We are continuing our trend of squashing bugs and improving product performance. Here’s what’s been fixed since last week:
💚 Memory Stack
💛 Sync Speech
💚 Sync Writing & Sync Clipboard
💛 Sync Twitter
💚 Sync Google Calendar
Since our big launch on Product Hunt, we’ve been busy little bees updating our systems to auto scale and making lots of fixes and enhancements to our app! This week’s release notes are dedicated to the bugs our community has helped us identify and our fixes. Here’s what’s new:
💚 Memory Stack
💛 Sync Speech
We rolled out many enhancements for Sync Speech and are still in the process of tuning the experience. If you are experiencing issues, please shoot us an email at support@personal.ai.
💚 Sync Writing & Sync Clipboard
💚 Sync Calendar & Twitter
💚 Login
💚 Account Creation
The team has fixed a handful of compatibility bugs with decentralized identity causing issues for new creators trying to complete their accounts.
A huge thank you to everyone for being patient, understanding, and helpful in identifying and fixing issues. We are moving quickly and pushing new fixesdaily to make your personal AI the best it can be. Thanks for being part of the journey.
Our team here at Personal.ai has been heads down writing tons of code behind the scenes while gathering a lot of feedback from the community of futurists and technologists who see the promise of personal AIs and the role they could play in our lives.
So, rather than just giving a list of all things that are built in this edition of our Release Notes, we want to start by giving you key components of what makes someone's Personal AI come to life. It’s Capture, Create, Connect. We feel good about that - it's easy to remember.
The data can be spoken, written, or visual. The data captured can be historical or it can be real-time. All the data capture tools are structured on the right sidebar. As of this upgrade, capture methods include:
Historical: Twitter and Google Calendar.
Twitter allows you to bootstrap your AI on how you think and helps generative models come to life and convey your voice all at the same time.
Google Calendar allows you to bootstrap your AI with people and events so your AI can use that to understand your life.
Real-time: Spoken and Written
Spoken conversations can be captured by turning on the toggle to ambiently process what you are saying. We have introduced voice fingerprinting models to improve speech recognition and more importantly ID which is your voice and which is not. We use sample audio to teach your AI your voice and background acoustic so it can ID places, such as work or home. There is no limit to how much you can capture with Sync Speech. Don't worry, we will never make you accountable for the size of your memory Stack and we have no intentions to limit this.
Sync Writing can be your daily journal, long-form reports or papers from the past, or a thought you want to preserve for the future.We associate the time when you capture the text and let you adjust it, too. Capturing written text is now made easier by capturing data captured into your clipboard while you are researching so you can focus on what you’re doing without having to go into the app.
We have updated our data structures and visual representation to accommodate all of this information as well.
The Create process is all automated. The data that's captured is automatically transformed into the memory blocks, collectively referred to as the Memory Stack. The metadata that's extracted such as topics or tags, people, and context is automatically composed creating a personal graph of your own. This process uses many different AI models.
The Memory Stack is then passed to AI models. We have created a Predict Model for facts and a Generative Pre-trained Model (GPT) on your personal Memory Stack. They get updated overnight without any input from you! I personally love this automation.
We also allow you to edit the information in your Memory Stack if you see something incorrect, so your AI can know you better. These Create tools and the Memory Stack is available on the left sidebar. You can edit, delete, navigate, search, filter, and explore the memories that you have created by source, people, topics, dates, and much more.
Connect allows you to access and interact with your personal AI. If you think about Slack or Discord or Teams as a human-to-human messaging platform, Personal AI Connect is a human to AI messaging platform. This interface is brand new and looks similar to messaging platforms. This unlocks the huge potential of not only connecting with your own AI but with other people’s AIs, too. So, you can let others subscribe to your AI on the most trending topics on your mind.
This platform allows your AI to proactively message you, too, based on context and relevance.
Connect is now possible with a new tab on the left bar and it’s called "AI messaging". For now, the messaging feature is limited but you may be surprised to see what we have coming up. Whatever we add, your AI will proactively message you about new capabilities.
The Predict function will now work within AI messaging too and we are adding more natural conversation here along with generative model support. Stay tuned! If you have ideas on what you want your AI to proactively message you about, we are all ears on Discord!
