CSV orders.csv
order_id,amount,placed_at
ORD-10231,149.90,2026-...
ORD-10232,38.00,2026-...
TXL
8 rules
4 transforms
JSON orders.json
id: "ORD-10231"
total: 149.9
ordered_on: "2026-03-14"
Zero Data Retention mode available
Bring your own AI keys & models
Runs fully offline
Multitenant & strongly typed
One ecosystem

Tools that work together

Start with a one-off transform or wire up a full pipeline. Every tool speaks the same TXL - and a magic link carries your session between them.

TXMStudio

Visual schema mapping editor. Infer schemas, wire fields, and edit TXL - progressively disclosed for experts and newcomers alike.

TXEData editor

Fix data issues in place. Failed rows pause and open here via a magic link - resolve, skip, or auto-correct, then resume.

CLIPipe in/out

Generate TXL and pipe data through it. Compose multi-step pipelines and stream straight into your destination.

APIClients

Integrate transformation into any stack. REST API plus generated clients across most programming languages.

MCPFor AI agents

A hosted Model Context Protocol server at /mcp - so your AI agent calls a transform tool instead of writing one. See setup.

The language

TXL - readable mappings, machine-precise results

A transformation is plain text you can read, version, and review. Simple field-to-field maps are generated for you; complex reshapes are written by AI that reasons about intent.

Field-to-field, nested-to-field, flattening, splitting and combining.
Normalization, value formatting, date reshaping and type coercion.
Aggregates and denormalization - count distinct, unnest, collect.
Custom operations and per-level error strategies, all extensible.
orders.txl.md
# orders -> orders
```txl-rule
id = order_id
customer = customer.name
email = lower(customer.email)
total = toNumber(amount)
currency = upper(currency)
ordered_on = date.format(placed_at, "yyyy-MM-dd")
item_count = count(items)
```
In practice

One mapping, run anywhere

Write the mapping once - by example, by AI, or by hand - then run the same TXL from whichever surface fits the job.

shell - nightly feed
# vendor feed -> your schema
transmute run orders.txl.md \
-i vendor-feed.csv -o orders.json
# or stream it in a pipeline
cat feed.csv | transmute run orders.txl.md
Repeatable pipelines. The mapping is plain text - version it, review it like code, ship it in CI.
shell - customer uploads
# every upload, normalized the same way
curl -X POST "$API/mappings/$id/run" \
-F "file=@customer-upload.csv" \
-o normalized.json
Customer file onboarding. Create the mapping once in the Studio, then run every upload through the REST API.
your AI agent
# connect once (Claude Code shown)
claude mcp add --transport http \
transmute https://transmute.online/mcp
# then just ask
"Transform feed.csv with orders.txl.md"
Agent workflows. Your agent calls a typed transform tool over MCP instead of hand-writing a 200-line conversion script.
Roadmap

Coming soon

What we're shipping next.

Zero Data RetentionAvailable upon request

Stricter guarantees around what's held during a session, for teams that need it today - contact us to enable it, ahead of full self-serve rollout.

Docker imageSoon

Pull and run Transmute in your own infrastructure - for fully offline execution and enhanced privacy. Same engine and TXL mappings, self-hosted.

Local MCP installSoon

A downloadable MCP server that runs on your own machine over stdio - full custom-op support, zero data leaves your device. The hosted MCP server is available today.

FAQ

Common questions

How Transmute converts CSV, JSON and XML between schemas - and what happens to your data.

What is Transmute?

A schema-driven data transformation ecosystem. It converts CSV, JSON and XML files into any target schema using TXL - generated algorithmically for simple mappings, by AI for complex reshapes - through a visual studio, a data repair editor, a CLI and a REST API.

How do I convert CSV to JSON with a custom schema?

Paste or upload a sample of your CSV and the JSON shape you want. Transmute infers both schemas, generates the TXL mapping, and runs it across all your files. Refine it visually in the Studio, from the CLI, or via the API.

Is my data stored or used to train AI?

No. Transmute offers a Zero Data Retention mode, lets you bring your own AI provider keys and models, and can run fully offline so data never leaves your machine.

Can I automate transformations in a pipeline or from an AI agent?

Yes. The CLI pipes data in and out over stdin/stdout, composable in shell pipelines, CI jobs and AI agent workflows. The API integrates the same transformations into any application, and the hosted MCP server gives AI agents transformation tools they can call directly. TXL mappings are plain text you can version and review like code.

Can my AI agent use Transmute directly?

Yes. A hosted MCP (Model Context Protocol) server at transmute.online/mcp gives agents four tools - validate_txl, transform, generate_txl and infer_schema - so they call a typed transformation instead of writing one-off conversion scripts. Setup takes one command.

What is TXL?

TXL (Transmute Language) is a readable transformation language: field-to-field mapping, flattening nested structures, splitting and combining values, date and type conversion, normalization, aggregation, validation and custom operations - with configurable error handling.

Is Transmute free?

The hosted Studio is free - schema inference, generated TXL mappings, visual editing, the CLI and the API. Paid engagements exist for teams that want the whole job done for them. See pricing.

Can someone implement data onboarding for me?

Yes. FOR IN SOLUTIONS, the company behind Transmute, offers data onboarding as a service: ETL implementation, embedding a self-service data importer into your product, and custom pipeline development.

Stop hand-writing conversion scripts

Open the Studio, drop in a sample file, and leave with a typed, versionable mapping you can run from the CLI, the API, or your AI agent. Free.

Open the Studio

A pile of customer files and no time? FOR IN SOLUTIONS takes the whole job - data onboarding, self-service importers, custom pipelines - fixed price, quoted after a short call.