7 Strategies to Master AI So Well That You’ll Be Mistaken for It

Sam Liang is shocked as I reveal my method for recording interviews: using the Voice Memos app on my iPhone and manually transferring the transcript to a Google Doc. The CEO of Otter, a transcription service aimed at meeting analysis, looks at me as if I attempted to join our video chat with a rotary phone. He understandably believes I should convert to Otter. He’s probably right.

This is all part of a fresh identity in both professional and personal realms: being AI native. Time-saving productivity tools, like innovative note applications, task-specific AIs, and conversational inbox assistants, are surging in popularity as they permeate various aspects of our digital existence. While it’s essential to remain mindful of security and hallucination risks with any AI tool, early adopters are cultivating a fluency that will likely yield benefits for years ahead.

Being AI native—or “agentic,” as adherents call it—means being flexible to new experiences. Despite transcription mishaps, I’ve welcomed experimentation, from generating AI podcasts to using Claude for organizing my desktop files. (Some of this was discussed in my newsletter series last year, AI Unlocked.) If you aim to master AI tools to the point where colleagues wonder if your veins run with data cables, here are my seven tips for ascending in the AI landscape.

1. Ditch Your Chatbots

ChatGPT is yesterday’s news. Nowadays, the trendsetters are embracing Codex. Your eyes may glaze over at the term “AI agents,” but compared to options even a year ago, automation tools like Codex and Anthropic’s Cowork are miles ahead in taking control of your computer to accomplish tasks. Don’t waste your efforts on a single chatbot when you could command a whole fleet.

2. Embrace Voice Mode

Oh, you’re still manually typing out every instruction for your AI, like it’s the ‘90s? That’s adorable. But heed Liang’s advice: “Voice will be increasingly dominant,” he shares. “People detest writing.” (He adds that I, as a journalist, likely don’t share this disdain for writing, which is mostly accurate.) This suggestion primarily applies to input, not necessarily to output. I seldom utilize the voice-only mode on ChatGPT, for example, yet I often voice a prompt into my phone and then review the written output.

3. Create a Sandbox

Even though these agents have improved considerably, they can still wreak havoc without appropriate limits. (Earlier this year, a Claude-powered agent wiped out a startup’s entire production database and its backups.) So, if you’re ready to let an external entity take control of your computer, spend some time researching what these tools can do and set up dedicated folders with the files you want them to access.

4. Share Everything

With apologies to our privacy-focused security writers, it’s a fact that the more information you provide to AI, the more tailored the outputs will be. Jo Barrow, chief of staff at Granola, an Otter competitor, explains: “I have a personal OS system, a set of files on my computer where my AI resides. When I ask questions, the context is readily available, and the agent can process it. I don’t have to keep restating myself.” Fair warning: Confidential conversations are still best conducted without a permanent record.

5. Develop an Impersonator

Barrow shares that she consolidates all her Slack messages into a document so that her bots understand her communication style on that platform, and she replicates this for her email and social media profiles. “People use AI to refine their tone of voice,” she states. “There’s only so often you can say, ‘OK, a bit warmer. OK, a touch less formal.’ That’s a significant time drain.” While these guides won’t completely mimic your voice, they can urge the bot to produce responses that are closer to your rhythm and tone.

6. Collaborate Across Teams

Data is potent, and adding more from those around you can further improve AI tools. Consider your colleagues: “Many are using a meeting note-taker, but they’re still applying it at the individual meeting level,” Liang notes. He highlights the “knowledge engine” that Otter can create when a whole workplace engages, from engineering to marketing. You can even do this at home: If family members funnel various notes from their day into a shared AI tool, it will yield more insights than isolated usage.

7. Master Jailbreaking

Effectively utilizing AI tools in 2026 won’t require crafting—rather, I mean speaking—perfect prompts. Nonetheless, initiating more intricate tasks with a well-structured request can be crucial. Play around with the phrasing, especially if you encounter unexpected blockages hindering the output. Recently, I tried persuading a bot to provide email addresses for niche experts, but it wouldn’t comply. However, when I initiated a new chat and explained my need for this information (for reporting purposes, naturally), it provided a list.

What’s Your Take?
Share your thoughts on this article in the comments below. Alternatively, you can send a letter to the editor at [email protected].