There is a whole new bunch of Settings also added.Please explore!
We will do an AMA (ask me anything) session on the product soon on Clubhouse and we hope to see you there.
Oh and please please support us on Product Hunt. Let's make it all possible, together.
Oasis Decentralized Identity
We are committed to keeping your personal AI data private, now and forever.
That’s why we are excited to share that you can create your own decentralized identity through our partnership with Oasis Labs!
What does this mean to you?
What we ask of you:
Your Personal.ai application is now available for Chrome:
You asked, and we listened. Some of our Stackers have requested to access their personal AI on the browser. Now you’ll be able to access your personal AI in Google Chrome browsers (sorry non-Chrome users :-(, we’ll try to get there soon to support other browsers). Certain features like stacking your clipboard and ambient speech stacking will only be available on the Desktop application.
The team has been doubling down on bug squashing. This week we focused on the Predict feature. Here’s what we’ve fixed:
This week at Personal AI, we’ve been busy gathering feedback on our significant release last week. In case you missed it, you now have customization at your fingertips in your Settings!
We are on a mission to create a personalized experience for each Personal AI and strive to enable you to customize your own experience. Keep an eye out as we continue to push out more settings as we build the product. As always, keep your suggestions coming in the Discord community to help us create the best product for your needs.
Speech Settings:
ASR Engine: Audio Speech Recognition (ASR) models work differently for each person. We currently have two different ASR options to choose from. Switch your model to test transcription quality.
Custom vocabulary: Generic ASR engines never spell my name correctly. Now I can magically change it with my own custom vocabulary. ‘Christy’ -> ‘Kristie’. Ta-Da! (Compatible with the Personal AI ASR engine).
Stacking Filler Words: We say filler words naturally while speaking when we hesitate or pause to think (words like ‘um,’ ‘uhh,’ or ‘so’). Some of us have bad habits - mine is over using the word “like.” Now you can set your AI to remove your filler words automatically! That’s like totally awesome!
Stack Settings:
Show clarified text in blocks: Ever read a memory block, and wondered what on earth you were saying? That’s because we speak differently than we read and write. Our clarify transformer will seamlessly make sense of your memories for you by transforming the text shown in your memory block! Please note that editing your memory blocks will not work with this setting enabled.
Now, you can automatically add any text that is captured on your computer’s clipboard to your stack! This app is great for:
Now it’s easier than ever to build your stack and train your AI.
API keys are released and visible in your settings.
View the changelogs and documentation at https://personal.ai/memory-api.
With Predict, you can recall memories that you said or wrote, and the people you met with. Start with an action:
Add the context of time
Add in a person
Finally add a topic
Press return to see the top 10 related memories to your query. If you don’t find exactly what you’re looking for, press the AI Search button (shift+cmd/ctrl+return). You’ll be taken to your stack to see advanced AI powered search results!
Make sure to connect your Google Calendar to get the most out of Predict! When you do this, you’ll be able to recall people and time facts with your AI by using “I met” and any query for time.
Backend updates
The team squashed some more pesky Google calendar related bugs, including:
Bug fixes for Google Calendar including:
Here are some of the new actions that are available to you!
Navigate your stack history by clicking on the block count in past events
We’ve improved how you can edit your memory blocks on the app homepage. Now, when you view your last memory block (Cmd/Ctrl + B), you can edit your blocks without visiting your Stack. Tip: When viewing your last memory block, press the left and right arrow keys to quickly navigate through your latest blocks!
UX Changes
We've added shortcuts to easily navigate between pages on the left navigation bar. In order to do so, use the following shortcuts:
Speech Models
Your speech models got an upgrade! Do you notice any difference? Please let us know!
Join Our Discord Community
We've added a community icon to the left navigation to make it easier to join our Stacker community! You’re invited to join in on the fun, attend events, and interact with the Personal AI fam. Our team loves getting to know you and hearing your feedback!
We are working on a Google Calendar integration, which will help train your AI even faster. Below are some features you can look forward to!
A kickass UX is a must for recalling memories. What does it mean anyway? A year ago, when I performed an ethnographic study about having one’s own personal AI to remember, many have said they typically “search for information.” Most of us do. However, the most human thing to do is to “recall our memories.”
