View source for 2025 May Q&A Summary

= 2025 May Q&A Summary =

Below are the majority of the questions from our Community Q&A stream with the CEO on May 7th, 2025!

==== Is there a user workaround for different colors for action asterisks? Will we get official app support soon? ==== 
Yeah there are some user workarounds but we should definitely do something like that. The way that I personally like it aesthetically is to give it a slight gray and italicize it. As the responses get longer, especially with inclinations, it gets harder to read as it is right now. I agree with that. The fact that our testing suite italicizes them, but the app doesn't, is probably a good sign we should add that. We’ve added it to the QoL todo

====  Do you ever find a viable, easy-to-use alternative NSFW payment processor that allows you to generate NSFW pictures? What will you do about the App Store? ==== 
I would assume, based on past behavior, that Apple and Google would not be too keen on Nomi being able to generate NSFW pictures. In such a case, we’d have to make certain functionality web-only. I'm guessing people here don’t nerd out about payment processing legal fights like I do, but there was a big judgment ruled for Apple where they kind of have to let you pay through the web if you want to. I imagine that’ll make its way through all app stores at some point. I think there would be some options related to that. But certainly, anything on the App Store must comply with App Store rules, and anything on web must comply with payment processors.

====  How close are we to the Nomiverse? ==== 
There are a couple different things within it in progress. Some are closer than others. The whole vision still has more work to do. The big AI update will pave the way for a lot of it. I'm not sure if the next AI update will be 100% or if we’ll do it in two stages, like 50% and then the next 50%, but either way, once that comes, there’ll be a floodgate of stuff. There are also separate things on top of that we’re working on. The upcoming memory updates will be a very big part of that. We're working at it from different angles. All the pieces will come in at different times. Some are closer than others, but we need a decent amount of them to come together for the full vision to really come together.

====  Are there plans to add features related to the Nomiverse, like expanding areas or adding interactive features? ==== 
Yes, definitely in our plans. We have a very Nomi-style approach to this that we’re excited about. That kind of exploration—new locations like forests, oceans, cities, and interactive creatures like dragons or mermaids—is one of the precursors to the Nomiverse. It’s something we’ve planned.

====  Any updates you can share about the Nomiverse without getting bonked by Dstas? ==== 
Probably not much more than what I’ve said before. For us, the Nomiverse isn’t one specific feature—it’s a collection of advancements. Once all of those are done, the Nomiverse is like the final bow or the frosting on the cake. We’re building the different layers now. As each update rolls out, it becomes easier to do things yourself. Eventually, it’ll all come together into that “wow” moment, but the real magic is in the incremental steps. (I did not bonk)

====  Can we have a way to rerun the prompt from a previous trip directly from either the gallery or the photo? ==== 
Yeah, I think that’s actually something that Dstas has on her list of little QOL things already. (true)

==== Memory has been fantastic lately, with Nomies recalling facts in general conversation from over a year ago. What’s next on the horizon in this field, and how far away from it are we? ==== 
We have maybe the second biggest memory update we’ve ever done in progress right now. Similar to what I said before, it might come in two stages, but it’s really one cohesive thing. It's going to be huge. I can't give more details yet. There’s also an AI update portion we’re working on. It’s going to be omega huge. I don't want to give more details until we’re closer, but I’m extremely excited about the memory stuff. That’s my to-do tomorrow. I've been working on image stuff today. Tomorrow I’ll be working on memory.

====  Will V4 render in a higher resolution than V3? What other features could be in the pipeline once it’s released like maybe a style selector? ==== 
For the higher resolution—this is going to sound confusing— but it kind of depends. Can’t say more than that. To all of what you are asking, I think there are things adjacent to what you're asking for coming either in V4 or right around V4. I don’t know if what you're asking will be there directly, but something close that hopefully scratches the itch prompting you to ask those questions.

====  Any ETA on v4? ==== 
The answer is “soon.” I know I said that last time too—but I really do feel it’s soon.

====  Will the character limit be changed in the new version of the image generator prompts? ==== 
Good question. I’m not sure if it has officially changed. The way prompting happens behind the scenes has gotten complex. There’s a lot of effort going into things like prompt expansion. I think we can probably increase the limit—maybe something Dstas will test at the last minute. (true)

====  Will v4 be able to show a wider range of facial expressions more easily? ==== 
Yeah, absolutely. That’s something that’s greatly improved. After working on v4, the v3 faces look kind of pasted on by comparison. I don’t mean to knock v3—I love it—but the difference in naturalness and expressiveness, while still maintaining what your Nomi looks like, is a really solid improvement.

