New Possible: Action Planning

Action Planning will be provided as a feature within New Possible, a people insight platform that helps organisations understand employee feedback and drive meaningful organisational change. The feature is intended to support clients in identifying, prioritising, and implementing actions based on survey results. Users will select a survey and may further segment responses by factors such as region, department, or team. The system will analyse quantitative survey data alongside qualitative free-text feedback using statistical techniques and AI-assisted text analysis. Based on these insights, the system may generate contextual questions to gather additional information from the user about their organisational context, priorities, and constraints. The responses provided by the user are then combined with the survey analysis to generate tailored recommendations. These recommendations are designed to help organisations develop action plans that address key themes identified within survey results and support continuous improvement following a survey.

Action Planning is not intended to be used for the following purposes: - Making or supporting employment-related decisions about individuals, including recruitment, promotion, performance management, disciplinary action, remuneration, or termination decisions. - Providing legal, financial, psychological, or any other form of regulated professional advice. - Producing definitive conclusions, instructions, or recommendations that must be followed without the application of human judgement and professional discretion. The system is intended solely as a decision-support tool. All outputs are advisory in nature, and responsibility for reviewing, selecting, prioritising, and implementing actions remains with the user and their organisation.

Action Planning uses survey data and user-provided contextual information to generate tailored actions. Data sources used during inference include: - Quantitative survey data such as ratings and response distributions. - Qualitative survey data, specifically free-text feedback provided by respondents. - Segmentation data, such as department, team, region, or other organisational groupings selected by the user for analysis. - Organisation details, including size and sector. - User-provided contextual information, collected through follow-up questions generated by the AI system. We have implemented the following safeguards: - Personally identifiable information (PII) is removed prior to contextual question generation and action planning. - Anonymity thresholds automatically apply. - Demographic questions are automatically excluded from analysis. - AI analysis and rationale is retained for audit and evaluation purposes. - Client data is not used to train the model. - All data is stored and processed within the EU, in New Possible's Microsoft Azure environment. The system uses a staged AI workflow comprising theme analysis, PII removal, contextual question generation and action recommendation generation.

Not provided

The system is intended solely as a decision-support tool. All outputs are advisory in nature, and responsibility for reviewing, selecting, prioritising, and implementing actions remains with the user and their organisation. New Possible will evaluate the performance of the AI system in the following ways: - % of actions saved vs ignored: Users can choose to save, edit, or ignore any AI-generated actions. Ignored and edited actions will be monitored to help identify trends and improve the quality and relevance of future outputs. - AI-generated actions will be regularly benchmarked against human analysis. This comparison will be used to help evaluate the relevance of generated actions, and monitor AI drift. - User satisfaction, including anecdotal feedback and surveys.

New Possible operates a proportionate AI governance framework appropriate to the size and risk profile of the organisation. Responsibility for this AI system sits with the Founder and product owner. New AI capabilities undergo documented risk assessment prior to release, including consideration of privacy, security, bias and accuracy risks. AI outputs are advisory only and remain subject to human review. Personal information is removed prior to analysis, minimum respondent thresholds are applied to protect anonymity, and demographic questions are automatically excluded. Significant product changes are tested and documented before deployment.