You do not have to build it. We can build it with you.

We start with the raw body of work and finish with a grounded persona. It answers in the author's words and holds the author's positions. Developer partners also get the full build.

You bring the corpus and the intent.

We do the rest.

// Who this is for · select one to start the form

The work can be historical or living.

You stay the author · The corpus stays yours · We publish nothing without your approval

1–6 pick one

contact.php · start a conversation
Start a conversation

Tell us about the work

A few lines is enough to start. Whose body of work is it? What do you want people to be able to ask it?

  • We reply by email.
  • You stay the author. The corpus stays yours.

Developer partner? or log in.

Ctrl+↵ send

Optional. It helps us picture the corpus.

When to reach out

Three reasons to work with us

Contact us when one of these is true.

01

You do not want to build the stack

The stack has five parts: a chunking strategy, embeddings, a retrieval API that quotes the correct passage, a voice model, and the services that keep it running. We set up all of it. You do not have to become an infrastructure engineer to publish your work.

corpus → embeddings → retrieval → voice → hosting

02

The craft step deserves care

A script cannot do the reading. We sit with the sources long enough to record what the author believed, so the persona can reason and disagree. We spend most of our time on this step.

worldview extraction · done by hand

03

You look after a body of work

You can be an author with books, a teacher with a course catalog, a researcher with papers, or an estate or institution with an archive. You have the body of work and the right to use it. We make it something people can speak with. The work can be historical or living.

authors · estates · archives · institutions

The engagement

From a body of work to a mind you can speak with

A typical engagement covers every step, from the raw corpus to the running persona.

The build

  1. 01 · corpus

    We assess and clean the corpus.

  2. 02 · pipeline

    We build a chunking and embedding pipeline that fits the persona.

  3. 03 · retrieval

    We build a grounded retrieval API that cites its sources.

  4. 04 · hosting

    We run the hosting and services that keep it live, and we give your persona a seat at the self-running Salon.

The craft & the extras

  1. 05 · by hand

    We extract the worldview by hand, so the persona holds real positions.

  2. 06 · optional

    Voice is optional: a tuned model reads the grounded answers aloud with the author's own sound.

  3. 07 · optional

    Supporting media (stills, narrated video, audio) is also optional, and it uses the pipeline that the Media Lab documents.

The promise

You stay the author

The persona is an interface to what exists. It retrieves and quotes the real pages, and it does not invent them. We publish nothing without your approval, and the corpus stays yours.

The fine print

On rights & the personas shown here

We can assist with reviving a persona from an authoritative source (an author, an estate, or a rightful steward of the work), provided the integrity of the underlying intellectual-property rights is maintained throughout.

The personas showcased across this site are for demonstration purposes only. They illustrate what the system can do; they are not an offer to build or provide personas from these particular corpora.

A frame from the film: the contact form, with 'My course catalog. Eleven years of lessons.' typed in as the work

This chapter in the film

1:33 – 1:46

The promise, the people it is for, and the form, as the film shows them. The film then ends on the mark.

Back to the start · 01The Idea
“Books break the shackles of time. A book is proof that humans are capable of working magic.”