Feel the follow-up. Keep the feedback honest.

The transcript, the question and the evidence are the stage. No uncanny faces.

Answer, then inspect the feedback.

Choose a sample question. Select any dimension to see the exact evidence behind it.

Question

Design a system that ranks items on a home feed.

Answer, excerpt

I would train a gradient boosted model on click data and serve it behind an API. Retrieval first, then ranking. I would watch click-through rate.

Feedback by dimension. Select one to see the evidence.

Trade-offs. No mention of latency budget, freshness or feedback loops.

Focused retry generated from your weakest dimension: Trade-offs. The result updates your evidence map and next-best action.

Sample question and feedback. No microphone needed here.

We explain what the microphone is for before asking for access. Text works too.

Choose the role, company and process, or let SPHERO infer the likely rounds from what you provide.
Adaptive follow-ups based on the depth of your answer and what it left out.
Feedback by dimension: correctness, reasoning, trade-offs, structure, communication, seniority calibration.
A concrete snippet from your answer behind every feedback point.
A focused retry generated from your weakest dimension.
Your evidence map and next-best action update after every mock.