# FY1 S2 W1 Product # Overview Get a personal team of AI assistants working just for you on your own data on Junwon. You can add notes and spreadsheets to it, or import data from your other devices or services like Apple Health or Chase Bank, then ask ChatGPT to answer your questions about your own data as opposed to the generic web. We bring the power of AI not just to the entire humankind but to each person. # Introduction ### **Database with GPT** You have a team of AI assistants working just for you on Junwon app, helping you remember your own notes, study patterns in your own spending habits, getting reminders about your own contacts. You can share any information you add to the private and secure database on Junwon app with GPT assistants you add or create. Bring the power of AI to your own data at just $1 a day. ### **Example Use Cases** 1. Make Fitness Assistant track trends from your Apple Health data, plan new workout routines, and log exercises for you. 2. Make Contacts Assistant remember notes like the name of your friend’s puppy, or the brand of wine that your friend likes to drink. 3. Make Travel Assistant suggest to you new nearby landmarks to explore this weekend based on where you have been and where you enjoyed visiting in the past. ### Adding New Data 1. Quickly create new notes on Junwon app on iOS, Android, or web browser. 2. Notes can be in plain text or in spreadsheets. 3. That is it! You can now ask GPT about or across any of your notes! ### Importing Your Data 1. Upload CSV files. 2. Use connectors to integrate with popular sources like Apple Health and Chase Bank. 3. Write Python scripts to scrape data the way you want. ### Making Custom GPT Whereas there are many GPT assistants that you can choose from on our AI Directory, you can also make new ones easily to fit your needs. In fact, making a new GPT assistant is so easy that you can do it in minutes, on the mobile app. No cloud deployment or API integrations. Just follow these simple steps. 1. Create a new assistant. 2. Give it access to specific datasets. 3. Write a plain text note explaining to it what role it shall play for you. When you feel comfortable making custom assistants, try pushing it to the limits by experimenting with different parameters and different base models. Sharing an assistant you made with your contacts is also as simple as publishing it on the AI Directory and sending a link to it to your contacts. You can also optionally choose to pay for their usage of your app if you want. ## Get started Give it a try! Add some notes, ask some questions, and you’ll feel the magic in minutes. Click here to get started. $28 for 28 days, and you can get a refund for every single day left in your plan if you choose to leave in the middle. Still not convinced? Check out this demo on our app to see what it can do for you. # Target Customers ### Alex Alex is a data scientist at Microsoft, living in Redmond. - Traveling: On weekends, Alex likes to drive out several hours to visit great hiking trails. He is spending too much time every week discovering where to go, finding best combination of destination and time based on weather and crowdedness and road condition and whether flowers have blossomed, and finding which roads to drive along and where to stop for meals and rests along the way. - Reading: Alex needs to read lots of news, respond to lots of emails, and attend lots of meetings. He has many small tasks that he needs to sort everyday. He wants to get better at keeping up with academic papers related to artificial intelligence. He had a great reading group to read academic papers with when he was in college, but he was not able to find one at Microsoft. - Cooking: Alex usually cooks his own meals during weekdays. Once or twice a week, he likes to eat out, sending text messages to friends, and suggesting that they visit a restaurant that he can find great reviews for. Alex is mildly interested in getting better at cooking, introducing more variety and flavor to meals he cooks, but he has not been able to find motivation to search for recipes or plan meals. Planning meals may help him meaningfully enhance quality of life. On a related but separate, Alex has tried recording his meals to understand his nutrition intake, but he has not been able to keep up due to the difficulty of finding out the weight and nutrition breakdown of every food item for every meal. ### Clara Clara is a frontend developer at Airbnb, living in San Francisco. - Contacts: Clara wants to get better at keeping in touch with her friends. She keeps forgetting to reach out to friends with whom she has not chatted for a long time, and she keeps forgetting to post on social media. She wants to get better at portraying herself on social media the way that is likable and still distinctly herself, but she is not sure what kind of person she really is or wants to be let alone how to communicate it well on social media. - Exercise: Clara wants to exercise more. She has been looking into going to the fitness center at her apartment building, running at the park, or learning yoga. She