This was an attempt to switch from ChatGPT to Google Gemini and consolidate everything into one system. After two months, the decision went in the opposite direction.
The Moment
The idea started from a practical place: I was already paying for Google services and at the same time relying heavily on ChatGPT for work, thinking, writing, and daily operations, which created a sense of duplication — two subscriptions, two environments, two systems growing in parallel — and I wanted to reduce that into one simpler setup, ideally inside the Google ecosystem, where I was already using Google Drive for some backups and considering expanding into documents, calendar, and other services.
The Experiment
I started a Google Gemini Pro trial and used it gradually for about two months, creating a few “gems” (equivalent of Chat GPT projects), running conversations, and testing whether I could realistically move my workflows there, while also briefly considering a broader shift — including moving deeper into Google’s ecosystem and even switching from iPhone to a Google Pixel device — since Gemini was the central piece that could make that transition coherent.
Adaptation / Friction
The issues didn’t appear as one clear failure but as repeated small frictions in daily use: while I could access past conversations on iPad, they were not grouped by gems, which made it difficult to navigate and continue specific contexts; voice input interrupted my natural pauses, forcing me to either rush or use a separate transcription tool; Gemini tended to respond with expanded answers, additional suggestions, and follow-up questions even when I needed a minimal response, which added cognitive overhead; and although document integration existed, in practice editing longer or precise documents required multiple iterations and verification, making manual editing faster and more reliable.
Observations
After this period, a consistent pattern emerged: nothing was fundamentally broken, but several small mismatches accumulated — lack of structured continuity around gems, limited tolerance for thinking pauses during dictation, a tendency toward overly verbose interaction, and reduced efficiency when working with documents — and while each of these issues on its own was minor, together they created a noticeable shift in how the tool felt in daily use; I could have adapted my workflows, but it was clear that those adapted workflows would be less efficient and less natural than what I already had, which made the trade-off visible. In parallel, I briefly tested the broader ecosystem shift on an Android device and considered moving to a Google Pixel, but this reinforced the same pattern: the experience felt less cohesive than what I am used to on iPhone and iPad, and moving away from an iPad-centered workflow would have been a step back for me; as a result, the advantages (integration with Google services, presence across tools, and the potential to unify my setup, including even the device ecosystem) remained more theoretical than practically useful in my daily workflow.
Conclusion / Current State
I decided not to switch to Google Gemini as my main assistant, closing this experiment as a completed evaluation rather than a failed attempt, and accepting a different trade-off: staying with ChatGPT despite its slower development in some areas and weaker integrations (compared to Gemini but also other LLMs), while continuing to build my own workflows based on how I work, instead of reorganizing everything around newer or more integrated tools, and for now choosing stability over switching.