User Guide & Manual is an example of an agent your team can build in Sylvie: point it at a product and it produces the full instruction set, a written guide with numbered steps, specs and troubleshooting, plus the how-to video script with chapters, on-screen text and voiceover. It works from your brand brain (the real product facts, what changed in this release, the safety and regulatory wording, and the terms your brand uses for every button and part), so the steps match the product customers actually have in their hands. Sylvie assembles the agent on your own memory, it is not a template you download.
Sylvie does not ship this agent pre-built. It is an example of an agent your team can build with Sylvie, powered by your brand brain, so it runs your workflow the way you actually run it.
User Guide & Manual is one of the agents a marketing team can build on Sylvie. Give it a product, a firmware version or a feature release and it drafts the whole instruction set: what is in the box or the plan, setup step by step, everyday use, troubleshooting, the FAQ and the spec table, alongside a companion video script with chapter markers, shot or screen-capture notes, on-screen callouts and a voiceover track.
Documentation is the deliverable everyone needs and nobody is staffed for. It gets written the week before ship by whoever is free, it goes stale the first time the interface changes, and the video version usually never happens at all. An agent built on your brand brain writes from the same product truth every time, so the manual, the help center article and the video all say the same thing.
Before drafting, the agent reads the brand brain for that product: the spec sheet, the feature list and what changed in this release, the terminology the brand uses for each button, part and plan, the warnings and regulatory lines that must appear verbatim, the tone the brand takes with customers, and the guides you have already published. Because that memory is permission-aware and isolated per client, one manufacturer's unreleased specs never end up in another's manual.
Then it runs the job inside Sylvie. It sequences the steps, holds to one instruction per step, marks where a screenshot or a shot belongs, times the video script against the written sections, writes the on-screen text and the voiceover, and cites the source behind every spec and claim. Anything it cannot ground (an undocumented step, a number missing from the sheet) comes back flagged for a human instead of invented.
Every instruction traces back to a documented product fact, so nobody follows a step that no longer exists. Gaps in the spec sheet get flagged, never filled with a plausible guess.
The written manual and the how-to video come from one source, so the wording, terminology and order of steps agree. Support stops fielding tickets caused by two versions of the truth.
Clear setup instructions and a real troubleshooting section absorb the questions that would otherwise reach a human. The queue gets shorter without adding anyone to it.
When a release renames a menu, the agent reissues the affected steps and the matching video segments across the library, instead of someone hunting for stale screenshots.
Warnings, disclaimers and regulatory wording are held in the brain as fixed text and reproduced exactly, so a rewrite for clarity never quietly rewrites a line legal approved.
Twenty guides written by nine people usually read like nine products. Because terminology and tone live in the brand brain, the whole library reads like one careful author.
A few of the documentation jobs teams hand to this agent once it is built on their brand brain. Each one runs end to end in Sylvie, from product facts the agent already holds.
The hardware is boxed and the printed insert is due Friday. The agent drafts the quick start guide from the spec sheet and engineering notes already in the brain, so the insert makes the print deadline and the setup video is live the day the first units land.
A redesign renames half the menu and suddenly forty help articles and a dozen videos are wrong. The agent rewrites the affected steps and reissues the matching video scripts in a single pass, turning a month of cleanup into a two-day review.
Someone recorded a rambling forty-minute walkthrough for the team. The agent restructures it into a chaptered script, a tight voiceover and a written guide to match, so a raw recording becomes a publishable onboarding video instead of sitting in a shared drive.
The same setup question arrives fifty times a month. The agent turns that pattern into a proper troubleshooting entry and a sixty-second how-to video, and volume on that step drops instead of the team answering it one customer at a time.
A guide has to ship in several languages without losing the safety wording or the brand's terminology. The agent adapts each version from the same source of truth, so localization takes a week of review rather than three rounds with an outside vendor.
You win an account with sixty SKUs and almost nothing written for any of them. Because the specs load into the brain during onboarding, the agent works through the catalog systematically, delivering a complete, consistent library in weeks without hiring a technical writer.
Both, from one brief. The agent produces the written guide (setup, everyday use, troubleshooting, specs) and the companion video script with chapters, on-screen text and voiceover, so what a customer reads and what they watch agree. When you build the agent you decide which outputs your version produces.
It writes only from what is in your brand brain: the spec sheet, the release notes, the documented behavior. Every instruction is cited to its source, and when something is missing or ambiguous the agent flags the gap for a human rather than filling it with something plausible.
No, and that is deliberate. This page describes an example of what teams build in Sylvie. You assemble it on your own brand brain and set the structure, the terminology, the reading level and the review steps, so it produces your documentation standard instead of a generic template.
You update the product facts in the brain once and the agent reissues the affected sections across guides, help articles and video scripts. Documentation stops being something you rewrite from scratch every release and becomes something you keep current.
Yes, when you build it that way. Mandatory warnings and regulatory text are stored in the brand brain as fixed language the agent reproduces exactly rather than paraphrases, and your reviewers still sign off before anything is printed or published.
Each brand and each product line gets its own isolated, permission-aware brain with its own terminology, tone and compliance rules. An agency can document a power tool range and a SaaS platform side by side with no chance of one product's language leaking into the other.
Book a demo and we will map the workflows worth turning into agents, and show how each one runs on your brand brain.
Request a demo