Rohit Begani on building AI that protects your judgment

What this conversation is about

Glenn Llopis has been saying it for a while:

"AI is the accelerant of standardization." — Glenn Llopis

It has never been easier to sound, think and lead like everyone else. Rohit Begani is building a product meant to push the other way. He is chief technology officer and co-founder of AceUp, where his team built Ally, an AI coaching platform designed to put leadership development into the flow of work for people who would never have had an executive coach.

What caught Glenn’s attention was a failure. Ally was first built as a pure coach that asked reflective questions and made people do the thinking, and people rejected it. They wanted the answer. Glenn’s read on that is pointed:

"That's not the technology's problem. That's a compliance problem" — Glenn Llopis

You can read more about this conversation in Glenn’s Forbes piece Will You Shape AI Or Will AI Shape You? (August 27, 2026).

The coach that asked too many questions

Begani explains the original design choice. Most AI is judged on its output: how fast and how accurately it hands you something. His team deliberately refused that:

"Ally cannot give you an output very quickly. It cannot just give you a decision, because then it will continue feeding into this standardization problem that you were mentioning earlier." — Rohit Begani

Coaches loved it in testing. Users found it exhausting, because an AI that makes you stop and think feels like more work. The lesson he took was not to give up on reflection but to stop imitating a human session:

"just asking questions and replicating what a human coach does is not going to work with AI." — Rohit Begani

The fix was rhythm: ask a reflective question, then give something back, then ask again. Ally watches where each person drops off and tunes itself, and it added role-play, so a manager can rehearse a hard feedback conversation with Ally playing a direct report who sometimes pushes back.

"The balance is not a standardized approach that we're using — it's personalized to the individual." — Rohit Begani

Standardize the framework, personalize the person

Glenn asked the question at the center of his own work: where do you make people the same on purpose, and where do you protect the difference? Begani’s answer is precise:

"the standardization is not at the people level, it's at the framework level" — Rohit Begani

Ally’s constitution is the International Coaching Federation standard, tested by coaches over many years. What personalizes it is memory: what it learns about each person over time, the way a good coach remembers what matters to you rather than treating each conversation as a transaction. It will never mandate an action. It also looks beyond the individual. If someone is working on being more decisive but their leader runs everything by consensus, coaching the individual alone will not work; the environment has to be understood too.

He is clear that none of this replaces people:

"I actually do not foresee — or even pursue — a world where AI coaching replaces human coaching." — Rohit Begani

AI brings continuity, presence in the moment and scale. Humans bring empathy, intuition and credibility for the consequential decisions. In AceUp’s model, a human coach meets someone every couple of weeks and Ally carries the thread in between, checking in on what was agreed. With the participant’s permission it joins sessions, and the human coach reviews and edits its notes before anything is used.

Your story is not the reality

One design decision says a great deal about how Begani thinks. Ally never uses what one person tells it about a colleague in its conversations with that colleague:

"what I share with Ally is my story of what's happening. It's not the reality. It's my story." — Rohit Begani

Perhaps the report who never updates you was never asked to. Carrying one person’s story into another’s coaching would steer them wrong, and it would destroy the psychological safety that lets anyone be honest with it. Control and transparency over data came first for the same reason: people fear a tool their employer deploys, and trust is earned by showing them exactly what happens to what they share.

He tells a story against himself, too. When a priority he had set with Ally slid for weeks, it checked in, and he chose not to answer because he did not want to be held accountable in that moment. It came back a week later, offered to drop the goal if it no longer mattered, and he rebuilt the plan. The measure of success he now trusts is not satisfaction scores but whether people keep committing to actions and following through.

What has to be true for the hopeful version to win

Glenn asked him to look ten years out. If coaching stops being an executive perk and reaches everyone, does it help people reclaim who they are, or just make them faster copies of everyone else? Begani is a believer in what AI produces; he uses it every day. His worry is what it takes away if nothing pushes back, and that is the line at the top of this page.

Intelligence, in his view, is becoming a commodity. What matters is the framework and the memory brought to a specific outcome, and he is explicit about which outcome his team is building for:

"that outcome is helping people make judgment, make decisions, and have clarity in their own convictions." — Rohit Begani