If you only want one takeaway: the best personal AI assistant is the one whose tradeoffs match how you work, and the three tradeoffs that actually matter are where your data lives, what it can automate, and how it charges you. I spent the last few weeks running the major options through real daily tasks, drafting emails, summarizing documents, planning my week, and capturing meeting notes, and the differences came down to those three axes far more than to raw model quality, which is now excellent almost everywhere.
This is a first-person, tested rundown of six options I keep coming back to. Five are the well-known cloud assistants most people already recognize. The sixth is the one my team builds, Routines, included here because it occupies a niche the others do not, a local-first assistant for the Mac. I will be upfront about that bias throughout, and I will be generous about what the others do well, because they do a lot well.
How I judged them
I held every assistant to the same three questions, because these are the decisions you live with after the novelty wears off.
Where does your data live? Your conversations, your notes, and whatever the assistant remembers about you have to be stored somewhere. For most cloud assistants that somewhere is the vendor's servers. For a local-first tool it is your own machine. Neither is wrong, but it is the single biggest difference between these products and worth deciding on deliberately.
What can it automate? Every assistant on this list can answer a question. Fewer can act on a schedule, react to an event, or run a multi-step task without you babysitting it. If you want an assistant that does work while you are away from the keyboard, this is the line that separates them.
How does it charge you? Most personal AI assistants are monthly subscriptions tied to a single model family. A few let you bring your own API key and pay only for what you use. I am not quoting specific prices here because they change often, but the pricing model matters as much as the number.
The best personal AI assistant options I tested
ChatGPT
ChatGPT is the assistant most people mean when they say "AI." It is fast, broadly capable, and the ecosystem around it is enormous, with custom GPTs, voice mode, image generation, and a steady stream of new features. For general-purpose questions, brainstorming, and quick drafting it is hard to beat as a starting point, and the mobile and desktop apps are polished. Your conversations and memory live in OpenAI's cloud by default. Automation is mostly request-and-response, though scheduled tasks and connectors have been expanding. If you want one familiar, dependable assistant and you are comfortable with cloud storage, it is a safe pick.
Claude
Claude, from Anthropic, is the one I reach for when the task is long, nuanced, or writing-heavy. It handles large documents gracefully, tends to follow detailed instructions closely, and its tone in longer drafting tends to feel more considered. Like ChatGPT, your data lives in Anthropic's cloud, and the core experience is conversational rather than autonomous. If your work is research, analysis, or careful writing, Claude is an excellent default. (Full disclosure on the model side: Routines can run on Claude, among other models, so I use it daily in both contexts.)
Google Gemini
Gemini's advantage is obvious if you live in Google's world. It is woven into Gmail, Docs, Calendar, and Android, so the assistant already sits where your work happens, and pulling context from your own mail or documents is smooth. It is a strong generalist with good multimodal handling. Your data lives within Google's account ecosystem. If Workspace is your home base, Gemini removes a lot of copy-paste friction that standalone assistants create.
Microsoft Copilot
Copilot is the natural choice inside the Microsoft stack. In Word, Excel, Outlook, and Teams it can draft, summarize threads, and turn a rough table into a formatted analysis without leaving the app. For organizations already on Microsoft 365 it is the assistant with the least adoption friction, because it shows up in tools people already open all day. Data lives within your Microsoft tenant or account. If your day is spent in Office apps and Teams, Copilot meets you there.
Perplexity
Perplexity is the one I open when the job is "find out and cite." It is built around answering questions with live web search and visible sources, which makes it my go-to for research where I need to trust and verify the answer rather than just accept it. It is less about drafting and automation, more about fast, sourced answers. If your main need is current, citation-backed information, it does that job better than a general chat assistant.
Routines (the local-first Mac niche)
Here is my disclosure again, plainly: I work on Routines. I am including it because none of the assistants above are built around keeping your data on your own machine, and that is the entire premise of this one.
Routines is a local-first macOS app. Its memory, your notes, meeting transcripts, and routine outputs, lives as plain markdown files and a local SQLite database in a folder you choose, and none of it is uploaded. You can point it at an existing Obsidian vault or notes folder and that becomes the assistant's brain immediately, and you can open or back up those files in any editor. The only server involvement is account sign-in, trial, and licensing, which never sees your notes or conversations.
On automation, it goes further than the chat assistants. You can schedule AI routines with cron or simple presets, or fire them on events like Mac wake, app launch, or an incoming webhook, so a morning briefing pulling your calendar and email can be waiting before you open the app. It also records meetings without a bot joining the call, captures both your mic and system audio with speaker separation, and turns transcripts into summaries and action items locally. Add global-hotkey dictation and a searchable clipboard history and it covers ground three separate tools usually handle.
On price model, it avoids stacking another monthly subscription on a single model. You bring your own Anthropic, OpenAI, or OpenRouter key, or sign in with a ChatGPT account you already pay for, and the app talks to the model directly from your Mac. There is a 7-day trial to start. The honest tradeoff: it is Mac-only, it asks you to think about which model and key you use, and it is a younger product than the giants above.
So which one is the best personal AI assistant?
There is no single winner, and anyone who tells you otherwise is selling something. Here is how I would actually choose.
Pick ChatGPT or Claude if you want a polished, do-everything cloud assistant and you are comfortable with your data living on the vendor's servers, ChatGPT for breadth and ecosystem, Claude for long-form and careful writing. Pick Gemini or Copilot if you already live in Google Workspace or Microsoft 365 respectively, because the assistant that sits inside the tools you already use wins on friction. Pick Perplexity when your core need is sourced, up-to-date research.
Pick Routines if you are on a Mac and the answers to my three questions matter to you in a specific way: you want your notes and memory to stay on your own machine, you want real scheduled and event-driven automation rather than only chat, and you would rather use a key or account you already pay for than add another subscription. If that is you, it is worth a look as an AI assistant for Mac, and you can browse the prebuilt recipes to see what the scheduled side actually does, or read more about the local-first approach to data.
FAQ
What is a personal AI assistant? A personal AI assistant is software that answers questions, drafts text, and automates tasks on your behalf using a large language model. The best ones also remember your context across conversations so you don't repeat yourself.
Which personal AI assistant is most private? Privacy depends on where your data lives. Cloud assistants like ChatGPT, Claude, and Gemini store conversations on their servers by default. A local-first app like Routines keeps your notes and memory as markdown and SQLite on your own Mac.
Can a personal AI assistant automate tasks on a schedule? Most chat assistants respond when you ask. Scheduled automation is rarer. Routines runs AI routines on a cron schedule or on events like Mac wake, so a morning briefing can land before you open the app.
Do I need a separate subscription for each AI assistant? Often yes, since each is billed monthly. Routines lets you bring your own API key or sign in with a ChatGPT account you already pay for, so you can avoid stacking another subscription.
Try the local-first option
If the cloud assistants already serve you well, keep using them, they are genuinely good. But if where your data lives, real automation, and avoiding another subscription are the parts you care about, the local-first approach is worth trying for yourself. Download Routines, point it at your notes folder, and let a morning briefing run before you sit down. Your memory stays on your Mac, and you can walk away with the files anytime.