Where the three weeks go
Bought weeks ago, still not sellable
The product sits in the ERP for three weeks before it can be published anywhere. Onboarding is the bottleneck, not the buying.
Done by hand, every time
- The supplier sends a spreadsheet, and someone normalises the units and taxonomy
- Missing attributes filled in, copy written for each market
- Brand and compliance checked one product at a time
- Published to the website, the app and the feed, each a different system
Every launch is a project
- Every new market is a project rather than a launch
- Campaign deadlines mean cut corners or overtime
- Specialists run the same check hundreds of times a week
- The page, the feed and the store screen each say something different
You need the launch backlog cleared before peak season without temporary staff. The constraint is not the buying decision or the supplier's data. It is the 30 manual minutes between the file arriving and the product going live.
From supplier file to live
Supplier file in, every market out
Speedtrain is an AI data orchestration platform that takes a supplier file and turns it into publish-ready product data for every market and channel in minutes.
- 1
Intake
Supplier Intake Gate normalises units and taxonomy from the messy feed, and records the gaps rather than silently filling them.
- 2
Enrich
Missing attributes are completed by per-attribute rules that are versioned and testable. Source Trail records where each value came from and how sure the system is, so guesses are countable.
- 3
Write and check
Content Amplify generates titles, SEO descriptions, translations and campaign variants from one product truth. Govern runs the compliance checks and flags what needs a human decision.
- 4
Publish
Core projects one representation into each channel dialect, and change detection reworks only what moved. Publish Ledger records what went live where, when, and what was rejected.
Your PIM, DAM and automation tools keep their jobs. Speedtrain reads from the sources you already run and writes back into them, so there is nothing to migrate and no new system to learn up front.
What XL-BYG measured
XL-BYG went from 30 minutes to 2
A Danish builders' merchant running three websites, thousands of products and constant assortment change.
Onboarding per product
30 min2 min
Quality checks included.
Approved automatically
99%
Reflects XL-BYG's own workflow and data quality. The general expectation is about 60 per cent approved first pass.
Time on site
+40%
Across four categories.
Catalogue processed
45%
In three months.
Read the 99 per cent with its caveat. It reflects XL-BYG's workflow and data quality, not a number to expect on day one. The general expectation is 100 products in under 10 minutes, with about 60 per cent approved on the first pass.
Our product specialists validate rather than write. That uses their expertise better.Kim Portz Knudsen, Head of Webshop, XL-BYG
What your specialists keep
The typing goes, the judgement stays
Product specialists validate rather than write, and nothing publishes until it has passed the checks.
Generation is grounded in your own source data. Your rules on tone, brand, legal language and forbidden words are inputs to Speedtrain rather than afterthoughts, and Govern flags gaps instead of inventing content to fill them.
The 80 per cent that is auto-approved earned it by passing the rules you set, not because nobody looked. What remains is approving the exceptions and validating what the system flagged, which is the work that actually uses a specialist's expertise.
The team stays in control of what publishes, and stays in the systems they already know. Speedtrain writes into the PIM, DAM and content tools you already run.
You either absorb rising volume without more hours, or you miss the campaign because content was late. Speedtrain removes the bottleneck.
