Stencil
On-brand creative · at scale · zero drift

The brand is locked.
The AI only writes the headline.

Teams already use AI for branded content — and it keeps drifting: wrong font, off colour, moved logo. So a human fixes every output, and it never scales. Stencil compiles your brand into the template as code. The model can only fill the slots you define, so an off-brand asset is structurally impossible. Hundreds a day. No designer in the loop. Nothing to QA.

1 template → unlimited assets · every export audited against the brand spec · correct by construction, not by review
PRESS PROOF · sample brand: TIDAL — cold-water swim club, Norwich
0/42
0/90
0/22
Locked by the template
#0C3B47 #EDF3F2 #FFD23F
Type Helvetica Neue 800 · logo 15px / .24em · margins 26px · grid 4:5 baseline. The AI can't touch any of it — only the three slots above.
PROOF № 0031 · 4:5 ✓ On-brand
COLD-WATER CLUB
NORWICH · EST 2021
Dawn dip, Saturday
Meet at the boathouse, 6:45am. Water's 11°C. Cake after, obviously.
Free for members
@tidalswim.nor
✗ font substituted ✗ colour off-palette ✗ safe margins broken

Type into the slots — the proof re-renders live. Flip to Raw AI to watch the same brief drift off-brand, then back to Stencil to snap it into spec. Export only works when the audit passes — drift can't leave the building.

The difference

Same brief. One drifts. One can't.

⚠ Prompted AI
  • The brand lives in the prompt — a suggestion the model is free to ignore
  • Invents fonts and colours it "thinks" fit; moves the logo; breaks the margins
  • Every output slightly different, so consistency is luck
  • A human reviews and fixes each one — that's the ceiling on scale
✓ Stencil
  • The brand lives in the template — compiled as code, not prompted
  • The AI writes into defined slots; it physically can't touch type, colour or layout
  • Every output identical in form, on-brand by construction
  • Nothing to review — it renders to spec or it doesn't render
Proof of scale

One template. A week of posts, in a click.

Feed it a content calendar and it fills every slot of every post — all identical in brand, all different in content. Eight from the same Tidal template:

8 assets rendered · 8 on-brand · 0 sent back for fixes
Under the hood

Four steps. One that never bends.

STEP 01

Compile the brand

Your guidelines become a locked template: colour, type, logo, grid and margins baked in as code — not instructions the model can shrug off.

tidal.brand.lock · 12 rules
STEP 02

AI writes the content

From a brief, product feed or calendar, the model drafts the copy and picks imagery — the parts that should vary.

brief → slots.json
STEP 03

The engine fills the slots

Content drops into the defined slots only, and the layout engine renders it to spec. Every output passes the same 12-point brand audit you saw above.

render(slots) · audit 12/12
STEP 04

Export anywhere

PNG, PDF, video, or straight into your scheduler through the API. One brief in, a whole campaign out — on-brand, every time.

png · pdf · mp4 · api
Example packaging

Priced per brand, not per designer.

Illustrative tiers — shaped to the client in the room.

Starter
£750/mo
  • 1 brand, up to 3 locked templates
  • Unlimited on-brand assets
  • Self-serve slot editor
  • PNG + PDF export
Studio · most useful
£2,400/mo
  • Up to 5 brands, 15 templates
  • AI copy + image selection from your feed
  • Automation into your scheduler
  • Video + carousel formats
Scale
Custom
  • Unlimited brands & templates
  • API + white-label
  • Brand-guardian approval flows
  • Dedicated pipeline & SLA