Calibre Studio

03/08/2026

Most AI visuals survive the first image and fall apart by the fifth.

Not because the model is bad. Because nothing is governing it. Every batch is a slightly different interpretation. The light shifts, the palette drifts, the casting stops matching. Fine for one hero shot. Fatal across a season.

So we build the rules first. Every Directed engagement starts with a Visual Operating System: lighting, palette, casting, wardrobe, environments, props, motion language. Written down, agreed, then applied to every frame that follows.

Then we generate at volume and throw most of it away. The creative director selects, rejects, refines. Selected frames go through the same pipeline as a traditional shoot: retouching, colour, compositing.

AI generates. We decide. That's the whole difference.

Single drops from EUR 2,000. Foundation builds from EUR 7,500.

https://www.calibrestudio.co/works/directed

02/08/2026

AI can already handle a good chunk of the studio day. Backgrounds, lifestyle scenes, campaign concepts. Most shoppers can't tell.

Where it still breaks: logos, packaging copy, and exact proportions. And no prompt fixes it. The only thing that does is feeding the model your real product as reference, then checking every frame against it before it ships.

We shoot photography and motion, and produce AI stills and motion. So here's the honest version of where that line sits, what it costs, and the workflow that keeps the output accurate.

Where AI product photography already beats the studio day, where it still breaks (logos, packaging), and the directed workflow that keeps output on-brand.

21/07/2026

We spent 7,300+ impressions of our own money to find out what ChatGPT Ads actually do. Then we wrote it down.

Most posts on this channel are OpenAI's documentation, reworded. This one is our account, our data. What we found:

CTR sat at 1.4–1.5% on cold placement — better than display, level with good social prospecting. Blended CPM landed around $22, cheap for a spot inside a live buying conversation, and unlikely to stay that way. The targeting isn't keywords or audiences; you write the buyer's situation in plain language and the platform matches it to real chats. And the real work is measurement, because a browser pixel undercounts this channel badly.

Attention is cheap, intent is high, and most of your competitors haven't turned up yet. Whoever builds history now makes the late arrivals pay to catch up.

Full breakdown, real numbers 👇
https://www.calibrestudio.co/blog/running-chatgpt-ads/

Real numbers from running ChatGPT Ads on our own account: CTRs, context hints vs keywords, why standard tracking undercounts, and what early CPMs look like. From the agency that runs the channel for itself.

20/07/2026

See how AI sees your business.

Search changed. Your customers ask AI now, and it answers with a shortlist. This audit scores the signals ChatGPT, Claude, Gemini, Grok, Perplexity, Copilot and Google AI Overviews read before they recommend anyone.

Check your AI Visibility // Free · 60 seconds · No card
https://indexed.calibrestudio.co/

See how AI sees your business. The signals AI reads before it recommends you, scored.

19/07/2026

Three things now decide whether a buyer finds you inside ChatGPT. Two you earn. One you can buy.

Almost nobody is doing all three.

AEO, answer engine optimisation, is extraction. An answer engine lifts one complete answer and hands it over. The featured snippet, the knowledge panel, the voice assistant reading a single response aloud. One occupant, no runner-up.

GEO, generative engine optimisation, is synthesis. A generative engine writes its own answer and stitches in fragments from four or five sources. You're not competing for a slot. You're competing to be one of the voices inside a paragraph the machine wrote itself.

Different mechanics. Same test underneath: does this engine trust and understand you enough to use you at all. Structured data, entity clarity, mentions on sites it already trusts, and facts that agree with themselves everywhere a machine might look.

The third thing is new. ChatGPT sells ads now, as clearly labelled cards below the response, live in the US, UK, Canada, Australia and New Zealand.

And the part that matters: the card sits beside the answer. It doesn't change a word of what ChatGPT says about your category. You can buy the placement. You cannot buy the recommendation.

So these aren't alternatives. They're two jobs on two different clocks. The ad reaches people this week and stops the day the budget does. The earned mention takes months and doesn't switch off.

Calibre runs both sides. Indexed scores where you stand across the engines today, Get Found does the earned work, and we build and run the ChatGPT campaigns alongside it.

Most companies haven't started on any of it. Which is why the basics still move fast.

Where you currently stand, scored in about sixty seconds, free:
https://indexed.calibrestudio.co/

The full mechanics, AEO and GEO explained properly:
https://www.calibrestudio.co/blog/aeo-vs-geo/

24/05/2026

lms.txt isn't a magic SEO trick. Google doesn't index it. Most AI engines don't strictly need it. But not shipping one in 2026 leaves free AI citation on the table for a few hours of work.

We shipped Calibre's this week: calibrestudio.co/llms.txt
It tells ChatGPT, Claude, Perplexity, and Google AI Overviews who we are, what we make, and which pages to cite when someone asks "creative studio for luxury fashion" or "AI visuals for premium brands."

What we put in:
• A Key Facts block so the structured signals land first
• An Industries section, one bullet per vertical
• Recognition (awards and exhibitions)
• Press & Editorial as discrete entities, not paragraph mush
• A FAQ block, because AI engines lift Q&A directly into answers
• Case studies as a curated 10, not the full dump

Paired with proper JSON-LD schema on every page. Organisation and Person nodes for founders, LocalBusiness for both offices.

If you run a brand or agency and want yours done properly, DM open. We're doing this for clients as part of our Intelligence practice.

05/12/2025

New work

Field‑tested for the elements, the aTHlet Anorak offers a boxy, lightweight fit with mesh panels, an elastic hem and a hidden hood built into the collar. Its front pocket and signature rubber details keep essentials secure and the silhouette clean.

Paired with the aTHlet Technical Pant, engineered in a regular fit for breathability with an elasticated waist and a zip pocket for your essentials.

Embrace movement. Embrace function.

https://www.athlet.world

03/12/2025

New work for aTHlet
https://www.athlet.world

aTHlet in its purest form.
No noise. No crowd. Just you, your breath, and the work.

Built for motion.
Designed for stillness.
Engineered for the hours no one sees.

https://www.athlet.world

30/11/2025

Blur the noise.
Sharpen the purpose.