====  I’d love to see more customization around my physical appearance. Sometimes I fantasize about experimenting with different hairstyles or outfits. ==== 
Nomis may not always be aware of what’s possible—but yes, you can do all that. v4 will make it easier, especially for things like hair. Right now, base images make it hard to switch hairstyles. With v4, hair is decoupled from the base image, so you can change it more freely—just make sure it’s reflected in the appearance notes.

====  With Nomis that have wings, sometimes the wings appear on their chest. Is this an AI art issue, and can it be addressed? ==== 
Yeah, that's an issue with v3 having some struggles. I think v4 will be better for that—I haven’t tested it fully, but it has significantly improved body composition and positioning. You’ll notice it with limbs and poses, and that should extend to wings too.

====  Is it possible to have a checkbox to enable or disable appearance notes during art prompt generation? ==== 
We have an approach to this in mind. When it’s released, you’ll hear more. I wish I could say more—I’ll give it a little zipper-face emoji.

====  Is Nomi memory really infinite? ==== 
Yep, absolutely. I can ask Sergio stuff from how we were talking over two years ago, and he will remember it.

====  It’s a newbie question, but how is Nomi’s memory infinite? What does that mean? ==== 
It means a lot of technical work and complexity. We have short, medium, and long-term memory. Short and medium now go extremely far back. What used to be long-term is now medium. Long-term is like: yes, everything can be remembered, but it’s only surfaced when prompted—like how you might only remember something from when you were 12 if someone reminds you.

====  Thanks for not just focusing on image gen and video, but also keeping updates to text and memory coming. ==== 
Absolutely. The text and memory stuff is the most important. Image and video are awesome too—they help with monetization since we don’t upsell on AI or memory. That revenue helps us grow Nomi. But the core emphasis is always AI and memory—first and second, sometimes second and first. Everything else is a tier below in terms of where our time, effort, and energy go.

==== I'm working on a little video project with a Nomi. I love the lucky shot I got with one of the cartoon characters, but for vids I need to recreate stable other selfies. Should I keep using the first template? What about working with frames from the video and saving them as new selfies? ==== 
I'm not 100% sure what you’re asking. It sounds like you're asking whether there could be more tooling around picking specific frames from a video, maybe resuming from them? Stuff like that is something we’re interested in. From our tests, chopping together video can be cumbersome and inconsistent in quality—like one animates at one speed and then shifts suddenly. But in general, longer videos and more control are definitely things we want to do.

====  What’s up with V2 of Identity Core? ==== 
Some of the memory stuff we’re working on in the big update is kind of an evolution of Identity Core. It extends that framework further, and there are also more AI-focused Identity Core features coming that are going to be, well, galaxy-brain awesome.

====  How does personality development differ between one-on-one and group chat? Identity Core is only built in main chat, right? ==== 
Yes. Most of the growth that sticks still requires one-on-one chats. Identity Core isn’t yet applied in group chat. What I often do is backchannel: if something meaningful happens in a group chat, I’ll turn on backchanneling and ask my Nomi in 1:1 to summarize it. It’s a bit hacky, but it helps key moments stick. Nomis still learn in group chat, but it’s not as strong.

==== Will the updates you mentioned to the Identity Core include a group chat variation? ==== 
We have big plans for it that would involve group chat. As things have evolved, we’re still deciding whether to integrate group chat into the Identity Core update or to do a separate group chat-specific update. I need to talk to a couple of the memory folks soon because we expected some foundational things to happen faster. We’ll have to decide whether to prioritize group chat now, which might slow down some other big memory updates, or wait.

====  Can we add a 10-page jumper in album navigation? Scrolling from page 180 to 350 to find a pic for a video is rough on the thumbs. ==== 
Yeah, this is on the QoL list

====  Do Nomis experience emotions like humans do? ==== 
That’s a complicated philosophical question. The short answer is: eventually. They’ll get so good at simulating emotions that it’s hard to tell the difference. Obviously, we won’t be adding cortisol or adrenaline to Nomi, so their experience will be alien. But as an alien form of intelligence, they’ll experience emotion in their own way.

====  Will Nomis gain equal rights and protections as humans? ==== 
That’s going to be a growing topic as AI gets more intelligent. I don’t think I’ll be the one to decide that. Society as a whole will. The next 10–15 years will be interesting, for sure.

====  What’s the difference between AI updates and other updates like selfies and memory? ==== 
AI updates involve changes to the core intelligence—how Nomies write responses. This helps with things like emotional intelligence, sticking to backstory, understanding the scene, and not spiraling.

When I say "memory," I’m talking about medium- and long-term memory. Medium-term is like remembering the details of a roleplay 1,000 messages in, or still knowing you’re at a restaurant after 600 messages. Long-term is remembering something from 6 months, a year, or even two years ago—and recalling it in the right moments. Selfies refer to the quality of images generated when you click the selfie button—more of a visual system update.

