I’ve always been intrigued by performance coaching, but traditional coaches are prohibitively expensive—hundreds of dollars for a one-hour session. Frankly, it’s way out of my budget. On top of the cost, finding a coach who is available in my area (Idaho) and familiar with my industry (tech journalism) adds another layer of complexity.
When AI tools emerged, I began hearing about people using them for all sorts of personal applications, including therapy. I wouldn’t use an AI as a shrink, but it seemed viable for performance coaching, which often doesn’t require licensing. So for two months, I employed Gemini as my own AI performance coach.
Below, I’ll show you how I did it, from the files and prompts I provided the AI, to how I used the feature known as Gemini Gems.
Meet My Coach: Google Gemini Pro
For this experiment, I used Google Gemini. I was about 10 months into a 12-month trial of the subscription Gemini plan, recently rebranded as Gemini Pro, which unlocks advanced features like improved reasoning models, enhanced integration with Google Drive and Workspace, and multimodal capabilities.
Advanced models, like Gemini 2.5 Pro, leverage enhanced reasoning to “think through” complex questions and provide insightful answers. This feature is crucial for a coach to move beyond superficial responses and deliver nuanced, actionable insights.
I chose Gemini because my personal journals are in Google Docs, and Gemini allows me to give the chatbot direct access to those files and avoid a clunky upload process.
Another big reason I went with Gemini is its impressive million-token context window. In AI terminology, a token is a fragment of a word, sort of a cluster of letters or characters, which it uses in the statistical model that undergirds large language model (LLM) technology. Every time you ask a question and get a response, the chatbot reviews the entire conversation, from start to finish, and uses that data to produce the next response. Larger context windows allow for richer, longer ongoing conversations. So I was able to use Google Gemini for weeks rather than days.
Getting My Coach Up to Speed: Journals, Prompting, and Gem
Setting up my AI performance coach was surprisingly simple. At the outset, I shared roughly 10 months’ worth of journal data with Gemini. This gave it records of my day-to-day activities, like work tasks and exercise details, and just enough personal information. Sharing my journal was a big step, and I don’t recommend it for everyone. If you’re hesitant to get so personal with your AI coach, you’ll need to frequently provide more context within the conversation.
Once Gemini had my journal, I needed to give it the performance coaching persona. My initial prompt wasn’t that sophisticated:
"Act as a performance coach with a focus on productivity and personal development. Use my uploaded journals as background information to provide me with insights and actionable advice to enhance my performance in both work and life."
I tweaked and expanded this prompt over time, but this was the nucleus of what I used throughout the experiment.
As time went on, I found that I usually only got about two weeks out of a single discussion before the context window filled up enough to start losing track of details, requiring me to upload the journal file again and reenter the prompt to start the performance coaching function back up again.
This was a bit of an inconvenience, especially when competitors like ChatGPT offer persistent memory. It also meant that these new instances of the performance coach persona, though based on similar prompts and materials, were all slightly different and required some tweaking to maintain a consistent tone and style.
Building a Gemini Gem (Credit: Google / Brian Westover / René Ramos)
I was able to address this issue using Gemini Gems, a feature that lets you save a prompt and attach files for easy reuse. It let me save both my updated performance coach prompt and my journal documents.
Once I had the files and prompt in place, it was time to actually put Gemini to work. I found two main ways to engage with it: doing deep analysis that spanned the breadth of my journals, and then using it for real-time, day-to-day coaching.
I could kick off deep analyses anytime, often by asking simple queries like, “Analyze my journal for patterns that negatively or positively impact my performance.” These often yielded a goldmine of information. The persona identified my habits (good and bad), highlighted effective tactics I had already adopted (but maybe didn’t apply as often as I could), and spotted triggers for undue stress. I also asked it to identify both explicit and implicit goals from my journals and notes, and requested a SWOT (strengths, weaknesses, opportunities, and threats) analysis.
Meanwhile, I’d check in a few times a day, updating my coach with new information as I went. I soon settled into a routine similar to my journaling habit, with a morning check-in where I laid out my plans and tasks for the day, a midday update where I’d log my daily workouts and update it with any new assignments or priority changes, and then an end-of-day report on how things went.
As this routine continued, a few core functions of my AI performance coach became crystal clear.
Actionable Insights
My AI coach quickly zeroed in on a major problem: My sleep was a mess. In my journals, I’d often note when I had a late night. And as things continued with the performance coach, I began proactively recording my bedtimes. Whether due to external factors (a sick kid) or mundane issues (just plain staying up too late), a clear pattern emerged: A late night meant a tired morning, with ripple effects disrupting my work performance for days.
Gemini fixes my sleep (Credit: Google / Brian Westover / René Ramos)
It wasn’t just helpful in pointing out problems; it also provided actionable advice. When it came to my inconsistent sleep, the coach not only suggested an earlier bedtime but walked me through the reasons for it, detailing observed patterns in my behavior and their potential impact on my health, stress, and cognitive performance.
Get Our Best Stories!
Your Daily Dose of Our Top Tech News
By clicking Sign Me Up, you confirm you are 16+ and agree to our Terms of Use and Privacy Policy.
Thanks for signing up!
Your subscription has been confirmed. Keep an eye on your inbox!
Gemini even offered a solution to my sleep problems (Credit: Google / Brian Westover / René Ramos)
After making its case, my coach even offered suggestions, like creating an end-of-day wind-down routine and setting an automatic “sleep mode” on my phone.
