Platform Upgrades –
May 2026 🚀

Added: Create ready-to-launch promotions by prompt.

Agents can now be created inside the AI Expert to generate recurring promotions via push notifications based on natural language requests.

The agent builds the promotion setup, including segment, trigger and content, and presents it for review and approval.

Examples include:

‘Create a daily push notification promotion with yesterday’s most talked-about stock.’
‘Create a daily push notification promotion with today’s top football match.’

Once approved, the agent creates the recurring push notification promotion, ready to launch.

Added: Email delivery for AI Expert analysis tasks.

The AI Expert can now send email summaries of analysis results.

Analysis requests can include data trends, campaign performance, user behaviour or other business questions. Once the analysis is complete, an email is generated with key findings, insights and recommended next steps.

This capability can be used directly within the AI Expert or as part of an agent configuration for scheduled or recurring analysis.

This enables analysis results to be delivered in a clear format without manually reviewing raw data or returning to the platform to check the outcome.

Added: Starter questions for new chats.

The AI Expert now displays suggested starter questions when opening a new chat.

Starter questions are adjusted based on the brand’s vertical, helping conversations begin faster and providing more relevant guidance from the agent.

This makes it easier to start working with the AI Expert without beginning from a blank prompt.

Added: Web search capabilities.

The AI Expert can now access web search when needed to answer questions or retrieve external information, such as upcoming games or asset rates.

This expands the AI Expert’s ability to support broader research, content creation and analysis beyond the data available within the platform.

Agent performance has been improved through more focused context handling.

Agents are now treated as distinct entities, allowing each to use only the context required to complete its task.

This reduces unnecessary processing and improves response speed, accuracy and overall efficiency.

Handling of temporary network issues has been improved. When a network interruption occurs before a response is completed, the system now automatically retries and continues the conversation.

This reduces disruption and ensures responses are delivered more reliably.

Churn and loyalty prediction models have been enhanced.

Churn prediction now supports 14, 30 and 90 day periods with improved accuracy.

Loyalty prediction now reflects expected user retention over the upcoming 90 days.

All models are available when creating segments, enabling targeted campaigns based on predicted customer behaviour.

Local time zone handling has been improved.

When a scheduled send time has already passed in a user’s local time zone, it is now possible to choose whether to send the content immediately or skip delivery for those users.

This provides better control over time-sensitive campaigns and helps prevent unintended deliveries across different local time zones.

The Visitor Activation module has been improved to recognise bots.

The Web SDK now identifies bots and automated traffic, ensuring communication is limited to real website visitors (non-registered users) and improving overall data quality.

The segment report has been enhanced with clearer sector separation and new activity reports.

The report now provides improved separation between sectors, making it easier to analyse events and understand performance.

The report interface has also been updated, and dedicated reports for trading activity and betting activity have been added.