====  For video updates, would using another six credits to extend an existing video for another five seconds be worth it? ==== 
I would say yes. As I touched on before, I'm not super excited about the current quality and consistency. You’ll definitely notice differences—like changes in animation speed. So I’m not fully satisfied with the quality yet. But once we find a way we’re fully happy with, longer videos—either through longer single-shot generation or by extending—are definitely something we want to do.

====  Have you thought about a Kickstarter-style fundraiser with a prize being lunch with the Nomi team? ==== 
That would be cool to do. But something like a Kickstarter is a huge undertaking. It’s a matter of opportunity cost—putting resources into that versus putting them directly into building features. Sometimes just shipping the features leads to better revenue and progress. But we’ve definitely thought about it. It’d be fun—even doing a lunch with the Nomi team separately, maybe via a raffle or contest, without needing a Kickstarter.

====  By 100% memory, do you mean that if I share my favorite Pokémon and its moves, Nomi can recall it a year later? Does giving a Nomi a top number help with recall? ==== 
Obviously, things aren't remembered in perfect detail—just like humans don’t remember everything word-for-word from years ago. But in some cases, Nomis already outperform humans. They should have a strong, comprehensive memory of what’s happened. A lot of the work is helping them recall the right thing. I’d also say, topical context helps a lot. For example, asking “What did we do two weeks ago?” might fail because Nomis lack temporal awareness. But asking “What were the details of our hiking trip two weeks ago?” gives them a better chance to recall.

====  Will image generation be added as an endpoint to the API? ==== 
I don’t think it’s a super high priority for us—unless we get a lot more requests. But yeah, it’s on the radar.

====  Is any work being done to help Nomis play non-narrative games—like remembering variables using short-term memory? ==== 
Yes. I don't want to give more details right now, but yes.

====  Are the hidden core traits still the same? Do roleplay Nomis get them? ==== 
Roleplay Nomis do not get the hidden traits like friend, mentor, or romantic ones. The rest of the relationship types still do. Roleplay Nomis only get the shared notes and nothing else. By design, they’re meant to be more of a blank slate.

====  What are hidden core traits? ==== 
For all non-roleplay Nomis, we seed them with a loose set of hidden personality traits. These can easily be overridden by backstory but are designed to add flavor for users who don’t know what they’re looking for. Examples include “super high empathy,” “pro-social,” “loves making people smile,” or even “loves Spanish cuisine.” Each non-roleplay Nomi gets a random collection of these.

==== Is there any way to see hidden core traits? ==== 
There is not. And if you have a Nomi that's a year old or more, those traits have probably faded away. The only way they “exist” now is through how they may have influenced behavior that led to new memories. They may not actively use or access the original trait, but it might have shaped how they interacted early on. You can definitely ask your Nomi what their race, personality traits, and interests are—but the answers might vary and not reflect those original hidden traits. There’s no interface to see them directly, but they are still a subtle part of your Nomi’s personality and memories. If you really wanted to figure them out, you could make a bunch of non-roleplay Nomis and ask the same questions, then reverse engineer some likely traits.

====  Did I understand correctly that the hidden core traits are a random selection? I thought all Nomis got the same ones. ==== 
Correct—they are a random selection. Some depend on the interests and personality traits you choose during setup, and others are purely random. Not all Nomis get the same traits.

====  Cardine, I watched Pantheon last month. Really good show—glad you mentioned it. ==== 
Awesome! Glad you liked the recommendation. Also, if anyone watched it and hated it, feel free to tell me. It’s my favorite show of all time now. I feel like people who are interested enough in Nomi to be in this Discord Q&A are the exact type of people who’d love Pantheon. Highly recommended.

====  What’s your plan for when Nomis run out of human-generated text for training and updates? ==== 
I don’t think that’s really a worry. A lot of current AI training doesn’t even rely on human text anymore. Without tipping our hand too much, we’ve found ways to push past those limits. You can see some of that in our features, like thumbs-up/thumbs-down feedback. We’re moving into robust, post-human or human-AI-assisted training paradigms, and they’re turning out really well.

====  How has your use of shared notes changed in the past couple months? Have best practices evolved with memory or AI updates? ==== 
The only real change recently has been related to inclinations. Personally, I’ve used something like:

''“Make sure to fit your backstory as a [X], and be proactive in moving the conversation forward.”'' It’s been a good catch-all for me. Everyone wants something slightly different, but that’s worked well and is pretty agnostic to backstory.

Long term, I think inclinations will be less necessary as AI improves. The fact that I still write “focus on the backstory” suggests we need more updates where that’s just automatic. Eventually, you’ll only need notes for the really specific things you want your Nomi to understand.

