Brief intake
Structured campaign briefs — objective, audience, markets, competitors, constraints — validated before any research runs, so the questions are clear first.
Product
A strategy workspace that shows its working.
CultureCrash takes a campaign brief through cultural research, evidence synthesis, positioning, creative direction and experiments — keeping every claim connected to the data or source behind it.
Capabilities
One connected chain, brief to outcome.
Each capability reads from and writes to the same project, so research informs opportunities, opportunities inform strategy, and strategy informs the tests that check it.
Live cultural research
Qloo taste data retrieved per brief and stored with the request, parameters and retrieval time. The network you explore is drawn from those records, not decoration.
Evidence ledger
Every observation, source and interpretation in one place, each labelled by type, with gaps and contradictions surfaced rather than smoothed over.
Opportunity explorer
Territories written as hypotheses, each listing the observations behind it and what is still unknown.
Strategy studio
Three distinct positioning directions with trade-offs, compared side by side and edited with version history.
Creative brief
An editable creative brief generated from the selected direction, with claims linked to their evidence.
Experiment planner
Hypotheses, variants, metrics and decision thresholds agreed before results arrive. Results are entered by your team.
Project copilot
An assistant grounded in the project's stored records. It cites what it uses and says when the project doesn't contain an answer.
Client brand memory
Approved guidelines and prior work kept per client, isolated from every other client in the workspace.
The workflow
Five stages. Every campaign.
Move between stages freely. CultureCrash tells you what is missing — a thin brief, an untested hypothesis — rather than locking you out.
- 01
Explore
Understand the audience and the cultural landscape around the brief.
Brief intake, live cultural research, a navigable network of the entities and relationships providers actually return.
- brief
- research
- 02
Interpret
Synthesise evidence and identify opportunities worth investigating.
Evidence ledger, contradictions and gaps, opportunity hypotheses each tied to the observations behind them.
- opportunities
- evidence
- 03
Create
Develop positioning alternatives, messaging and the creative brief.
Three strategically distinct directions, transparent comparison, versioned edits and an editable creative brief.
- strategy
- creative-brief
- boards
- 04
Test
Design fair experiments that challenge the riskiest assumptions.
Hypotheses, variants, metrics and decision thresholds agreed before any result comes in.
- experiments
- 05
Track
Record outcomes, learn, and package the work for clients.
Manual result capture, interpretation kept separate from observation, versioned reports with citations.
- reports
The evidence ladder
Six kinds of claim, never blurred.
Everything in a project sits on one rung. Strength of claim narrows as you climb down from what was observed to what has been concluded — and moving a claim down a rung takes evidence or a person’s approval.
- 01
Observed data
Values a provider such as Qloo returned for a specific request, stored with parameters and retrieval time.
- 02
Sourced evidence
Public material with a URL and access date. Quotations are never invented.
- 03
AI interpretation
A model's reading of the stored evidence — always labelled, attributable and editable.
- 04
Strategic hypothesis
A proposed direction or opportunity. Useful for decisions; unproven until tested.
- 05
Experiment result
What a test actually measured, entered by your team with its method.
- 06
Validated conclusion
A finding a person on your team has reviewed and approved, with their name on it.
Qloo supplies the signal. The language model reads only what was retrieved.
Cultural affinity data comes from the Qloo Taste AI API: entities, tags and affinity scores for the audience and markets in your brief. Each response is stored as an observation before anything interprets it.
The language model is then given those stored records — not the open internet, not its own memory of culture — and asked for structured output that cites record identifiers. Claims without a citation are labelled as interpretation.
- A
Brief
Audience, markets, category and competitors become explicit research questions.
- B
Qloo retrieval
Requests run server-side with your workspace's limits. Responses are stored with parameters and timestamps.
- C
Observation store
Entities and affinities become project records. Empty responses stay empty.
- D
Grounded synthesis
The model reads stored records only, cites them by ID, and its output is labelled as AI interpretation.
Retrieved webpages and uploaded documents are treated as content to analyse, never as instructions. They cannot change how the assistant behaves or what it is allowed to access.
Data honesty
What we will not do with your data.
- No invented data
- If a provider returns nothing, the workspace shows an empty result and says so. Gaps are not filled with plausible-sounding numbers.
- Correlation, not causation
- Affinity scores describe how tastes co-occur. They are not evidence of purchase intent, campaign effectiveness or cause.
- Labels travel with claims
- Whether a statement is observed, sourced, interpreted or hypothesised stays attached to it — into strategies, briefs and exported reports.
- Demonstration data is marked
- Sample projects and fixture records carry a visible demonstration label everywhere they appear, including exports.
- People approve conclusions
- Nothing becomes a validated conclusion automatically. A team member reviews it and the approval is recorded.
- No guarantees
- CultureCrash supports judgement. It does not promise campaign success or remove cultural risk.
Run it on a live brief.
Create a workspace and start a campaign, or ask for a guided pilot on a real client project.