KWGMate

Our idea, KWGmate, is an AI teammate specialising in facilitating discussions amongst employees and with employers. It leverages prompt engineering from AI models to provide contextual opinions and timely communication between employee and employer. With KWGmate, companies will have a reliable team member that provides a grounding voice to steer conversations to the right direction, depending on their respective needs. Improving communication, strategy, and employee engagement Leverage frameworks like ADKAR and incorporate real-time feedback mechanisms to drive continuous improvement KWGmate splits communication into discussion and review. Prior to any discussion, KWGmate is given the context of the job. A checklist of information is equipped to serve as a colleague to employees or assistant to managers. This checkline includes updated information on the persona it should adopt, goals of the organisation and project team or department, timeline of goals to be met, manpower capabilities and the dynamics of the team. For instance, ‘You are part of a team of XXX, your goal is to XXX, you have existing knowledge of XXX eg. ADKAR framework. You are to give opinions from the perspective of someone prioritising the ADKAR framework’. During discussion, KWGmate uses a speech-to-text api to transcribe all the words spoken. It will save these as prompts, which will be fed to openai when generating opinions. To generate an opinion, KWGmate prompts openai with the context, prompts and a request for a contextual opinion. This response is then verbalised to the team. Contexts (what is currently being discussed, overall progress of team with respect to the goal etc. - this is being returned from openai) are being updated and displayed to the employee teammates. To improve communication between managers and employees, KWGmate allows contexts to be snapshotted. In review, managers can review these snapshots, much like a linked list, with each node containing a chronological snapshot of the discussion context. This keeps the manager aware of the ideas and train of thoughts that the discussion has gone through. At the end of discussions, KWGmate summarises the context and prompts and asks for 3 summaries from openai, focusing on 3 separate metrics (topic dependent, ai decide metrics by itself) that can be beneficial to the flow and goals of the discussion. These are sent to the manager for selection, after which KWGmate can reinforce the group’s working style using this information from the discussion and manager’s feedback. Managers are also able to provide real-time feedback by looking at the discussion checkpoints (context snap shots) in real time. They do not need to be physically part of the discussion, but they are always able to see where the discussion is headed to and send a comment to the team via KWGmate and provide feedback. Overall, managers can change organisational ‘culture’ using the targeted feedback from KWGmate, and provide targeted recommendations to each segment of the organisation based on the needs of the company and current capabilities of its employees.