This page documents PUBXX 1.0, the AI engine owned by SEW Capital, and the format models built on it — PUBXX 1.0 Prose, Narrate, Drama, Press, and Lingua. Each model is a single-prompt system: give it a source manuscript and an instruction, and it starts and finishes that format's draft in one pass, with no separate steps to chain together. None of this is tied to a single platform or token; it is the infrastructure underneath SEW Capital's publishing output generally. It is an operational reference, not an offer or investment content.
PUBXX 1.0 is not one single tool — it is an engine with a set of named models built on top of it, each trained to take a source manuscript and a single prompt and return a finished draft in one specific format. This page describes the engine and its models on their own terms, independent of any platform or token that may draw on their output.
Every model described further down this page — Prose, Narrate, Drama, Press, and Lingua — runs on the same underlying engine. PUBXX 1.0 is not itself a finished-format tool; it is the shared drafting and consistency layer that each named model specialises for its own format.
PUBXX 1.0 is SEW Capital's proprietary AI engine. It handles the parts of production that are skilled but repetitive and that a format model would otherwise have to solve from nothing: structural and language consistency, formatting logic, and metadata handling. The named models are built on this shared layer rather than being separate, unrelated tools.
It produces top-specification, error-free, to-date drafts at a consistency that hand assembly across separate tools cannot sustain, and it does so at a fraction of the cost of producing each format from scratch with a dedicated, unconnected workflow.
PUBXX 1.0 is what makes five named models viable instead of five unrelated production systems. One engine, consistently maintained, underneath all of them.
Each model takes the same source manuscript and the same underlying engine, and applies the specific structure, pacing and technical conventions that its format requires. There is no separate chain of steps to manage: one prompt in, one finished draft out. They are described here independently of any platform that might use their output.
Converts a source manuscript into a structured digital edition — reflowable layout, internal navigation, and the metadata a digital edition needs to be catalogued correctly — from a single prompt.
Of the five models, Prose is generally the fastest single-prompt draft to return, since it works with text structure directly rather than needing a separate audio or print layer.
Takes a source manuscript and, from a single prompt, drafts a narrated audio edition — voice performance, pacing and chaptering — then applies audio mastering so the draft is technically finished.
Narrate is trained specifically on narrated long-form audio, distinct from the dramatized, multi-voice work the Drama model performs.
From a single prompt, drafts a dramatized, multi-voice adaptation structured for serialized, episode-by-episode release, rather than a single straight narration of the source text.
Because it is built for serialization, Drama's output is structured in episodes from the start, rather than being a single file split up afterwards.
From a single prompt, handles the structural difference between a single standalone work and a periodical: issue-based layout, recurring sections, and the identifier conventions periodicals use instead of a book's.
Press is trained on periodical conventions specifically, so an issue is drafted with the right section and masthead structure rather than being treated as a single long document.
Each model returns a complete draft from a single prompt. A separate, optional editorial review pass can be added afterwards — see usage & pricing for how that is priced — but it is not a required step to get a finished draft out of any model.
Translation is handled by a named model of its own, PUBXX 1.0 Lingua, rather than being a feature bolted onto each format model — because the same translation pass needs to apply consistently whether the source becomes a Prose, Narrate, Drama or Press draft.
From a single prompt naming the target language, Lingua returns a complete translated draft of a source text. It is built on the same PUBXX 1.0 engine as the four format models, which is what keeps terminology and structure consistent across formats and across languages for the same underlying work.
Because it sits across the format models rather than inside any one of them, a single Lingua pass can feed Prose, Narrate, Drama or Press for the same source work, rather than each format needing its own separate translation step.
Every model — Prose, Narrate, Drama, Press and Lingua — follows the same general pattern: a source manuscript and a single prompt go in, PUBXX 1.0 drafts, and the named model shapes that draft to its own format. There is no multi-step chain to manage, and no required human pass before a draft is returned. This describes the pattern in general; it is not tied to any one platform's workflow.
A single source file and a single instruction are the starting point — no separate stages to queue up beforehand.
The core engine produces the structural and language-consistent base draft that the named model then shapes.
Prose, Narrate, Drama or Press applies its format-specific structure, or Lingua applies the requested language — and returns a complete draft.
Editorial review is available as a separate, optional step after a draft is returned, priced on its own — it is not part of the single-prompt generation described above. See usage & pricing for details.
These figures describe PUBXX 1.0's own output and sales record to date, independent of any single platform. They are a record of engine performance, not a forecast for any individual title produced with it.
| Metric | Figure | What it describes |
|---|---|---|
| Titles produced through the engine | 350,000+ | Across all named models to date |
| Of catalogue sales from engine-assisted assets | 90% | Company-reported, across the engine's output to date |
| Returns, clawbacks & sales returns | 0 | Reported on the engine's AI-assisted assets to date |
| Production cost reduction | Up to 80% | Against an all-human workflow, not a revenue claim |
These figures describe SEW Capital's own engine performance to date. They are not evidence of how any individual future title will perform, and producing more drafts does not guarantee more sales.
SEW Capital owns 100% of PUBXX 1.0 and the five models built on it — Prose, Narrate, Drama, Press, and Lingua — along with the IP and output those models generate. No token confers any ownership of the engine, the models, or their output.
This page describes the engine and its models as infrastructure in their own right. It does not describe, and should not be read as describing, the utility of any specific token or the mechanics of any specific platform that may use this engine's output.
This page describes PUBXX 1.0, the AI engine owned by SEW Capital, and the five models built on it — Prose, Narrate, Drama, Press, and Lingua. It is an operational reference. It is not a prospectus, not an offer, not a solicitation, and not a recommendation to acquire, hold or dispose of any token.
Not tied to a platform or token. This page intentionally does not describe how any specific platform sequences its workflow, nor the utility, mechanics or settlement of any specific token. Those are documented separately, elsewhere in the insights list, and token usage pricing for these models is set out on the usage & pricing page.
Not a forecast. The figures on this page describe the engine's track record to date. They are not evidence of how any individual future title, format or language edition will perform.
We do not publish price information, return expectations, yield, appreciation claims, or any statement of value for any token on this page. Nothing here should be read as financial, legal or tax advice.
We can take you through the engine, how each named model specialises it, and how Lingua fits across them. This is documentation, not a pitch.
hello@sewcapital.co.uk See usage & pricing