Quick answer: Faster AI skills are named, repeatable ways of working that the assistant follows — built-in ones cover every corner of the workspace (website, CRM, posts, journeys, payments, bookings, and more), and custom ones encode your recurring processes: the monthly newsletter routine, the way you quote jobs, the review rules every draft must pass. You never invoke a skill by name — describe the result you want in chat, and the right skill shapes how the work gets done.
There's a gap between an AI that's smart and an AI that works your way. Smart gets you a competent newsletter draft. Your way gets you the newsletter assembled the way your team always does it — the same sections, the same tone, the same "always include one customer win," the same review step before anything sends.
That gap is what Faster AI skills close — and with today's update, you can teach your workspace skills of its own.
What a skill is
A skill is a repeatable way of doing a kind of work — part process, part tone, part checklist, part review rules — that the AI applies consistently every time that kind of work comes up. Not a macro, not a template: a working method. When the assistant uses skills from chat, what you experience is simple: the draft comes back already shaped like your business does things.
The important design choice: you never name a skill to use one. You describe the result — "get a promo out for the new evening classes" — and the assistant picks the right capabilities and working methods for the job. Skills are how the AI works, not another menu you manage.
What ships built-in
Every workspace comes with skills covering the whole surface: website pages, customer records, forms, social posts, blogs, journeys, broadcasts, media, payments, bookings, courses, webinars, plugins, and workflows. That breadth is why one plain-language request can cross tools — "set up a waitlist for the spring workshop and email past attendees" touches forms, audiences, and broadcasts, and the built-in skills know how each piece should be prepared and what stays in review.
You've seen built-in skills at work in everything we've written about this season: updating pages by describing the change, customer work with scoped context, drafting content that keeps your voice. The constant across all of them: the AI prepares, a person approves.
Teach your workspace a skill of your own
The new part. If your team repeats the same instructions — for posts, pages, quotes, follow-up, reporting — that's a skill waiting to be written. Creating one is closer to writing a good onboarding doc than programming: name the recurring job, spell out the steps, the tone, the must-includes and never-dos, and the review rules. From then on, every request in that territory follows the method. A few that earn their keep immediately:
Three sections, one customer win, one practical tip, subject line under 50 characters, never more than one promotion, always drafts to review by the 25th. Write it once; every month's draft arrives pre-shaped.
What a quote includes, how options are presented, the language for deposits and timelines, what never gets promised in writing. Every AI-prepared quote follows the house rules — including the ones you usually have to remind people about.
Your channel rules, claim policy, and brand lines — applied to every social draft before a human ever sees it, so review time goes to judgment instead of housekeeping.
Skills are operating rules — review them like rules
A skill change affects every future piece of AI work in its territory, which is why skills get the same discipline as any operating policy: review and publish skill changes deliberately. Edited the tone rules? Reworked the quote process? The change is reviewed like the policy it is — clear, current, and safe for the whole team — before it starts shaping output. And when a teammate says "the AI stopped doing X the way we like," the skill is the first place to look and the single place to fix it.
This is also where skills and workspace memory divide the labor: memory holds facts (your offers, your audience, your tone in brief); skills hold methods (how the newsletter gets built, what a quote must contain). Facts inform every task; methods govern specific kinds of work.
And for the technically inclined: Faster CLI
One more surface worth knowing about. Faster CLI is an advanced chat for runtime-oriented help — the place to ask why a workspace action didn't complete, or to run a more technical review of configuration and workflow behavior, separate from routine drafting. Most teams will never need it; the one who debugs your setup will love it.
Key takeaways
- Skills are repeatable working methods: process, tone, checklist, review rules — that the AI applies consistently.
- Built-in skills cover the whole workspace: you describe results, never invoke skills by name.
- Custom skills encode your recurring processes: newsletter routines, quoting rules, review passes — written like an onboarding doc, not code.
- Skill changes are operating-rule changes: reviewed and published deliberately, fixed in one place.
- Memory holds facts: skills hold methods. Together they're how AI output stops sounding generic.
- Faster CLI: the advanced diagnostic surface for when something needs a technical look.
Frequently asked questions
Do I need to know how to code to create a skill?
No. Writing a skill is writing instructions — the same ones you'd give a careful new hire. If you can describe your process in a checklist with a few rules, you can write the skill.
How do I know which skill the AI used?
You mostly don't need to — the output is the evidence, and review is where you judge it. If a draft doesn't match your process, that's the signal to check the relevant skill's rules rather than re-explaining in every chat.
What's the difference between a skill and just telling the AI what I want?
Telling works once; a skill works every time, for everyone. The payoff is consistency — a teammate's request goes through the same method as yours, and a process improvement lands in one place instead of in everyone's habits.
Can a skill let the AI publish without review?
Skills shape how work is prepared — they don't bypass the draft-and-approve model. The review rules inside a skill typically make review stricter and faster, not optional. Who clicks publish remains a person, by design.
Where should we start?
With the instruction you've repeated most this month. If you've typed "remember to keep subject lines short and always include the booking link" more than twice, that's your first skill — ten minutes to write, saved on every request after.
Skills are live in your workspace today. Start with one recurring process, write it down the way you'd teach it, and let every future draft arrive already knowing the house rules.