EN
For brand & category marketing teams

One prompt in.
A full campaign of creative out.

SOMIN reads your category on the live web, finds unclaimed territory, writes propositions, builds creative in your Figma components and scores every frame before spend.

Book a brand demo Talk to sales
SOMIN · Prompt to Creative LIVE
Step 01 of 11
One prompt in
a single chat message
inputOne message naming the client, congress and two skills. No research pack, brand kit or brief.
outputEverything after this is fetched automatically.
0/180 min machine time
Real run · MedTech congress launch, Sep 2026Client anonymized
1prompt
Brief to finished creative library
180min
Machine time, idle waits excluded
108frames
Brand-locked creatives in the client’s Figma
Trusted by brand teams in
11+ markets across APAC, MENA & Europe
P&GAudiNestléAbbottGojekKFCMothercareMcDonald’s
The creative performance equation
Campaign performance=Open territory×Brand-locked execution×Pre-spend validation
Most stacks optimise only the third term, after spend. SOMIN settles all three first.
The brand-team reality

Making creative is cheap now. Knowing which creative isn’t.

Generative tools made options cheap but cannot say which to run, whether the category has saturated it or whether it is on brand.

01

The brief starts from last year

Territory comes from the old deck, not current category spend.

02

AI output that legal won’t clear

Invented logos, drifting palettes, unlicensed imagery — rebuilt by hand in production.

03

Validation happens in-market

The first honest read comes after spend, when cuts are costly and late.

Watch it check a brand's own feed for variety

Check · Is Grab's own feed repeating itself?/content-diversity-scorer
Condensed replay of a real run: SOMIN agents in Claude on Grab Singapore data, shown live at Marketing Mondays on 12 Oct 2026. No sound.
From prompt to production

Four moves, run as one line.

Each move feeds the next; nothing is re-briefed by hand.

01
Diagnose

Brand Tracker · Studies

Read the category.

Brand and scope read from the live web, then 15 parallel studies of category paid output.

  • Palette and type from live CSS
  • 15 studies, 6,502 content items
  • Share of voice: crowded vs. open
02
Decide

Insights · Propositions

Choose the ground.

Persona, tension and angle form a story per pillar; propositions cover only what you own.

  • Whitespace verdicts, not ranked adjectives
  • 21 propositions, each tied to a study ID
  • No promise the brand cannot keep
03
Produce

Expressions · Figma MCP

Build it on brand.

Expressions generated under a locked prompt suffix, then rebuilt in your Figma file (optional).

  • 108 frames from your own components
  • Licensed photography only
  • Wordmark composited, never generated
04
Validate

Scoring · Meta refresh

Judge before you spend.

Every frame is scored; only those clearing the gate go live. Measured CTR reopens the loop.

  • Predicted clickability with confidence
  • Gate: class and confidence
  • Bottom tier cut in review, not in-market
What one run produced

Measured on the clock, not on the calendar.

One real run; machine time excludes time spent waiting on people.

180min
Total machine time
Brand read, 15 studies, whitespace, propositions, 108 expressions, the briefing deck, Figma production and upload.
6,502
Content items analysed
Across 15 parallel studies: two probes, six trend windows and three whitespace cohorts.
75%
Lower briefing cost
On-brand quality in a fraction of agency hours, traceable to source studies.
Case study · MedTech × congress launch

A global surgical-technology manufacturer, one congress, one line of instruction.

From one chat message naming the client, congress and two skills to 108 frames in the client’s own Figma component set.

The clinical field spends 28.8% of its paid output on safety and standards, the client 3.4%. On imaging and the integrated OR, the field bought nothing at all in 90 days — so the creative went there.
Global surgical-technology manufacturer · Congress launch · Paid media
Client, handles & spend anonymized — under NDA
Studies fired
15→1
Parallel, merged into one read.
Propositions
21
Only for capabilities the client owns.
Creatives built
108
36 per pillar — 12 upper, 12 mid, 12 lower funnel.
Figma layers
531inst.
Plus 402 text layers, in the client’s file.
Inside the whitespace read

Four territories. The two thin bars are the opportunity.

Brand versus category share of paid output. Buy where the brand is quiet and the field loud, or where nobody is.

Territory AWhitespace

Safety & standards

The largest gap, and the one the campaign took.

Brand 3.4% vs field 28.8%NDA
Territory BEmerging

Imaging & the integrated OR

No buys in 90 days and no proof yet. A test, not a bet.

Field 0.0% in 90 daysNDA
Territory CCrowded

Training access

Over-served by the brand itself. Defend, do not expand.

Brand 50.8% vs field 17.2%NDA
Territory DCrowded

Procedure literacy

Loud, but in a space the category has left.

