Change management when you have to disrupt yourself: Aydin Mirzaee on pivoting Fellow to AI

Change management is usually about persuading an organization to fix something broken. This is the harder version: persuading an organization to abandon a product that has genuine product-market fit, paying customers and healthy renewals, because of where the market is going rather than where it is.

Aydin Mirzaee is co-founder and CEO of Fellow, returning roughly a year after his first appearance. In between, the company pivoted from a meeting productivity platform into an AI meeting assistant, which he describes as transforming into an AI-first company.

Michael has used the product for five years, which makes this an unusually well-informed conversation about what actually changed.

Why they had to

Mirzaee's reasoning is stark and worth taking seriously even if your business looks nothing like his.

He believes that had they not completely transformed the company and the product, they would have become extinct.

His general claim: AI will affect every company, in waves. Meetings and AI happen to be an obvious pairing, which put Fellow in an early wave and let them get ahead of it. But he thinks the transformation they went through is one every company will eventually need.

His illustration comes from his own history. His previous company was sold to SurveyMonkey, and he thinks even something as established as surveys is due for disruption, pointing at companies building AI interviewers that conduct long, deep voice conversations at scale, going several levels into why someone wants what they want.

Everything that has to change

Michael draws an important distinction first. In technology, pivot usually means something was not working. Fellow had demand and product-market fit, and made this move on a judgment about where the puck was going.

Mirzaee's answer to why that is hard is a list, and the list is the point.

The product needs to change, which he calls the easy part because it is obvious. Then the onboarding flows. Then pricing and packaging. Then how you service customers. Then the sales team structure. Then how you describe yourself on your website. Then how customer success works. Then how the teams are organized.

And then the genuinely hard part. Going back to existing customers and telling them that the thing they bought is not what they should use, that they should move to the new thing, that it is better but they will have to be convinced of that, and that it costs more because it is now powered by AI.

For the new Fellow to be born, the old Fellow must die

The internal language was deliberately severe.

Mirzaee describes a pivotal moment where they brought everybody together and set out where the market was heading, with the explicit framing that if they did not change they would go extinct. His analogy: being in the phone book business as the internet arrives.

Then they ran a hackathon, focused the development teams on building with AI, and shipped a beta.

The sentence they used internally, repeatedly, was that for the new Fellow to be born, the old Fellow must die.

His explanation for the severity is the useful part. The purpose was to jolt people out of thinking in terms of how things used to be done and which use cases used to be supported, and toward what AI makes newly possible.

His framing of what AI actually is, organizationally: a new set of building blocks. You are still solving the same problems. But now that these blocks exist, not using them is close to irrational, because the next company to start certainly will.

How they tested it

Michael asks how you validate a new direction when you already have satisfied customers.

Mirzaee did talk to customers, and got real pull. Forward-thinking ones were asking whether Fellow could suggest what to discuss in the next meeting given everything discussed before, or generate feedback for a direct report given a year of one-on-one notes, or surface the top three feature requests across every customer conversation.

But he names the limitation honestly: information from existing customers is weighted toward what they already know about you. It starts from a base level rather than from what is possible.

So alongside MVPs tested with customers, they made a structural decision. Every new customer coming into the product got the new Fellow.

Legacy customers required real work. Some migrated themselves. Others needed customer success teams to engage and show them how their current usage compared to what was now possible: why would you write something down yourself, why would you track someone else's action items, when the product can do it.

His observation about why this is counterintuitively difficult is worth keeping. Once someone is using a product, it is genuinely hard to get their attention for something new, because people develop tunnel vision around how they already use it.

His supporting anecdote is a familiar one from Microsoft Word, where customers asked for a list of features and the overwhelming majority already existed and nobody knew. Products that have been around a while accumulate inertia that takes effort to overcome.

Retiring a value to unblock the change

The most transferable idea in the episode is that they changed a company value to make the pivot possible.

Fellow had a value called Pace Quickly, reflecting that speed and urgency matter in startups. They archived it and replaced it with Fast and Fearless.

Both halves of the change did work.

Fast, because quickly was no longer fast enough given how rapidly the AI landscape was moving.

And fearless, which Mirzaee identifies as the real constraint. A great deal of the slowness came from fear of making these changes. What if this upsets legacy customers? What if it does not work?