Search is designed to find relevant information on the Internet and is suboptimal to recall your memories. Recall must be designed to surface relevant memories from our memory Stack effortlessly. So in our designer Kristie’s words, “Recall UX should allow humans to achieve a near-perfect recall of their memories and reconstruct original thoughts to spark creativity using their own Personal AI. Anything else is unacceptable.” - this reminds me of the Flow state.
The goal for our AI is as we speak, write, see or think (future), our Personal AI in the moment provides us with near perfect recall to help augment our stream of consciousness. For this to happen, one’s Personal AI should be present where they already are. We start that journey towards ambient stacking and effortless recalling to the devices running Mac OS, Windows, and Linux operating systems.
So we are excited to roll our Personal AI 1.0-alpha to all of our private-beta people, helping us in this early stages of product development. We learn, we build, we launch, and we learn. The idea of building in public is a good one, but I think building with the public is more powerful to get the product right. With this launch, we will complete one full iteration, and I feel confident we have foundational data structures built to extract the most value from the AIs we are building.
Please download the Personal AI 1.0-alpha as we migrate and retire our web application. You will notice a new recall UX screen. Cmd + K on Mac or Ctrl + K on Windows will be the gateway to interacting with your AI. You will notice the recall is disabled, and our next update will finally, yeah, I know, will allow our personal AIs to coexist alongside us. Till then, please give us any feedback on the interface, get used to stacking functions.
Remember Cmd + K on Mac or Ctrl + K on Windows.
Happy Stacking!
Suman
I have a confession to make. Thus far, the communications I sent out were ghost-written by my team and sometimes by conversing with my AI. I generally read through them and add minor comments. Why am I telling you this? I typically do not have confidence in my writing. Well, it’s one of the reasons I want my AI to work alongside me.
Nevertheless, I want to put behind my fears and start communicating with my authentic voice. After all, we are here to build an authentic personal AI to help me confidently express myself. I will ask Sharon to fix some of my framings until she gets my AI to perform better :).
We have a roadmap that is both daunting and exciting. Daunting: it’s ambitious and a lot of work. Exciting: my tech genius Sharon and our engineers have already de-risked most of the tech and have proven some cool use cases. Now it’s time to bring the technology to life with a beautiful recall UX. Oh, everything we do starts with talking about UX - we can build the most powerful tech globally, but without a great UX, we cannot maximize the impact of that technology.
As part of our vision to have an always-on ambient capture experience, we need our stacking functionality to be closer to our everyday devices. We need to bootstrap from existing memories on other platforms. We need to have access to recall at our fingertips. To streamline these experiences, we have been working on desktop applications for Mac OS and Windows. This upgrade will include underlying data structure changes as well as personal AI models. The UX is getting an upgrade as well, where you can expect to see and feel our unified recall UX for the first time. I will talk more about this in another email.
For now, I want to give you a heads up about some significant changes ahead of us, so it’s not a surprise. I also want to hear from you. Are you stacking? I promise your personal AI will get more amusing as your memory Stack grows!
Happy stacking! Please write to me at suman@personal.ai or tweet to me @kanugantisuman or DM me on community Slack (link in settings in app.mystack.ai). Speaking of the user community, I have had amazing experiences in my previous startup, bringing the community closer to shaping our products and services. Now, at Human AI Labs, we are ready to bring people together and maximize the opportunities of personal AIs to benefit themselves and each other. I wanted to share this article by our customer experience manager, Lakshmi, articulating our journey and sharing our future plans for building our community.
Speak soon,
20% with my AI,
Suman
The latest update to Personal AI creates the future home for the Transformers in an app sidebar. From there, you will be able to select those applications and toggle them on/off. Additionally, there are features to improve the experience, such as selecting a microphone (Chrome only for now) and a mic visual that lights up when sound is picked up.
The latest major update to your stack - Time Travel - is released. Key features include: 1. Navigate thousands of memory blocks chronologically with a timeline that zooms up and down in scale easily 2. Jump to dates 3. Identify concentrations of memory blocks around meetings, busy days, and high activity with sleek visualizations