85% of organisations increased AI spending last year. Engagement, NPS and loyalty — the metrics that matter before revenue — are barely moving. The problem is not the technology. It is what the technology is being fed.
Modules, outcomes, and long-term co-creation: a new AI business model. The Palantir model for everyone — co-creating AI value at any scale.
Transaction data tells an agent what a customer did. Mindset data tells it who they are. Without that second layer, even the most capable AI agents are making decisions in the dark.
Most organisations are using expensive frontier models for tasks that don't need them. ACPU — AI Cost Per User — is the metric that exposes the waste, and a pre-agentic data strategy is what fixes it.
Badges and points are not engagement — they are decoration. The real opportunity is using small, well-timed interactions to help customers understand themselves, express regret, and feel genuinely known by the services they use every day.
Push notifications are the SMS of the 1990s — mature technology used well below its potential. Every institution already has the channel, the permission, and near-zero cost. What's missing is interactivity and context.
When AI analyses the what and never the who, personal data never enters the system. Faster deployments, simpler legal reviews, DPOs who say yes. Privacy by design is not a constraint — it is the smartest architecture decision any regulated industry can make.
Banks, telcos, utilities and streaming platforms share the same data quality problem. 0.5 party data captures what zero and first party miss: honest context at the moment it matters, building the behavioural foundation AI agents will need.