Skip to content
AI infrastructure · model customisation

A model made
to measure.

Kethra builds your company its own private, fine-tuned model, without an ML team. We take your data, tailor and evaluate a private model on your voice and standards, then deploy it. Yours alone.

You own the weights.

Prompt

Draft the opening line of a renewal email to a client whose contract lapses in 30 days.

Base modeloff the shelf

Hi there, just a quick heads up that your contract is going to expire soon! We'd love to keep you on board, so please get in touch to renew.

eval58/ 100
Kethra · tailoredyours alone

eval58/ 100+0

one prompt · two models · the fitting is the difference

02The problem

Everyone sounds the same

Most companies build on the same shelf of base models. The voice, the defaults and the blind spots arrive pre-set, and real fine-tuning has meant hiring a team you do not have.

1 shelf
the same base models

Most companies build on a handful of shared base models. The voice, the defaults and the blind spots arrive pre-set, and so does every competitor's.

6 to 9
months to hire an ML team

Real fine-tuning has meant standing up a research team, a data pipeline and an eval suite before a single private model ships. Few teams can wait.

0
weights you actually own

Rent a model and you rent its behaviour. When the shelf changes underneath you, your product changes with it, and you were never holding the thread.

03How Kethra tailors

Three steps, one owned model

You bring the cloth. We do the tailoring. You keep the garment.

  1. 01

    Bring your data and your standard

    Hand us examples of the work done well, the documents that hold your knowledge and a written note of the standard you hold. No pipeline to build, no labels to prepare.

    kethra fit --data ./corpus --standard ./voice.md
  2. 02

    We tailor and evaluate a private model

    Kethra fine-tunes a private model on your material, aligns it to your standard and scores every version against a held-out eval built from your own bar. Nothing ships until it clears.

    eval pass 47/50 · voice +31 · refusals held
  3. 03

    You deploy it, and you own it

    One command puts the model behind a private endpoint. The weights are yours, exportable, on our infrastructure or inside your own. Refinement continues as your standard moves.

    kethra deploy --private → wss://api.basisresearch.tech/v1
04Before and after

Drag the thread, watch it fit

The same prompt, answered by a base model and a Kethra-tailored one. Pull the slider to fit the private model, and watch the eval score cross the bar.

Prompt

A customer writes: my order is three days late and no one has told me anything.

Base model0% fitted

I'm so sorry for the inconvenience! Your satisfaction is our top priority. Please allow 3 to 5 business days for delivery. Let me know if there's anything else I can help with!

BaseFit the modelTailored
held-out eval56
base 56+0
Voice54
Accuracy66
Standard52

Fitted to a house voice that owns the mistake, gives a real date and never hides behind a canned window.

05What it does

The whole atelier, in one line each

From the first fitting to the deployed endpoint, every part is measured to you.

01

Fine-tuning on your data and voice

fine-tune

We fit the model to the way your company actually writes and decides, from your own corpus, not a generic instruction set. The result reads like you on your best day.

02

Alignment to your standard

alignment

You write the standard once. Kethra aligns the model to it, so tone, refusals and house rules hold under pressure, not only in the demo.

03

Private, owned weights

weights

The tailored weights belong to you. Export them, host them, keep them behind your own walls. Owned, not rented from the same shelf as everyone else.

04

Evaluation against your bar

eval

Every version is scored against a held-out eval built from your examples. You see the delta over the base model before anything is deployed.

05

One-click deployment

deploy

Ship the model to a private endpoint in a single command, with versioning, rollback and an OpenAI-compatible API your engineers already know.

06

Ongoing refinement

refine

Standards move. Each cycle re-fits and re-evaluates the model on fresh material, so the fit stays true rather than drifting out of shape.

06Ownership

You own it

A tailored model is not a subscription to someone else's shelf. The weights are yours to export, to host, to keep behind your own walls, and to carry with you if you ever leave. Kethra does the tailoring. The garment is yours.

the deed of ownership

  • The tailored weights are your property, in writing.
  • Export them at any time in a standard format.
  • Host on our infrastructure or entirely inside yours.
  • Your data trains your model only, never the shelf.
How we keep your data yours →
07From teams

Owned, and it shows

Two of the teams building on a model of their own, worldwide.

We stopped sounding like the default. Our support model answers the way our best agent would, and the eval score is the reason our board signed it off.
AD
Anaïs Delacroix
VP Customer Experience, Rivenne, Paris
The weights are ours. That single fact changed the conversation with legal, and it is why a private model finally cleared review after two years of no.
MB
Marcus Bell
Head of Engineering, Harbourline, Singapore
08Pricing

Priced to your currency

One tailored model, a house of them, or the whole atelier inside your own walls. Prices show in your local currency.

Shown in

Studio

One model, tailored end to end

$1,900/month

A single private model fitted to your data and your voice, evaluated against your bar, and deployed to an endpoint you own.

Book a fitting
  • One tailored private model
  • Fine-tune on your data and voice
  • Alignment to a written standard
  • Held-out eval suite, scored monthly
  • One-click deployment endpoint
  • You own the weights, exportable

For a team ready to replace a rented base model with one of their own.

Most chosen

Atelier

A house of models

Custom

Several tailored models across teams and tasks, a shared eval bar, and a private atelier that keeps refining as your standard moves.

Talk to the atelier
  • Multiple tailored models
  • Distillation to smaller, faster weights
  • Shared eval bar across the house
  • Ongoing refinement each cycle
  • Priority fitting and review
  • Named model tailor and SLA

For a company standardising on private models across more than one team.

Enterprise

Private, on your ground

Custom

The whole atelier inside your perimeter. Train and deploy on-prem or in your own cloud, with the weights never leaving your estate.

Contact sales
  • On-prem or your own cloud (VPC)
  • Weights never leave your estate
  • SSO, audit log, data residency
  • Bespoke eval and safety review
  • Procurement, DPA and security review
  • Dedicated solutions engineering

For regulated and security-first companies that keep everything in-house.

Prices are shown in your local currency, converted from the USD home price at a fixed rate. Sales tax is added where it applies. Fitting, data handling and deployment are included in every tier.

09Questions

Before you book a fitting

A model made to measure.
Yours alone.

Start free and tailor a sample yourself in a minute. Or book a fitting and we will build you a private model of your own.