// PRICING

One flat price. Quoted up front, and capped.

You see the number before any work starts, and the cap is the ceiling. No metering, no surprise bills. Refine your data or synthesize new examples — priced simply, capped either way.

FLAT · CAPPED

Refine

Your messy data, cleaned and certified. Priced by size, complexity & data type.

$20flat — example

your number depends on the job — quoted before any work

Get my quote →
Clean ShareGPT / ChatML JSONL
PII masking, dedup & schema validation
Runs autonomously within your cap
Local DGX Spark inference included
Signed, re-verifiable certificate (JSON + PDF)
Unlimited certificate re-downloads

Synthesize / Augment

New training examples, generated and filtered. Priced by target rows.

by rows

quoted by target rows — capped, same as refine

Get my quote →
Net-new examples for the cases your model fails
AutoData Δ-filter — keeps only what a strong model solves and a weak one fails
Every row labeled synthetic, with signed provenance
Metered & capped, just like refine
Signed, re-verifiable certificate (JSON + PDF)
How the cap works

Your quote is flat and capped before any work. Accepting it pre-authorizes the agent to run autonomously within that budget — it can’t bill you past the cap. The only time it stops to ask is if finishing would exceed the cap; then you approve a top-up or keep your cap. Either way, what was quoted, spent, and the margin are itemized on your certificate.

WHAT YOU PAY
The quote

The trusted middle path

Manual cleaning
Unsupervised agent
Aegis

Questions