Skip to content
Inkstand

Copy that can’t say the wrong thing.

Your writing tool doesn’t know your brand has a banned-words list, a required disclaimer, or a regulator. Inkstand does — and it will hold the release until someone deals with it.

Two studios over one rule engine and one record: Copy Studio for documents and emails, Carousel Studio for social.

For agencies and in-house teams in health, finance and legal

Invite-only · EU-hosted · No tracking scripts

An animation of the editor. A headline is typed out, and when the word “cure” appears the checker marks it on the slide, opens a blocking finding against the rule “Do not claim anything cures a condition”, and refuses the release.

How much of “our AI keeps you compliant” is a fact?

For us the answer is a number, per rule, on every export. Nobody else in this category will even tell you the question exists.

10 rules in this brand kit

  • banned phrase
  • required disclaimer
  • reading level
  • contact details
  • closed period

verified

a fact

Runs on every export. No model involved.

  • unsubstantiated claim
  • advice to an individual
  • someone’s name

reviewed

an opinion

Catches what patterns cannot. Will miss things.

  • house style
  • tone of voice

steered

nothing yet

Written into the brief and never inspected.

A rule that needs judgement counts as reviewed only if the review actually ran. If it was switched off or it errored, it drops to steered and the audit record says so.

House rules that nobody below you can switch off

Set rules once at account level and every brand inherits them — a client brand if you are an agency, a product line if you are not. What an editor may do to an inherited rule runs one way only.

set once, at account level

“Never claim a cure.”

Blocking. Applies to every brand below it.

every workspace inherits it

Allowed

A workspace editor makes it stricter

Narrower wording, a harder severity, more surfaces. Local rules can always add.

Refused

A workspace editor removes it, or drops it to advisory

The attempt is reported back rather than silently undone, so nobody finds out at export.

Approval that means something

Sign-off is bound to a fingerprint of the exact content reviewed. Edit a headline afterwards and the approval voids itself.

Every release is on the record

Who exported, what they acknowledged, which rules were checked, and the reason for any rule they waived. Append-only.

The engine doesn’t only run here

Content gets written in a dozen places and reviewed in one. The same checker that holds the line inside this editor answers over HTTP, so the check can sit where the sentence is actually typed.

The same function, not a copy of it

Two implementations of “is this allowed” eventually disagree, and the day they do you have content one surface cleared and another flagged. There is one, and it is the one you have already seen work.

Overrides come back to one place

When somebody publishes over an objection, the reason lands in the same append-only record whether they did it here or somewhere else. Five tools’ worth of overrides is not an audit trail.

How the review engine works — the tiers, what a connection carries, and the answers procurement asks for.

And it doesn’t look like a template

Governance is worthless if nobody wants to use the output. Six typeface pairings, eight colourways, six compositions — and no small uppercase label sitting above every headline.

Field Notes Poster colourway, Wide type

What should this plug into first?

One tap · no email

  • Something else

Every brand you run sits in Stockholm — yours, or your clients’. No analytics vendor in the stack, on this site or in the product. Ask for your data back or erased and you get it within thirty days.

Privacy policy →

Currently invite-only.

We onboard a small number of teams at a time, each set up with its own rule set. Tell us what you’d use it for and we’ll get back to you.