounds, objects, artworks, scans and logos — and shows the scores even where Pixian loses, such as 88.1% and 94.8% against a competitor's 100%. A free comparison tool then lets any customer run the same test on their own images, and the API is a drop-in match for remove.bg, so migration is a matter of changing the endpoint and API key.
The Show HN on 24 May 2023 drew 383 points and 115 comments. Commenters benchmarked Pixian against open-source U-2-Net and Meta's Segment Anything, compared it with Apple's built-in background removal, and several said the combination of speed, accuracy and per-image pricing made them paying customers. The launch thread also became a feedback session, with developers fixing issues users flagged in the comparison charts.
Why it lands
- Publishing benchmark losses is the counterintuitive move: admitting 88.1% and 94.8% against the competitor's 100% makes the categories where Pixian wins believable.
- A free tool that tests the customer's own images moves the argument from vendor claims to the prospect's data — the comparison that actually decides a purchase.
- Drop-in API compatibility with remove.bg removes switching cost, attacking the incumbent's lock-in instead of only its price.
- Basing the business on commoditization — 'their margin is our opportunity' — converts the industry's downward pricing pressure into the company's own pitch.
What it did
The Show HN on 24 May 2023 drew 383 points and 115 comments. Testers compared the service with open-source U-2-Net, Meta's Segment Anything and Apple's built-in tools, and several commenters said the speed, accuracy and price made them paying customers.
What you can take
When you cannot outspend the category leader, remove the customer's reason to distrust you: publish your losses with your wins, let prospects test on their own images, and make switching nearly free.
Since then
The transparent-comparison tooling stayed at the centre of the product: the site continues to publish per-class comparison reports, offers the free compare-your-own-images tool, and still advertises the remove.bg migration guide. Its FAQ explains the same philosophy — the team optimizes model, architecture and overhead for quality per dollar rather than building a subscription business around lock-in — and describes data retention that keeps uploaded images only briefly, with API images never persisted.
Sources
spotted an error? The archive wants to know.
Your turn
You just read one. Describe the brief you are staring at, and see who has been given the same problem.
Free account · 3 free questions · no card