Latest project

Transmute: schema-driven data reformatting

Whenever two teams or systems need to cooperate, the data never arrives in the shape the other side needs. Transmute turns an example file and a target schema into a working transformation.

The problem

Whenever two teams or systems need to cooperate - share data, consolidate it, or hand it off in a new format - the files almost never arrive in the shape the other side needs. A customer's CSV export, a partner's XML feed, a legacy system's nested JSON - someone has to look at an example file, work out field by field what it means, and repair or remap it by hand before it's usable for review, comparison, or import into whatever comes next. It's usually a one-off job, done once for this exchange and then done again from scratch the next time it comes up. That recurring, manual reformatting problem - not any single client project - is what Transmute solves.

The digital solution

Instead of treating every reformatting job as a one-off, we built a tool for it. Transmute is a schema-driven data transformation engine: point it at a source (CSV, XML, JSON, Excel) and a target schema, and it generates a typed transformation - algorithmically for simple field-to-field mapping, AI-assisted for anything that needs reasoning about intent. A human-in-the-loop editor (TXE) lets someone fix the records that don't fit cleanly instead of blocking the whole batch. It runs as a free hosted tool, a CLI and API for automated pipelines, or fully offline with Zero Data Retention for teams that can't send data anywhere.

The business result

Transmute is our philosophy made tangible. Anyone can try it free and see, in minutes, how we think about data work - schemas first, humans in the loop for the records that don't fit, privacy by default. That's the credibility a proposal alone can't provide, and the same thinking carries into two service offers:

  • Data onboarding as a service - send us the messy files, we return clean data in your schema, fixed-price and fast.
  • Embed a data importer in your product - SaaS teams get a self-service importer under their own brand, so their customers can onboard themselves instead of filing support tickets.

Every engagement teaches us new connectors and TXL transformation patterns, which make the free tool better and the next conversation easier to have.

Try Transmute free Talk about data onboarding Back to projects

Transmute schema mapping preview
Tech highlights
  • CSV, XML, JSON, and Excel input and output
  • Algorithmic mapping for simple cases, AI-assisted mapping for complex ones
  • Human-in-the-loop repair (TXE) instead of fail-fast batch rejection
  • CLI and API designed for shell pipelines and AI agents
  • Zero Data Retention and offline/self-hosted options

Live at transmute.online.