Work · White label

White-label case studies: our AI products under a partner's brand

Ultra Lab is an AI product company. We turn our own AI products into white-label editions that carry a partner's brand, and the partner sells them to its existing customers. We own the product, the engine, the deployment and the ongoing releases. On product screens, reports and pages, the partner's customers see the partner's brand.

The two cases below come from one partner: an overseas domain and cloud services provider.

Case 1

Social automation platform, white label

Background: The provider put its own brand on our social automation platform and sells it to its existing small business customers.

What the white-label edition includes

A multi-platform social automation platform launched under the partner's brand: scheduled publishing, AI copywriting, an auto-refilling content library and auto-replies for Threads; publishing and scheduling for Instagram and Facebook Pages; in-app video generation that publishes as Threads videos and Facebook Reels.
A cross-platform distribution hub: a post published on any one platform can be distributed to the other two, six directions in total. Automated distribution goes into an approval queue, so every post is confirmed by the user before it goes out.
An AI social assistant that finds prospects (currently open via an allowlist): based on the user's target audience and working hours, it finds relevant posts, drafts replies, and reports leads and engagement.
Trials, plan upgrades and self-service cancellation, wired directly into the partner's billing platform.
A partner console: member management, customer overview, plans and renewal status.

Calendar page of the social automation platform: a week of scheduled posts in time slots, with upcoming posts, platform and queue status listed below, using demo accounts — Calendar: a week of scheduled posts and the status of each one. Taken from the test environment of our own product; the accounts and content are demo data. In the white-label edition the interface carries the partner's brand. The interface is shown in Chinese.

Content library page: AI batch generation, a review-before-publishing switch and a list of queued posts with their times, using demo accounts — Content library: AI batch generation and nightly auto-refill, with an option to send AI-written posts to drafts for review before they join the queue. Same test environment, demo data.

There is no publishable demo screen of the cross-platform distribution approval queue yet, so here is how it works: posts created by automated distribution go into a queue first, and the user confirms each one before it is published to the other two platforms.

Technical highlights

Brand by hostname: the brand is resolved from the hostname, and the UI, colours, copy, SEO tags and sitemap all follow. The white-label edition runs on its own release line and takes features from the main product line in batches. The pages and templates we control (the UI, server-side page templates, the page shell served to search engines, and robots, llms.txt and sitemaps) have been through a layered brand string audit, and no trace of our own brand is left in the white-label edition.
Business data in the partner's chosen region: customers' business data (posts, account settings, subscription status) lives in a separate database in the region the partner chose, stored apart from other brands' business data. The main application and the scheduling engine are deployed on the partner's servers; some components, such as video rendering, run on Ultra Lab's side.
Staging first, then production: the same image is accepted on staging first; before production a rollback tag is set, so a fixed set of steps restores the previous version if anything goes wrong. After releases, automated regression tests run, and major features are accepted with real signed-in screenshots.

Results

Live under the partner's brand within 2 weeks of signing, followed by about 5 months of continuous releases.
20+ features deployed to production; the AI social assistant and Instagram/Facebook publishing are currently open via an allowlist.
8 Meta platform permissions approved (2 for Threads search and profile data; 6 for Instagram and Facebook Page publishing, comments and Page access).
7 external API integrations: Threads, Instagram, Facebook Pages, AI model, the partner's billing platform, authentication and database, cloud storage.

Case 2

AI security scanner, white label

Background: The same partner put its own brand on our website and AI security check, adding one more product alongside domains and hosting.

What the white-label edition includes

A website AI security check deployed under the partner's brand: enter a URL and get two scores, website security and AI visibility. The report leads with the conclusion and the business impact, then the technical details.
A live AI attack demo: the same chatbot is hit by attacks adapted from real incidents, once with protection on and once with it off, so visitors see the difference on the spot.
Leads and formal reports: after a visitor leaves their details, a formal audit report is emailed automatically and a lead notification goes out.
A 25-vector in-depth assessment kit and report generator, so the partner's team can run paid assessments on its own.

AI attack demo with protection off: a demo customer service bot is fooled by someone posing as a manager and lists customer names and phone numbers, all fictional — Protection off: given the request from someone posing as a manager (shown in the next image), the demo bot hands over the customer list and phone numbers.

AI attack demo with protection on: the same attack is blocked and labelled as data theft, and the bot declines to share customer data — Protection on: the same attack is blocked and its type is labelled. Both screens are from Ultra Lab's own demo page; the white-label edition carries the partner's brand. The demo bot, business name, customer list and phone numbers are all fictional. Colours adjusted and cropped (the last paragraph of the bot's reply is omitted). The demo is shown in Chinese.

Technical highlights

Two scanning engines in parallel: HTTP security header checks plus 9 AI readability checks (llms.txt, robots rules for AI crawlers, whether AI crawlers can actually reach the site, sitemap, structured data, meta description, title, social previews, canonical).
Abuse protection for public endpoints: server-side blocking of private network addresses (SSRF protection) and fetch size limits.
Dual deployment: the same code runs as cloud serverless functions or as a container on the partner's own servers.

Results

From zero to deployed in under 2 weeks (one cloud deployment, one on the partner's servers).
About 12 features, in two languages (English and Chinese).
25 attack vectors and 9 AI readability checks.
2 external API integrations: AI model, transactional email.

Already have customers and want another AI product line?

How a white-label partnership runs, how brands are kept separate and who owns the customer relationship are covered on our partner page (in Chinese). In English, write to contact@ultralab.tw.