Feature
Campaign memory
Campaign memory keeps each workspace's voice, rules, and campaign context in versioned records instead of scattered docs. The brand voice guide, the versioned brand profile the review checks enforce, memory items, and structured campaign briefs all live in the workspace, so planning starts from what the brand has already decided.
What gets remembered
Voice, rules, and briefs that stay with the workspace.
Brand voice guide
Each workspace has one voice record: summary, target audience, approved phrases, banned phrases, and approved examples, with who approved it and when.
Export the voice guide
Download the voice guide as Markdown or open a print-ready version to save as PDF, so writers outside SignalDeck work from the same reference.
Versioned brand profile
Audience, voice examples, style rules, prohibited claims, required disclosures, vocabulary, source preferences, and approval policy are saved as numbered versions. Saving an unchanged profile doesn't create a new version.
Rule changes re-check open work
When a new profile version changes the rules, every draft, in-review, changes-requested, or approved version that hasn't been published is marked for re-evaluation and can't be approved until the checks run again.
Memory items
Append memory items to a workspace with a kind, polarity, source, and visibility, for example a phrase to avoid after a client rejected it. Items are scoped to the workspace that created them.
Structured campaign briefs
A brief records the objective, audience, evidence and source links, dates, and destinations. Missing pieces are listed as unresolved items, such as no evidence attached or no destination chosen.
Proposed plans from the brief
Ask the AI assistant to propose a campaign plan from the brief. Plans are labeled AI drafts with source links, and missing facts stay unresolved.
Repetition check
Each new content version is compared with the workspace's recent versions, and near-duplicates fail the review check.
How it works
How memory shapes the next campaign.
1. Write the voice and rules
Fill in the brand voice guide and the brand profile. An owner or admin approves the voice examples.
2. Brief the campaign
Create a campaign with an objective, audience, evidence, dates, and destinations. Unresolved items show until the brief is complete.
3. Propose a plan
Request a proposed plan from the brief. It appears on the campaign as an AI draft for the team to revise.
4. Review against the rules
Posts written for the campaign are checked against the current brand profile, recent posts, and attached sources before approval.
5. Update when the brand changes
Change a rule and unpublished work under the old version is flagged for re-checks, so nothing is approved under outdated rules.
Isolation and limits
Where memory stops.
Per workspace
Voice, brand profile versions, memory items, and campaigns are stored per workspace. Checks read only the current workspace's profile and recent posts.
Per tenant for sources
Source snapshots used in generation or checks must belong to your organization; a reference to another tenant's snapshot fails closed.
Not automatic learning
Memory is recorded deliberately. Rejections and performance results aren't turned into memory items on their own.
Who uses it
Context that survives team changes.
Agencies
Keep each client's voice, banned phrases, and required disclosures in their own workspace so a new account manager starts from the same rules.
Brand teams
Version the brand profile when legal adds a required disclosure, and see unpublished posts flagged for re-checks right away.
Nonprofits
Record required program disclosures and approval policy once, then brief each appeal with the evidence behind its figures.
Creators
Write down the phrases you never use and export the guide for editors and brand partners.
FAQ
Frequently asked
Does memory leak between client workspaces?
No. Voice records, brand profile versions, memory items, and campaigns are stored per workspace, and the review checks only read the current workspace's profile and recent posts.
Does SignalDeck learn from rejected copy automatically?
Not yet. You can record a rejected line as a memory item to avoid, but rejections aren't captured into memory on their own.
What happens to posts in review when we change the brand rules?
Unpublished versions created under the previous profile are marked for re-evaluation, and approval is blocked until the checks run again under the new version.
Can we export our brand voice?
Yes. Download it as Markdown or open the print-ready version to save as PDF.
Is campaign memory included on every plan?
Brand voice, brand profiles, and campaign briefs aren't gated by plan. Proposing a plan with AI uses AI credits, which are included on Agency, Scale, Command, and Enterprise.
What isn't supported?
Automatic learning from approvals, rejections, or performance, and sharing memory across workspaces, aren't supported.
Keep the context
Start every campaign from what the brand already decided.
Set up a workspace's voice and brand profile, then brief your next campaign.