Living marketing memory

A marketing memory that shows its work.

Confirmed facts stay confirmed. Inferences arrive for review. Every useful detail keeps its source. Every teammate reads the same approved truth.

Start with Stash →See the Stash product
€19/mo · Handler included · Exportable memory
A memory change you can trust
See what was learned, where it came from, and whether it is approved.
Proposed memory
Primary buyer objection
“The team worries AI will erase the founder’s voice.”
Source · Parley interview · 18 min ago
After approval
Shared with 12 agents
Chief addresses the objection in next week’s plan.
Ghost uses the founder’s words—not a generic trust claim.
Cite looks for evidence that answers it.

Why prompt context fails

Memory without governance becomes confident drift.

It forgets

The next session starts from zero, so your team keeps rebuilding the same brief.

It conflicts

A website, a call, and an old deck can each claim to be the newest truth.

It hides provenance

The model remembers a claim, but nobody can tell where it came from or whether it is safe.

01

Capture

A site, call, document, founder correction, or agent discovery enters with a source.

02

Propose

New inferences stay visibly separate from confirmed memory.

03

Approve

You keep, edit, or discard the proposed change before it becomes shared truth.

04

Use & learn

Agents read the approved context; useful learning can return as another reviewable proposal.

The memory contract

What every teammate can inherit.

Not every agent needs every field. They all start from the same governed source.

Positioning
One-liner, wedge, why-now, and comparisons to avoid
Audience
Buyer pains, triggers, objections, and language
Voice
Tone, banned phrases, signature language, and examples
Proof
Claims, metrics, stories, provenance, and expiry
Narrative
Beliefs, points of view, stories, and content pillars
Constraints
Approval rules, brand boundaries, and what must never be claimed

One source, specialist views

Every teammate reads the context its job needs.

Chief
Marketing Operator
Handler
Social Scheduler
Ghost
LinkedIn Writing Agent
Pulse
X Writing Agent
Stitch
Threads Writing Agent
Cite
Long-form Article Agent
Parley
Expertise Interview Agent
Herald
Newsletter Agent
Whisper
Employee Advocacy Agent
Pixel
Brand Visual Agent
Clerk
Performance Analyst
Scout
Conversation Discovery Agent

Living memory, in straight answers.

No. Agents can only propose changes. A suggestion stays out of the record agents read until it is confirmed, and a discarded suggestion is not proposed again.
No. Stash structures your facts, proof and voice rules into fields, and gives each claim a status that decides whether agents may use it.
Ghost, Pulse, Stitch, Chief, Herald, Cite, Whisper and Pixel draw on it, and Handler reads your posting windows from it.
Yes. Stash connects to your AI assistant through FeedSquad MCP. Your assistant can read your record and propose additions, which are held until they are confirmed.
Not today. Stash belongs to one account.
Stash includes Handler for scheduling and publishing, with no monthly publishing limit. Daily posting limits and spacing between posts still apply.
Yes. You can edit what Stash holds at any time.
A normal knowledge base stores material for people to find. Stash structures the specific context agents need at write-time and keeps provenance and approval state attached to it.
Stash is the shared context layer across the team. Each agent reads the parts relevant to its job, while the public page makes the shared-memory contract explicit.
Stash is €19/mo as the DIY memory and publishing path. It is also included by plans whose canonical configuration lists Stash, including Chief.

Correct it once. Brief the whole team.

Start with Stash alone, or get it as the shared memory inside Chief’s operating loop.

Start with Stash →See Chief