likes to listen to podcasts during exercise sessions. - Shopping: Clara likes K-pop, K-drama, and fashion. She likes to shop fashion for fun, just browsing to relax, and without a strong purchase intent. She is interested in better managing her spending habits. She has a suspicion that she is spending too much on subscriptions and luxury items that she does not need to spend money on. ### Both Alex and Clara are 27 years old, studied Computer Science at Stanford University, earns $12,000 of disposable income in a year after paying for housing, transportation, and food. Alex and Clara like to develop side projects for fun and to learn and practice technical skills. Alex and Clara are both interested in using Stable Diffusion models or Open AI GPT API to develop programs, but have not had a chance to develop any yet. It is easy to make the AI models do something for them, but there is a big gap to delivering that utility to consumers. Alex and Clara are strongly interested in applying GPT to live more conveniently. # Customer Story Workday desires are more important to us since people live more workdays in a week, get work done on workdays, and workday desires are likely to also appear on weekends. This story assumes Polaris technology and not the minimum product. ### Alex: Weekend Trip It is Friday lunch time. Alex is already thinking about what to do over the upcoming long weekend. He is planning to explore nature. He can go anytime on Saturday, Sunday, and Monday, but he will be too tired to go more than once. He wants to depart from home and from the hiking trail and from the rest stops along the way such that he can avoid traffic, get home ideally before sunset, leave hiking trail definitely before it gets dark, and avoid visiting any businesses outside of business hours. He is also fine with driving along new roads and check out new restaurants along the way even if it takes a little longer to drive. He wants to drive along scenic routes. He is interested in mountains, beaches, national parks, and less popular hiking trails. He can drive towards the North to the Mount Baker or North Cascades, East to Mountain Loop or Leavenworth, West to Olympic National Park or Olympic Peninsula, or South to Mount Rainier or Mount Saint Helens. He wants to pick a place where weather will be nice and sunny, not too hot and not too cold, hiking trail is open and snow has melted sufficiently. Alex opens Junwon app. He opens a trip planner assistant. It suggests him to first choose when to go where. It makes a table where rows are destination candidates and columns are hour candidates. He tells it he is considering of going anywhere that is nature and is within 3 hours of drive from Redmond. It remembers that he was interested in weather, trail condition, driving hour, and seven other criteria from last time he planned his trip. It asks him if he wants to search the same. He says yes and the matrix is filled out. He chooses to visit Maple Pass on Sunday afternoon. He asks it to help him plan the itinerary. It asks him if he wants to get lunch on the way or before he departs home, and dinner on the way or after he arrives home. He says he will get lunch on the way, but not dinner. It says there are three restaurants he will drive by between 11AM and 1PM that fit his usual taste preferences, but there is also a famous Italian place if he is interested in trying new cuisines he normally does not try. He likes the idea, and it generates an itinerary for when to depart where. Based on the trail condition, it suggests to him what to pack. The night before his trip, he follows its instructions to pack. He also adds sunblock, and asks it to remember to suggest sunblock starting from next time. On the day of, it streams songs generated to his taste to make his driving time more fun, helps him pick a menu at the restaurant, and alerts him to turn back and head to his car to get to it before the sun sets. He is now back home, reflecting how helpful the trip planner assistant on Junwon app has been to making his weekend trip great and safe. ### Alex: Workday Morning The long weekend is over and a new workday begins. Alex sits down at his desk, and opens Junwon app, his workplace office, and his morning briefing room. In the room are several AI assistants working together to make sure he is on top of what he needs to know. His Calendar Secretary informs him what meetings he has and what tasks he had planned from last workday so that he can get back into working mindset. His Messages Secretary informs him that he has 80 unread emails and 16 unread messages, and suggests that he pay attention to 5 emails and 8 messages that look urgent, and then dedicate an hour after lunch to catching up on the rest. He says yes, and the Messages Secretary reads to him summaries while displaying the full text. Based on his past behavior pattern, the app recommends that he now ask News Assistant to read headlines. He has instructed News Assistant to pull information from Hacker News, his Twitter feed, specific Subreddits, email newsletters, and several RSS feeds. It shows