Rehearsal AI
Interview practice, feedback and skill learning. I founded Rehearsal and lead its product development.
Examine the product work → Can agreed guidance settle the choice? No
Would getting it wrong have serious consequences? Yes
1 Consider a leave request that passes through several people before it reaches a manager. Much of that work consists of checking records and applying a policy.
2 An agent with authorised access can perform those checks. Once the process is reliable, routine steps can proceed within the authority already agreed.
3 Even an unusual request can follow a rule. If the policy defines when additional leave is allowed, the exception does not automatically need a manager’s judgement.
4 Two requests now qualify under that policy, but approving both would leave a critical service unstaffed. The policy provides no adequate way to resolve the conflict.
5 The manager must decide how to maintain cover while responding to the requests. This requires judgement because guidance cannot settle the choice, and getting it wrong has serious consequences.
6 The manager remains accountable while the agent handles the agreed routine work. Staffing and reporting lines can now be examined against the volume of human work that remains.
I am Dr. Shiva Kakkar. I hold a PhD from IIM Ahmedabad, am former faculty at XLRI and lead product at Rehearsal AI.
I work with leadership teams on AI strategy, managerial decisions and the redesign of work.
Interview practice, feedback and skill learning. I founded Rehearsal and lead its product development.
Examine the product work → A working worksheet for mapping what an agent can perform, which decisions require a person and who remains accountable.
My co-authored research in Business and Professional Communication Quarterly examines how learners use AI while developing their own judgement.
What an AI system needs to know depends on how the organisation records its work, decisions and responsibilities.