Skip to content
Repic is live. Create your account
Repic
LoginSign up
Repic AssistantLive

An assistant that doesn't just answer, it does the work. With your permission.

Available on every screen, launcher or Ctrl+K. No separate page to visit.

Follow the thread

It routes, it plans, and then it asks.

The interesting part of an assistant is not what it can do. It is what it does before it does it.

  1. 01

    The request arrives

    In your words, from anywhere in the studio.

  2. 02

    It classifies

    What kind of job this is, and which engine owns it.

  3. 03

    It plans

    The steps it intends to take, in an order it can explain.

  4. 04

    It asks

    Anything impactful stops here for a yes. This is the beat the whole design is built around.

  5. 05

    It acts

    And what it did is in your history, attributable, and reversible.

Ask, it answers
“What's weak in my brand profile?”“List my recent runs”
Reads and navigation run instantly, nothing to approve.
Tell, it acts
“Queue a carousel about launch week”“Regenerate my palette warmer, keep the logo”“Turn on the weekly digest”
Impactful actions wait behind the Allow/Disallow card below.
The safety contract

Five points. Always visible. Never fine print.

1It explains its intent first.2Impactful actions wait behind an Allow/Disallow card.3Disallow = nothing changes. Verified.4“Just do it” phrasing can't skip the gate.5Your exclusions (“keep the logo”, “not salesy”) are respected and forwarded.
Needs your approval
Change

Remember this Emoji usage: sparing → none (you had sparing). Every tool will use it from now on.

AllowDisallow
A real approval card from the product, static preview, faithful to its brand profile.
Impact levelWhat happens
Read & navigateQuestions, tours, listing your runs
Runs instantly
ChangesEdits to your brand, queued content, settings
Gated
DeletionsAnything that removes your work
Gated
Always on: a deterministic planner keeps the assistant responsive even when no live model is selected.
Go deeper

2 min read · 5 sections


An assistant that does the work, and asks before it changes anything

Most product assistants are a search box with a personality. You ask a question, you get a paragraph, and then you go and do the thing yourself. The Repic Assistant is built to do the thing, which immediately raises the only question that matters about such a tool: what stops it doing something you did not want?

The answer is a confirmation gate on every action that changes your brand. The Assistant can read your persona, score your kit, draft copy and run engines, and when a turn would mutate something durable it shows you the specific change and waits. Nothing is written to your brand behind your back, and a declined action leaves no trace of itself.

It is available from every screen through the launcher or Ctrl+K, deliberately rather than as its own destination. An assistant you have to navigate to is an assistant you stop using by the second week.


Step by step, with nothing skipped

  1. Open it anywhere

    Launcher or Ctrl+K, on every screen. There is no assistant page to visit, because context is the whole value and a separate page throws it away.

  2. Ask in plain language

    It already knows your brand, so "what is weak in my brand?" or "draft three carousel topics for next week" are complete instructions.

  3. It proposes, you confirm

    Anything that would change your persona, kit or settings is shown as a specific proposed change and waits for you. Read-only answers just answer.

  4. Actions are traceable

    What it did, and when, is visible after the fact. An assistant with write access that keeps no record is not one you can trust with write access.


What this does not do

It does not act silently

No mutation without an explicit confirmation. This is the constraint that makes the rest of the capability acceptable.

It is not a general-purpose chatbot

It is scoped to your brand and to Repic. It is not the place to ask it to write your emails or debug your code.

It does not replace the Quality Engine

It can tell you what is weak, but the scoring, the alerts and the staleness tracking are the Quality Engine doing its job.


Straight answers

Can the Assistant change my brand without asking?

No. Every action that mutates durable state is gated behind an explicit confirmation that names the specific change. Declining leaves your brand exactly as it was, with no partial write.

Copy link to this question
Does it have access to everything in my account?

It works within your own brand context and your own permissions. It cannot reach another account brand, and it cannot do anything you could not do yourself in the interface.

Copy link to this question
What can it actually do, as opposed to answer?

It can read and explain your persona and kit, surface what the Quality Engine has flagged, draft copy and topics, and run the creation engines. The dividing line is not read versus write; it is whether a write is confirmed.

Copy link to this question
Why is there no dedicated Assistant page?

Because the useful version of this is available where you already are, with the context of the screen you are on. A separate destination costs you that context and, in practice, costs you the habit.

Copy link to this question

Free, no signup

Not ready to sign up? Run the brand voice generator instead.

Paste two or three things you have written and get a measured voice profile: sentence rhythm, register, who you address, plus rules and a rewrite.

Try it free →
Repic is open

Delegate the work. Keep the keys.

Sign up