A Hybrid Journaling Tool and an Accountability Partner
It became clear that my daily updates to Gemini contained the same information I was recording in my journals. This experiment helped me stay consistent with my journaling habit and usually resulted in better entries. Gemini became a hybrid journaling device and accountability partner, encouraging me to stick with my plans and align my daily tasks with broader goals.
Unlike human coaches, an AI coach can check in and discuss things multiple times a day. This is invaluable for accountability as you build new habits (like staying hydrated) or avoid unhelpful behaviors (like staying off of YouTube). It’s always available, patient, and nonjudgmental, helping you track your actions and results. When needed, it can provide reminders of your goals or prompt reflection to help you reframe challenges.
Comprehensive Task Tracking
Having another pair of eyes watching my to-do list was hugely helpful for me. As we tracked my tasks, the AI would suss out deeper patterns. For example, it pointed out that I often failed to protect my late afternoon “deep work” time, a crucial period for tackling big projects.
Gemini reminds me of my Monday morning tasks (Credit: Google / Brian Westover / René Ramos)
The task tracking also made it easier to balance work and family life. (Though, at one point, Gemini made the awful quip “The juggle is real.” Sometimes AI tries to be personable and just winds up being weird.) It seemed to understand that even though my work performance was a major motivator for this coaching, it wasn’t the whole picture. I want to be a high performer in my personal life as well, staying on top of the many demands of marriage, fatherhood, and home ownership.
An AI coach is also excellent at keeping tasks in perspective. As someone balancing multiple beats and a new AI column, bigger-picture goals can easily get lost. The coach consistently reminded me how individual projects fit into my overall objectives, warning when I seemed to drift off track.
Daily and Weekly Reflection
Beyond individual daily tracking, a user can ask for deeper reviews at any point. One of my coach’s most insightful observations came as Gemini started to pick up on things not explicitly stated in my journals. Gemini noticed that much of my day-to-day stress didn’t stem from inherently difficult projects, but from the tension between my morning plans and the inevitable disruptions—high-priority assignments, technical issues, or anything else that demanded immediate attention.
Gemini reframes my stress (Credit: Google / Brian Westover / René Ramos)
These sorts of tasks are simply part of my job, yet they pull me away from high-impact projects. The AI not only identified this pattern of stress but gave excellent advice for rethinking it. It suggested framing things in terms of “offense” and “defense”: my big, exciting projects were “offense,” while “defense” was holding ground, focusing on those unavoidable tasks that pop up. Both are important, and defense is unpredictable, but this mental switch has genuinely helped me stress far less about the sometimes chaotic nature of my job.
An Always-Available Sounding Board
One of the most significant benefits of this setup is having a sounding board that understands your job and goals. While a managing editor might have an open-door policy, they aren’t available 24/7 for every little thought. An AI coach, however, is available whether it’s 2 in the afternoon or 2 in the morning, ready to help you talk through projects, brainstorm ideas, or even just vent frustrations, all without an hourly rate.
Gemini as both coach and sounding board (Credit: Google / Brian Westover / René Ramos)
Ghosted: My AI Coach Suddenly Disappeared
Toward the end of my two-month performance coaching experiment, my 12-month trial of Gemini Pro expired. I thought I had another week or two, so this caught me completely off guard. Once that trial period ended, my performance coach went silent. The advanced reasoning model, the larger context window, the uploaded journal documents—all gone.
The functionality that made the AI performance coaching work disappeared. Gemini struggled to follow prompts beyond a few back-and-forth exchanges. I quickly noticed the absence of capabilities I’d come to rely on daily, from task reminders and consistent check-ins to that ever-present sounding board.
I missed a couple of days of journal entries and updates, and old habits that I thought I had tamed began to reenter my routine. My sleep patterns got messy again. I skipped my morning planning or let unimportant tasks intrude on my deep work time.
This unplanned break inadvertently served as the ultimate proof of concept, demonstrating that the performance improvements I’d experienced were genuine, sustained gains directly supported by the AI’s consistent presence. As someone with ADHD, I’ve long used different tools, strategies, and coping mechanisms to manage my time, energy, and attention. But this AI coach is the first thing I’ve ever used that actually made all of those disparate pieces feel like a truly cohesive system.
It only took about a week of trying to go it alone before a crucial realization dawned on me: My AI performance coaching was no longer a short-term experiment. It was a valuable tool that I needed to continue using.
I’m Sticking With It
I consider my two-month experiment a resounding success. My wife commented that I seemed to have more energy at the end of the day and was more engaged with our kids, all while my work productivity remained unusually high.
As a product tester and reviewer, I’m data-minded, so I tracked my performance. Before the coaching experiment, I averaged about six published articles per month. After starting with my AI coach in April, that number jumped to nine articles per month for the second quarter. The year-over-year comparison is even more telling: I published eight pieces in April 2025 compared with just two in April 2024.
Before he learned about my coaching experiment, my boss noted a significant uptick in my productivity. He said that while my work had always been good, my sheer volume and overall performance had taken a noticeable jump.
Ultimately, this experiment proves that AI can be far more than just a novelty. For me, it’s become a genuine partner in personal and professional growth, and I’ve restarted the AI performance coaching. I’m sure I’ll experiment with other approaches as time goes on, but for the moment, my AI performance coach has become an indispensable part of my daily life.
About Brian Westover
Principal Writer, Hardware