him the articles ranked by the ranking algorithm that he instructed it to follow. For this morning, he says he actually wants to see all headlines mentioning Microsoft and OpenAI in the same article. News Assistant filters articles down to those, and he reads on, selecting what he likes and what he dislikes so that News Assistant can better rank articles in future days. He wants to add TechCrunch to the sources from which News Assistant scrapes. He says he wants to add techcrunch.com, and News Assistant asks him to give it top articles he sees, such as title and content. He gives it several examples, then it parses the website to make a custom scraper for that specific website. From that day on, News Assistant adds new articles from TechCrunch to daily briefings. He is working hard. Then Junwon app sends him a notification advising him to go for a 30 minute run in 15 minutes, since he has no other time today to go for exercise. ### Clara: Weekend Lunch with Friends Clara gets a notification from Junwon app that she has not reached out to her friend Diana for a long time. Clara opens the app, and her Contacts Secretary informs her that she chatted with Diana about certain topics last time, and that she met Diana along with these other people last time. Clara starts a group chat with all of them and suggests that they catch up over lunch next week. Contacts Assistant asks them what times they are free. There is no overlap, but if Diana can make it on Saturday, there is a time that works for all. It asks Diana if she can make it on Saturday, and she responds yes. Contacts Assistant asks them what they are interested in getting for lunch. Most have no preferences, but Diana likes Asian food, and Clara likes Sichuan food. Contacts Assistant finds restaurants in the area that serve Sichuan food, filtered by price range, ranked by recent reviews, and adds other options in case group is interested in trying something new. It also shows them photos from recent reviews on Google Maps and Yelp. Group picks the top restaurant, and Contacts Assistant gives Clara an instruction on how to make the reservation. If this was a chat or a call on a computer, Contacts Assistant would have been able to automatically record conversations and inform Clara next time. This time, the meeting happened offline, so it asks Clara how the lunch was. Clara recollects the event, interesting stories she heard from friends, and how Diana got a new Pomeranian puppy named Snowball. In several weeks, Contacts Assistant will remind her to ask Diana about Snowball. ### Clara: Workday Evening Clara opens Meal Planner Assistant. It asks her if she ended up using all the ingredients from last week, and lists the meals she had to remind her what ingredients may have left. She says she still has a radish, but that’s it. Meal Planner suggests some flagship cuisines to try cooking in the coming week. The suggestions are a balance of what she likes and what she has not yet tried. This week, there is a suggestion for a Spanish cuisine, because people who like cuisines that Clara likes usually also like this cuisine. Clara picks it. Meal Planner Assistant records nutrition information of that Spanish cuisine, and then informs Clara to pick several more. Clara does. Meal Planner Assistant now informs Clara that she is missing Omega-3 from this week’s diet, and suggests several options for adding it to her diet, along with why it is important for her health. When she has a perfect plan, she asks it to give her a grocery shopping list. She visits the grocery store, buys as instructed, then finds a fascinating meal kit for Japanese Soba. She tells Meal Planner Assistant that she wants to have Japanese Soba for dinner. It responds she will get more carbs and less protein than planned for tonight, but can make changes to the remaining week to make up for it. She says yes, takes a photo of the nutrition label on the Japanese Soba meal kit, and heads home for dinner. # Study Plan ### Study 1: Needfinding Interviews 1. Can you pick three workdays from this week, and tell me what you did from the moment you woke up to the moment you fell asleep? Tell me about what you did, but outside of what you did for work. Also tell me about a weekend day. What are routine tasks you do on a daily basis? 2. What kind of information do you record on a daily basis? Notes? Exercises? 3. How do you record the information? What apps do you use? 4. What kind of information do you study on a regular basis? Do you review your banking statements? How you are spending time by looking at your calendar? 5. How do you store the information? What databases or devices do you store the information on? 6. How do you process the information? How do you extract insights? Do you visualize the insights? 7. What kind of insights do you get? 8. What pain points do you experience when recording the information? While storing the information? While processing the information? ### Study 2: Introduction Page and Waitlist Registrations 1. Describe the utility proposition of our product, and allow people to sign up on a waitlist.