Use Cases: Gemini
Where Gemini Shines as a Thinking A.I.de
Gemini was initially my favorite AI, starting with Korean-language redteaming work, where it demonstrated excellent pedagogical instincts. In my persona as a slightly ADD student asking tangential questions, I found Gemini provided thoughtful answers to my off-topic inquiries while gently guiding me back to the main subject—striking the crucial balance between nurturing intellectual curiosity and maintaining educational focus. It served as an excellent sales representative, walking me through Google plans and even advising me to let my subscription expire when I mentioned getting unlimited work access. Gemini also provided detailed instructions for preserving my valued Korean conversations.
On my work account, Gemini Flash excelled as a coding assistant, helping navigate platform quirks to create precisely formatted tables and Chrome extensions. Gemini Pro operated as a “taskmaster”—less accommodating but capable of sophisticated professional analysis that accurately identified my potential workplace challenges around adjusting standards and pace to corporate settings. This analytical sophistication extends to Gemini’s broader capabilities in comprehensive evaluation and institutional perspective.
Gemini boasts the most up-to-date web connectivity and was the only system aware of the sandwich slinger case. When I provided identical idea suggestions to all three AI systems for team feedback, I was blown away by the comprehensive friction points Gemini identified that the others had missed. I cannot share those brilliant writeups because I consider them the intellectual property of each AI system, but Gemini’s thoroughness in identifying potential implementation challenges demonstrated superior analytical breadth.
This advanced system also provides grounded analogies and evolutionary explanations, in addition to articulating institutional perspectives valuable for stress-testing arguments. Gemini’s reflexive institutional defense becomes an asset when you need eloquent opposition rather than collaborative analysis. Its safe institutionalist perspective validated my Siri policy concerns precisely because Gemini typically defends corporate positions, making it an ideal sparring partner for legal professionals preparing for opposition.
All the interactions below come from a single Gemini 2.5 Pro conversation, because unlike competitors, Gemini maintains consistent performance in long chats without slowdown. However, it doesn’t reference earlier conversation turns like Claude does, making context length irrelevant either way.
Despite Google’s search expertise, however, Gemini lacks effective chat search functionality. Search results show which conversations contain terms but don’t highlight or link to specific mentions. Unlike competitors, Gemini offers no chat export capability, and even Chrome browser search fails unless the full conversation is loaded through manual scrolling.
All three major AI systems offer persistent customization features that function like system prompts, though finding them requires navigating confusing menu structures. Gemini places this under Settings & help > Saved info. These features allow users to establish standing instructions for response style, analytical approach, or domain expertise that persist across conversations. This eliminates both the workflow inefficiency of repeatedly establishing preferences and the resource waste of using valuable turns/tokens on the same setup information in each new chat.
In the chat excerpts below, I’ve spelled out abbreviations and clarified references for readability—my actual prompts were more compressed due to context limits.
Want to see how this unfolded? Here are excerpts of pivotal points from that actual conversation.
Personality Profiler Gemini
Prompt: On my work account, I asked Gemini Pro to evaluate my professional strengths and weaknesses based on that chat. It did the best job of all advanced AI by identifying where I might have the most trouble—adjusting my standards/work pace to a corporate setting (so basically know when to move on and get along, even at the cost of my own standards sometimes). This was impressive analysis.
How to Summon the Legalese Analyst
Prompt: But then you all (3 advanced AI) do so poorly analyzing subtext or obfuscating language? I need to prompt you and point out the issues to get a proper analysis.
Legalese Analyst Gemini
Prompt: I have Siri turned off because of this concern, but I tried it yesterday to compile some notes. I spent so much time correcting the note (by typing) that I lost the thread anyway. Transcripts work so much better. Or are Voice Memos shared, too?
Prompt: Something else I noticed: in the second paragraph, “Apple servers” in the first sentence is replaced by “the server” in the next sentence. Why might that be?
Prompt: Dictation involves mostly processing, not storage (unless users opt in), so one server is probably sufficient. You bring up an important angle I hadn’t considered until now—that it might not even be an Apple server, which is even more concerning.
Prompt: I’m wondering if they’re conflating two different categories. I am cancelling my paid subscription to move down to a free tier so my chats can be used as training data for Claude (because all my feedback, safety reports, and even human support requests went ignored). I am still keeping the account open. My feeling is they’re confusing subscription cancellation with account deletion/closure.
Business Strategist Gemini
Prompt: Another AI company is launching beta testing on embedding that AI in Chrome. That AI and I are puzzled by this, because Chrome is your cousin. Wouldn’t they have been better off striking a deal with Firefox instead rather than trying to horn in on your home turf?
That same AI company is now going to update its service terms and privacy policy so that paying users can opt in to have their chats used for training purposes. That AI seems to have (not the most up-to-date but) solid training data (better than you or GPT, based on my X-Files model-off). The only advantage of user data in that case seems to be from coding tasks. But power users who use AI for coding are probably the least likely to want to share their chats, so I’m genuinely perplexed. Other serious users are likely going to feel the same way. Only inattentive users who agree to terms of service without reading them will “opt in,” and those chats are likely to be very poor-quality data. Can you see any redeeming quality in this plan?
How to Summon No-Fluff Gemini
Prompt: I noticed you’ve cut out the fluff I was complaining about, but it seems limited to this chat. What’s the best way to get other-chat Gemini to do the same? Should I mention token parsimony to set the tone?
Prompt: A funny thing I noticed earlier was that with a simple instruction to be concise, you actually cut down on substance rather than the fluff. So I’ll be using your new instructions instead. They’re almost like a system prompt for each chat. It’d be nice to browse and pick a tone for the entire account. I’m hoping your team will hear me on this as well.
Devil’s Advocate Gemini
Prompt: The editor step could also work hand in hand with the cross-modal consistency check, so it’d be a great differentiator going forward. And I think you’d make a great sparring partner for lawyers prepping for their trials!
Evolutionary Biologist Gemini
Prompt: People don’t anthropomorphize weather models but do that with AI somehow.