Brand 45.8% vs field 2.5%NDA
Built for how brand teams work

One engine. Both sides of the room.

For brand integrity or media efficiency, SOMIN works where the decision is made.

For brand & creative teams

On brand because it uses your components.

Built on your real Figma components: your brand, not a model’s impression of it.

  • Your components, type and licensed photography
  • Wordmark composited, never generated
  • Palette and copy density locked by prompt suffix
  • Optional — only with your Figma link
For performance & media teams

Validated before the money moves.

Every frame is scored against the category corpus; only passing frames reach the live ad set.

  • Predicted clickability with a confidence figure
  • Class and confidence both required
  • Pushed to the aspect-matched Meta ad set
  • Aligned with Meta & Google ad-diversity rules
“Creative is the most important lever for performance in the modern advertising ecosystem. To reach diverse audiences, brands must shift from granular manual targeting to creative differentiation — letting machine learning find the right customer through the right visual message.”
— Meta for Business
Beyond human-in-the-loop

SOMIN is agentic AI-ready.

After typically 60–90 days of signal, parts of the workflow can run autonomously, with SOMIN as Strategy Agent via MCP.

Orchestrated through Claude Agent SDK Manus Zapier Agents

Rated by users on G2 and Capterra

Brand and growth teams on SOMIN.

G2★★★★★★★★★★4.5Capterra★★★★★★★★★★5.0
★★★★★★★★★★
“SOMIN replaces guesswork with grounded, data-backed direction…”
Katerina S.Director · Consumer goodsVerified review on Capterra (invited by SOMIN) · May 2026
★★★★★★★★★★
“What I value most is that the insights are data-backed, not speculative.”
Angeline N.Director, GTM consultancy · Small businessValidated review on G2 · Oct 2026
★★★★★★★★★★
“I really appreciate how accurate and precise the data they gather is…”
Wai Yeong L.Mid-marketValidated review on G2 · Jul 2026

Short excerpts from validated reviews, quoted as written. Read every review in full on G2 and Capterra.

Peer-reviewed research

Built by seasoned marketers and an AI professor.

SOMIN's co-founder, Prof. Aleks Farseev, publishes research on audience profiling, personality-driven content and explainable AI for advertising at AAAI, ACM Multimedia, SIGIR, WSDM and The Web Conference. Those models run inside SOMIN, next to playbooks from marketers who work with brand and agency teams every day.

AAAIACM MMSIGIRWSDMWWWACM TOIS
ACM TOIS · 2025Fusing Predictive and Large Language Models for Actionable Recommendations in Creative MarketingPairing ad-performance prediction models with LLMs to turn creative scores into recommendations marketers can act onJournal article · read the paperAAAI · 2017TweetFit: Fusing Multiple Social Media and Sensor Data for Wellness Profile LearningCombining social media posts with wearable-sensor data to learn users' wellness profilesFull paper · read the paperACM MM · 2025SOMIN: An Explainable AI and LLM Platform for Real-Time, Data-Driven Digital Marketing StrategyThe SOMIN platform: a performance-prediction model plus an LLM that recommends marketing strategy in real timeIndustry demo · read the paper
Watch the talksIndustry expert talk · 2025Will AI Make Agencies Obsolete? Rethinking the Future of AdvertisingKeynote · 2025The Future of Generative AI in Marketing CommunicationsLecture · 2025INSEAD: How AI Cut Marketing Costs for a Beef Brand

All papers and talksGoogle Scholar profile

Common questions

“We already have AI creative.”

Most brands do. What is missing is knowing which to run.

How is this different from a generative creative tool?
They produce options. SOMIN decides what those say and whether they beat the category baseline; most tools cover only production, one of four steps.
Will it stay on brand?
Yes. Frames use your Figma components, licensed photography and a composited wordmark, never a generated identity; a locked prompt suffix fixes palette, type and copy density.
Do we have to give you our Figma file?
No. It is optional; without it you get the strategy, propositions and scored expressions and build the final frames yourself.
What does “scored” actually mean?
Each creative gets a predicted clickability class against the category corpus and a confidence figure. It passes only if the class is above average and confidence reaches 0.80 — a check most pre-flight scoring skips.
How fast does a team adopt it?
There is nothing to learn or prepare: name the brand, moment and market, and the first whitespace read returns in minutes.

See it run on your category.

A 30-minute walkthrough on your brand, competitors and next media moment, plus an optional live whitespace read.

Book a brand demo Talk to sales
30 min · Live demo · No obligation
You found the secret button! As a reward for your curiosity, enjoy 14 days of SOMIN on us. Just click and drop your details in the form below.