The walkthrough
Watch @Cohenta run a campaign.
A live Slack thread: goal and sources in, the engine working in plain sight, the finished campaign at Approve.
in short03 facts
A working run, not a reelStatus ticks in the threadAd creative lands paused
Just mention it
A campaign starts as a mention.
The intake
Your part is small, by design.
What you bring
A goal and your facts. Research, KPIs, claims — every number downstream traces to a source you gave.
- A marketing goal — awareness, demand, a launch
- Your source material — research, facts, KPIs
- Your site, so the run learns your brand and audience
- Optional: voice, palette, messaging pillars, banned terms
- Optional: named competitors to measure against
What happens next
One mention covers the whole function. Nothing is re-typed; every asset stays traceable from brief to publish.
- Research — keywords, trends, audience and competitor signals
- Planning — a multi-channel campaign mapped to the funnel
- Authoring — source-grounded copy with per-channel format rules
- Rendering — the right format for every channel, most of it model-free
- Governance — quality gates, then your approval before publish
What comes back
Not drafts — posts, articles, visuals, short video, decks, and landing pages, cut native to each channel.
- Numbers verified, never invented
- One identity across every asset
- A paper trail from brief to publish
- Heavy formats render on demand
After you hit send
Then three things happen.
A confirmed briefA short intake pins down goal, audience, and claims
One governed runThe whole pipeline, checkpointed from first step to last
Your approvalPublishing waits for your explicit confirm
In practice
A real thread, unedited.
you @Cohenta launch our new analytics tier — drive signups
# in-thread: your site · q3-kpis.csv · positioning.md
▸ brief ✓ validated · audience & claims locked
▸ run research · plan · author · render · govern
▸ status 03/12 assets done · ticking in this thread
◆ campaign ready — paused at the gate [ Approve ] [ Request changes ]
A button, not a guess: the exact run resumes, even after a restart. Nothing publishes without the press.
Before you start
Questions, answered.
What exactly is the walkthrough?
A real campaign, run start to finish in Slack — brief in, finished assets out, pausing at the approval buttons. Ask questions as it works; it's a thread, not a stage.
What should I bring?
A marketing goal, plus the material you want to stand behind — research, facts, KPIs — and a pointer to your site. Brand voice and named competitors are optional but sharpen the result.
How does the run produce assets?
One mention flows through research, planning, authoring, rendering, and governance in a single run. Assets land as they finish, and the pipeline saves its place after every step.
Who approves what publishes?
You do — the run holds at the gate, surviving restarts, until you press a button. Publishing is a separate, deliberate yes.
Will the output stay on brand and accurate?
Yes — every statistic is checked against your facts, and anything untraceable is flagged for review. Brand rules cascade into every step, so a post, a deck, and a landing page share one identity.
Does Cohenta see my keys or credentials?
No. Cohenta never receives or stores your provider keys, publishing credentials, canonical assets, or campaign state. Data is shared only with the platforms and providers you have chosen.
Is Slack the only place this works?
The walkthrough runs in a live Slack thread. @Cohenta also works inside Microsoft Teams, ChatGPT, Claude, and Copilot — plus a CLI and API — and a campaign started in one can be checked from another.
What happens if something fails mid-run?
Each asset fails on its own and every step has a grounded fallback, so coverage narrows but output never hits zero. A crash or timeout resumes from the last completed step — no re-research, no re-generation.
Ready when you are
One message starts it.
Tell us your goal and where your facts live. The campaign ends at a button only you can press.