-
-
Personal Quest Rank three suggested directions, choose a duration, and receive a personalized three-, five-, or seven-day learning journey.
-
Overview One Daily Ten becomes a complete learning journey: read or listen, test your understanding, and continue with a Personal Quest.
-
Growth dashboard A privacy-safe Admin dashboard turns aggregate user activity, paid conversion, and AI-generated Growth briefs.
-
The day began without a lesson. The agentic system found a timely topic, researched and checked it, then published it automatically.
-
AI discovers topics, creates lessons, verifies evidence, passes quality gates, and builds Personal Quests, with human review when required.
-
A consent-based Mailjet campaign connected 190 delivered emails to 44 visits, 14 paid signups, and USD 16 in attributed revenue.
-
Ten Minutes Wiser in one view: daily 10, personalized quests, autonomous content production, and the operation of a real AI-native business.
Inspiration
I love reading and learning new things. When a topic catches my attention, I can easily spend hours—or days—following one question into the next. I enjoy that kind of deep dive, but I also know how demanding it can be. Finding good sources, understanding a difficult subject, and deciding what is actually worth remembering takes time and energy.
I often found myself wishing for an app that would do the searching with me: something that could surprise me with a fresh, relevant topic every day and explain it in the time it takes to drink a cup of coffee. I wanted to be able to read it like a short newspaper article or listen to it like a mini podcast. Most importantly, I did not want another endless feed. I wanted one good idea, condensed into no more than ten minutes.
I have a medical background, not a programming background, so for a long time this remained just an idea. When I heard about the Build with Gemini XPRIZE, I thought: this may be the moment to finally try. It felt like now or never.
That is how Ten Minutes Wiser began.
On August 17, the product surprised me in exactly the way I had hoped it would. There was no Daily Ten prepared, so I opened the protected Admin area and asked the Topic Scout for suggestions. One of them was Black Cat Appreciation Day, approached through the genetics of melanistic coats and their possible evolutionary trade-offs. I had never heard of the day, and I certainly did not expect it to lead me into genetics and viral immunity. I immediately wanted to know more.
That small moment captures the reason I built Ten Minutes Wiser: it found a connection I would probably have missed and turned it into something I could genuinely learn from.

What it does
The free experience is the Daily Ten: one timely, non-political, evidence-backed topic shared with the whole community each day. You can read it, listen to it, or follow the text together with the narration. At the end, a short multiple-choice Wisdom Check helps the lesson stick. Progress is saved across devices, and the whole experience is designed to fit into ten minutes.
This is exactly the size of learning I wanted for myself: substantial enough to teach me something real, but small enough that it does not take over my day.
Sometimes one Daily Ten is not enough. Sometimes it opens a door and I want to keep going—or I am simply in the mood for something different. That is where Personal Quests come in.
Instead of asking the learner to write the perfect AI prompt, Ten Minutes Wiser presents three possible learning directions. The learner ranks them, chooses a three-, five-, or seven-day journey, and the Quest Director builds a short curriculum around those choices. In the production journey shown here, the black-cat Daily Ten led into a three-day Quest called From Sundials to Atomic Clocks: Measuring Human Time.

How Gemini helps run Ten Minutes Wiser
Gemini does much more than generate a block of text. It helps with the repeated work that would be difficult for one person to perform every day: finding promising topics, researching them, organizing claims and sources, assembling lessons and quizzes, preparing narration, checking quality, and creating personalized learning journeys.
For each Daily Ten, the workflow starts by researching current Google Trends and identifying topics with strong learning potential. The selected topic then moves through research and production and a separate evidence-audit stage. The system creates the article, quiz, and audio narration, then checks the result for factual confidence, safety, duration, quiz quality, and audio readiness.
If a low-sensitivity Daily Ten passes every required gate and reaches the configured quality threshold, the agentic system has permission to publish it by itself. If it does not pass, it cannot publish. Instead, it is placed in the human-review loop in my protected Admin area, where I can inspect the lesson, its evidence, and the failed or incomplete checks.
The August 17 black-cat edition is a good example. It carried five sources and seventeen tracked claims, passed every displayed gate, received a 98% quality score, and was published by the system.
Personal Quests use the learner's ranked interests and chosen duration to create a connected sequence rather than a pile of unrelated articles. A separate Marketing Growth Director works only with aggregate signals. It prepares KPI briefs, points out bottlenecks, and suggests measurable experiments for me to review. Seven production Growth briefs were available in Admin at the evidence cutoff.
This division of work matters to me. Gemini can do the repeatable research, production, and analysis at a scale I could never manage alone. I define the publication rules and quality thresholds, review anything that is escalated to the human loop, approve every campaign and budget, communicate with customers, and take responsibility for what the business does.

