Is remote work inherently bad?

No. Remote work performs very well in companies in stable execution mode — with mature processes, large teams, and low need for collective creativity. For a company in AI First transition (radical model change), in-person accelerates decisions. These are different contexts.

Does Fhinck not trust remote workers?

Trust is not the issue. Speed of decision-making under radical ambiguity is the issue. In an AI First transition, dozens of difficult decisions arise each week. In-person converges faster. Remote adds latency.

What about the right to disconnect?

In-person work does not negate the right to disconnect. On the contrary — when the workday ends, it ends visually. In remote work, the boundary becomes diffuse and burnout increases. Fhinck manages disconnection protection through its Journey Assistant.

Can large companies do the same?

Nubank, Amazon, Apple, Tesla, JPMorgan, Goldman Sachs, and several others did so in 2024-2026. It works. The cost is turnover from those who do not accept it; the gain is institutional speed. It is a strategic decision — there is no absolute right answer.

How should the transition from remote to in-person be managed?

Direct and advance communication (3-6 months), a clear rationale (what changes in the business), support for the transition (commuting costs, infrastructure assistance), and accepting some team attrition as the cost of the decision. There is no painless path.

Why We Ended Remote Work at Fhinck 2 Years Ago (and Why Other Companies Are Returning Now)

In 2023, against the prevailing current, Fhinck ended remote work. In 2025-2026, Nubank, Amazon, Apple, JPMorgan, Goldman Sachs, and many others followed the same path. Why this decision is particularly critical for an AI First company.

"Many companies are reaping good results today, but will not survive the next 5 years because they did not build the right foundation. (...) When I see movements like those from Nubank, Amazon, and other companies returning to in-person work, I recognize that we made the right decision at Fhinck when we chose to end remote work two years ago."

— Paulo Castello, November 2025

The Unpopular Decision of 2023

In 2023, any decision by a tech company to end remote work was controversial. The pandemic had "proven" that remote works. Several companies competed on the basis of "we are 100% remote." Good talent chose employers based on remote work policy.

In that context, Fhinck ended remote work. It took criticism. It lost part of the team that did not accept it. It was accused of being "retrograde."

In retrospect, it was the right strategic decision — not for ideological reasons, but for one very specific reason: the company was entering a deep AI First transition.

Why AI First Transition Requires Physical Presence

In stable operation mode (a company with mature processes, large teams, executing a known routine), remote work performs very well. Remote teams consistently deliver.

In deep transformation mode (a company redesigning its operational model, making weighty strategic decisions weekly, integrating new technology at an accelerated pace), decision-making speed under radical ambiguity is the bottleneck.

In-person work addresses that bottleneck in four ways:

1. Parallel Conversations Become Decision-Making

In the office, two Fhinckers cross paths at the coffee machine, quickly discuss a problem, and close a decision in 10 minutes. The same decision in a remote setting: meeting request, scheduling, context recap, decision reached in 3 days.

In an AI First transition, with dozens of decisions per week, this speed difference becomes a brutal competitive differentiator.

2. Sharpening the Axe Culture Happens at Greater Depth

An entire Friday in a classroom together is a different experience from watching a Zoom. Collective discussion, parallel hands-on work, observing each other's work — all of it is amplified in person.

3. Onboarding New Fhinckers Is Dramatically Better

Someone joining Fhinck needs to absorb a specific culture, an intense pace, and radical autonomy. By osmotic presence this happens in weeks. Remotely, it takes months — and sometimes never fully happens.

4. Difficult Conversations Happen Better In-Person

In an AI First transition, several hard conversations arise: someone who is not keeping up, a role redesign, eventually a termination. In-person preserves the relationship and dignity. Remote degrades both.

What Changed in the Market in 2024-2026

Fhinck's decision became a trend:

Amazon (Sept/2024): Andy Jassy mandated an end to remote work, citing the need to be "more agile, like a startup," reduce bureaucracy, and cut down on excessive meetings.
Apple (2024): return to 3 days in-person, evolving to 5 days in many areas.
JPMorgan, Goldman Sachs (2024-2025): return to 5 days in-person for directors and senior staff.
Tesla (always): never adopted remote work, position reinforced.
Nubank (2025): reduction of remote work, more in-person for product and core engineering teams.
Google, Meta, Microsoft (gradual): 3 days in-person, evolving toward more.

The market message in 2026: companies that intend to lead in AI are consolidating in-person work as the standard, with remote as a limited exception.

Why? The same reason that made sense at Fhinck in 2023: decision-making speed under radical ambiguity.

How Paulo Responded in the CBN Interview on the Topic

In January 2026, in a CBN interview about Itaú's mass layoffs (related to this transition), Paulo positioned:

Layoffs are not the end of the world; they are part of market dynamics
Those who reinvent themselves, prosper
Companies that continue making work model decisions (in-person vs. remote) based on short-term comfort will struggle
Companies that decide based on transformation velocity win

It was an unpopular position, but aligned with what the market was showing.

The Most Common Objection (and the Answer)

"But won't in-person work limit our talent pool?"

Yes, in part. People who strongly value total flexibility will not choose an in-person company.

And? The choice for an AI First company in 2026 is not to maximize the "total talent pool." It is to maximize the density of talent that accepts radical transformation pace. Those two sets are different.

Fhinck lost people who valued flexibility. It gained people who valued being at the center of building Brazil's first AI First company. A conscious trade-off.

What This Article Is NOT Arguing

This piece is not saying:

That remote work is inherently bad (it is not — it works well in many contexts)
That companies doing remote are doomed (they are not — it depends on context and phase)
That remote professionals are less committed (they are not — they are different)
That the return to in-person should happen without support (it should not — it requires clear communication and transition time)

The piece is saying: in an AI First company undergoing deep transition, in-person accelerates what matters. This may change once the company reaches a stable operating mode.

Conclusion

The 2023 decision was unpopular at the time. By 2026, it had become a consolidated market trend.

Companies that make work model decisions based on comfort win only in the short term. Companies that decide based on what serves the transformation win in the long term.

Fhinck went through the AI First transition combining tough strategic decisions (in-person, headcount reductions, deep redesign) with humanity (transparency, support, dignity). If you want to discuss how to apply this in your context, schedule a conversation.