His example of a decision that required the new value: putting unlimited AI on every paid plan, which is a large commitment when you do not know how much people will use or what it will cost. The instinct is to approach it slowly, put it on the highest tier first, and see what happens.

His diagnosis of where that instinct comes from is precise and applies to any established business. You become methodical. Every decision acquires multiple stakeholders. Things get done incrementally. Which is reasonable, until the world shifts drastically, at which point incrementalism stops working and you have to make big bets.

The counterweight he gives people: most decisions are reversible and most things can be backtracked.

And the reinforcement mechanism matters as much as the value itself. Every time someone made a bold decision, they called it out at the town hall as the Fast and Fearless value in play.

The communication structure underneath the speed

Michael's observation is that this makes internal communication vital, and often overlooked at companies that have not reached this operational maturity.

Mirzaee's cadence, as a fully remote company:

Weekly town halls, covering what is happening, wins across the company, and new product demos. Given how much is shipping, this is where marketing learns what needs updating, help center articles get flagged, comparison pages get revised, and the sales team learns the new language.

A weekly email from him to everyone, covering what is on his mind, how he is thinking about things, and important anecdotes.

Monthly business reviews, where people from across the company, not only executives, present their objectives, where they are struggling and where they are succeeding, and the executive team coaches on how to tackle it.

Quarterly business reviews covering the whole quarter.

Twice-yearly in-person gatherings for everyone, focused on vision and direction.

His argument for why the structure is load-bearing: if you want people to execute at ground level without seeking approval from a hundred others, and you want to move at that pace, you need communication strong enough that everyone knows what is happening.

The 5 things I took away from this conversation

1. Pivoting a working product is harder than pivoting a failing one. There is no crisis to point at. Aydin's list of everything that has to change alongside the product, onboarding, pricing, sales structure, positioning, team organization, is the honest inventory most pivot conversations skip.

2. Change the value that is blocking you. Retiring Pace Quickly for Fast and Fearless is the sharpest move here. They correctly diagnosed that the constraint was not speed, it was fear, and they changed the stated value to name the thing actually holding people back.

3. Then celebrate every instance of it out loud. A new value on a page does nothing. Calling out each bold decision at the town hall as that value in action is what converts a word into a behavior people can copy.

4. Route new customers to the new thing immediately. Rather than migrating everyone at once, every new customer got the new product from day one, while customer success worked existing accounts individually. That splits an impossible transition into a default and a project.

5. Existing customers tell you about the product they already know. Aydin's caution on customer research is the one I will remember. The feedback is real and it is anchored to your current shape, which means it will never point at the thing that replaces you.

FAQ

What is change management in a company pivot? The work of moving an organization from one operating model to another, including product, pricing, sales structure, customer communication and internal culture. Mirzaee's case is a pivot made from strength rather than failure, which removes the obvious burning platform and makes the persuasion harder.

How do you pivot a product that already has product-market fit? Fellow's approach combined a clear internal case that the market was moving, a hackathon to build the new capability, routing all new customers to the new product by default, and a deliberate customer success effort to migrate existing accounts one at a time.

How do you get a team to embrace a difficult change? Mirzaee's answer is largely cultural rather than tactical. They replaced a company value that was no longer serving them, named fear explicitly as the constraint, reminded people that most decisions are reversible, and publicly celebrated each bold decision as an example of the new value.

Why is customer research misleading during a pivot? Because existing customers answer in terms of the product they already know. Their input is genuinely useful and it starts from your current shape, so it will tend to suggest improvements rather than the replacement that a new entrant would build.

What communication cadence supports a fast-moving company? Fellow runs weekly town halls, a weekly all-company email from the CEO, monthly business reviews including non-executives, quarterly business reviews, and twice-yearly in-person gatherings. Mirzaee's argument is that autonomy at speed requires the communication structure to carry the context.

Also mentioned

• Fellow, and its transformation from meeting productivity platform to AI meeting assistant
• SurveyMonkey, which acquired Mirzaee's previous company, and the survey category he expects to be disrupted next
• The Microsoft Word feature-request anecdote, on product inertia and what customers do not know they have
• Pace Quickly and Fast and Fearless, the values retired and introduced to enable the pivot

Listen to the full episode

Aydin Mirzaee on Between Two COO's

Between Two COO's is hosted by Michael Koenig. Subscribe on Apple Podcasts, Spotify, or wherever you listen.