The production system runs on Google Cloud using Gemini 3.6 Flash through Vertex AI, together with Cloud Run, Firestore, Cloud Storage, Cloud Scheduler, Cloud Text-to-Speech, Cloud Logging, and Cloud Monitoring. Separate workloads create Daily Tens, narration, Personal Quests, and marketing briefs. Failed attempts and retries stay visible in the operational record, because a trustworthy system should show where it struggled as well as where it succeeded.
A real business with real users
I did not expect the business side to move as quickly as it did. By the final evidence cutoff, Ten Minutes Wiser had 292 registered accounts, including 281 verified accounts and 32 current paid accounts. Three of those paid accounts were Friends and Family and are reported separately. The other 29 independent customers generated USD 37 in arm's-length revenue. Total revenue, including USD 5 in related-party revenue, was USD 42. Actual marketing and customer-acquisition spend was USD 0.
Mailjet handles consent-based email delivery. Only verified free accounts that explicitly agreed to marketing are eligible. Gemini can turn an approved Growth brief into a campaign draft, but I control the test delivery, approval, final audience confirmation, and send. The marketing dataset contains campaign-level results rather than recipient email addresses or payment identifiers.
The latest captured campaign was delivered to 190 eligible accounts. It recorded 44 first-party CTA arrivals, 14 completed checkouts, and USD 16 in attributed revenue. That is a 7.4% delivered-to-paid rate and a 31.8% CTA-to-paid rate. Across the product, paid accounts represented 11.0% of registered accounts at the cutoff.
These are very early numbers, and I do not want to pretend they predict the future. But they made the project feel real. People I did not know were willing to sign up and pay for something that had started as a personal wish: a calmer, more focused way to keep learning.

How Codex helped me build it
I am building Ten Minutes Wiser alone, and I am not a programmer. That made Codex especially important. It helped me translate what I wanted the product to do into implementation steps, understand and review code, find and fix problems, test production behavior, and organize the evidence behind this submission.
Codex did not choose the product for me. The original idea, the learning experience, the editorial judgment, customer communication, approvals, and final decisions remained mine. What it changed was my ability to turn those decisions into a working product. It gave me a way to build something that, not long ago, I would have assumed required a full technical team.
Jobs and economic opportunity beyond the founder
Ten Minutes Wiser currently has no employees or contractors, and personnel expense is zero. Any job creation is therefore a future opportunity, not a current claim.
If the product grows, I can imagine hiring subject-matter reviewers, educators, illustrators, motion designers, localization specialists, audio-quality reviewers, and performance-marketing partners. My goal is not to remove human expertise. It is to let a small team spend more of its time on judgment, creativity, and quality while AI handles a larger share of the repetitive research and production work.
For learners, I also hope affordable personalized education can make it easier to build knowledge and useful learning habits without needing a private tutor—or several free hours every day.
What is next
The current marketing system already produces recommendations and briefs for email, LinkedIn, and experiments on the Ten Minutes Wiser site. Email drafting and Mailjet delivery are implemented with human approval gates. The next step is to bring the same controlled workflow to LinkedIn and other social channels.
I also want to make the lessons more visual. My idea is to add topic-specific 2D or 2.5D illustrations and short animations that make difficult concepts easier to grasp. AI could help propose scenes and production assets, while art direction, factual review, licensing, and final approval would remain explicit human steps.
Personal Quests should also become more adaptive over time, using consented learning signals to understand which subjects, formats, and levels of difficulty work best for each learner.
Ten Minutes Wiser began with a very personal question: could I build the learning companion I had always wanted, even though I was not a programmer? The answer is now live at tenminuteswiser.com. Every day it gives me one new reason to be curious—and, ten minutes later, a chance to be a little wiser.


Log in or sign up for Devpost to join the conversation.