Shubham Korde
← All projects

AI · LEARNING

Practice with Leo

Practice a consulting case with an interviewer that follows your questions and reviews your final recommendation.

Live Practice with Leo homepage showing an adaptive case interview and sample exchange about cold-chain unit economics.

Why I built it

When I was preparing for consulting, the practice that helped most was usually simple: a friend, a printed case and enough of an evening to work through it. The useful part was the conversation: the questions I asked, the assumptions I missed, and how I brought the answer together. Practice with Leo is my attempt to make more of that kind of practice available to anyone who wants it.

The interviewer follows your analysis

Pick a case and take it where your reasoning leads. Leo can follow a valid alternate path, answer focused questions, and push back when an assumption has outrun the evidence. The case library now has 119 scenarios across profitability, growth, operations, market entry, M&A and public-sector strategy. A focused session takes about 30 minutes.

The case controls its facts

Each case has a pool of facts. Leo reveals them in response to your questions. The runtime keeps those facts separate from the conversation, so the model can answer without giving away the diagnosis or inventing data. I spent the most time getting this boundary right: Leo can keep the discussion moving while the case runtime decides what information is available.

Finish with a recommendation

At the end, you have to make the call. Leo asks for the recommendation, the evidence behind it, the risks and the next steps. The debrief points back to your interview, showing where the analysis held together and where it needs another pass.

Try it

Practice with Leo is free. Pick a case that interests you and start with the question you want answered. You can work it out as you go.

Built with Python, FastAPI, LangChain, ElevenLabs.