Long-form studio
What it does
It is PULSE’s long-form copywriter: it writes the pieces a specialist reads, or listens to, to the end.
- Writes whitepapers in Studio: an executive summary, sections that each make one argument, and a conclusion with next steps.
- Writes the blog posts the social engine drafts when the blog channel is armed.
- Writes the weekly audiobook, as prose meant to be read aloud.
Shorter Studio pieces (podcast and short-video scripts, outlines, reports) are written by the channel content creator.
What it reads
- The brief: topic, industry, length, and any instructions you give.
- For a whitepaper, the product corpus: markdown fact files shipped with the
deployment, searched for the topic’s terms and read whole, with the passages
that matched put in front of the agent labelled by the file they came from.
Files whose name begins with
_are standing rules and go in every time, matched or not. - For the daily blog draft: the campaign’s name, objective, and audience, and that day’s web research.
It is deliberately not given your marketing diagnosis. A whitepaper is a document about a product, and grounding it on a marketing diagnosis turns it into a channel plan for someone else’s business.
What it produces
Long text, shown once it has passed the brand and output checks: markdown with headings and short paragraphs for written pieces, and plain flowing prose with no lists or tables for audio. Audio itself is produced by the platform’s text-to-speech, not by this agent.
The length it can write in one pass depends on the deployment’s time budget. See Sovereign inference.
Human oversight
- You edit and approve every piece. Saved pieces wait in Drafts to approve on Campaigns like any other content.
- Blog posts, podcast episodes, and the audiobook publish only when an administrator has armed that channel. See Publishing controls.
Model
The deployment’s base model, shared with every other PULSE agent — on Runink Server, a Qwen-family model under the Apache-2.0 licence. See Sovereign inference.
Where it runs and data handling
On the deployment’s own inference server: Runink’s shared machines on a Trial, your own cloud on Cloud and your own servers on On-premises. No outside AI service is used on any plan.
Guardrails
- Value first, no hard sell. It teaches the topic; your product appears at most once, as an honest example, never as the headline.
- Facts only from the brief and the corpus. Every capability claim has to trace to a passage it was given. It never invents statistics, studies, quotes, customer names, or case results. If a point needs a number it was not given, it makes the point without one.
- No verified source, said out loud. When the corpus has nothing on the topic, the agent is told so explicitly and writes from the topic and industry alone, rather than inventing capabilities, integrations, certifications, customers, or partnerships.
- No hype words, no clichés, no fake urgency.
- Will not handle payments or credentials.
- Its output is screened before it reaches you.
Getting good results
- It writes from the brief and the product corpus, so a detailed brief gives a richer piece.
- Very long documents follow the deployment’s time budget; see Sovereign inference.
- It writes text; audio comes from the platform’s text-to-speech.