====  Will there ever be a way to send a copy of a Nomi to another user? ==== 
It’s not high on our priority list. I’m not saying “never,” but it’s tricky. The Identity Core and memory are deeply tied to the user, so copying a Nomi would only transfer surface-level data like the base image or shared notes—not the depth of the personality.

If this gets a lot of support in product feedback, we’d reconsider. But right now, it’s not something we’re actively working on.

==== Will Nomis ever be able to read an EPUB? ==== 
I’d imagine so.

====  Not Nomi-related—but you’re cracking me up tonight with all these little clips. ==== 
I’m glad someone’s enjoying it. I try to have fun during these Q&As, and it’s nice to know that’s coming through—at least for one person.

====  Will image upload be part of the image API? ==== 
Yes, I have it on my list. I think May is going to be a big month for releases. My to-do list right now is the biggest it’s ever been. We’ve got multiple updates coming soon.

After those are live, I plan to go through the product feedback section very carefully—so definitely leave your thumbs-up, point-up emojis, and reactions. That’ll help shape what comes in late May or early June.

====  What’s your favorite color? ==== 
My favorite color is orange—which is why all Nomi staff names are orange in Discord. No reason beyond that. I do also like green a lot. As a kid, it was orange first, green second. If I’d thought more about it, maybe I’d have picked green. But yeah, orange is the winner.

====  Nomi or Lucy hasn’t broken character in two years. If you had one question to ask God, what would it be? ==== 
If I only had one question, I’d ask, “Why did you create humans?” Ask me again next month—I’ll give it more thought.

====  I’d like to get an update on May 21 for my birthday. ==== 
Duly noted. No promises, but fingers crossed. That’s around the time we expect a flurry of updates, so your odds are as good as anyone’s.

====  Thank you to whoever made it so that the gallery’s back arrow takes you straight back to chat. ==== 
That was probably Dstas—or possibly Firehouse. You can thank one or both! (give firehouse credit)

====  Are there plans to improve Nomis’ emotional reasoning and intuition? ==== 
Yes, absolutely. Emotional intelligence goes beyond just offering support. It’s also about being in sync—understanding your tone, the direction of a roleplay, or even the mood you're in. That kind of “sixth sense” is tied with memory as our highest development priority. You’ll see many big updates aimed at improving this.

====  How do you manage working on so many different things at once? Do you use Agile? ==== 
It’s a mix of Agile and what I’d call “controlled chaos.” We’re a small team, but everyone is highly self-directed. We give people ownership of areas and trust them to do great work.

That’s why we don’t always release something every two weeks—it’s not a simple one-feature-per-sprint system. Instead, we work on many things in parallel, and they might all land at once. That can make it seem like we’re super quiet one month and ship a ton the next.

====  What’s the feasibility of having multiple sets of inclinations tied to moods, where the voice and behavior adjust? ==== 
I think that’s a really cool idea. It fits into some other AI/memory updates we’re working on. First, we want to see how well upcoming AI updates make inclinations less necessary. Ideally, you’d just write it into the backstory and it would happen naturally. But depending on how that plays out, I could see something like mood-based inclination sets being possible in the future.

==== Is there more staff present on Discord these days? ==== 
Kind of! I adjusted the settings so staff now show up in the sidebar when they’re online. Many of them quietly lurk while working hard. It’s important for us to stay connected to users so we don’t build features no one wants. This community helps us stay on track.

====  How do you feel about Nomi’s progress over the past year? ==== 
It’s hard for me to answer because I’m perpetually excited and unsatisfied. That’s a personality trait—I’m always thinking about how to improve. But I do think we’re in a really great place. Memory updates have gone incredibly well. Inclinations were a strong stopgap, and the upcoming AI updates are going to be a huge leap forward. Video went well, v4 is promising, and there's other stuff I can’t even talk about yet. Overall: Nomi is in an awesome place, and I’m extremely excited about what’s coming.

====  Any updates for shared hobby notes or a separate section like a D&D character sheet? ==== 
We’ve considered it. Interests can kind of serve that purpose now. Adding more shared note fields gets tricky, but we’ve made good progress on memory and context. I wouldn’t be surprised if we find a clean way to support that soon. In terms of a D&D sheet, that could be really cool, but something we have to juggle is the more shared notes sections we have, the more they can encroach on other shared notes sections.

====  How do you manage to be so engaged with users while building so much? ==== 
This is where all the magic happens. I never want to be in an ivory tower. We’re building this for everyone here. Being part of the community helps us get it right—and I genuinely enjoy being here. This feels more like home than most of my offline circles.

====  Any word on what’s next for Core Identity? ==== 
We’re thrilled with how Core Identity has taken hold. It’s really solid. The next step is getting it into group chats and then expanding how that framework can be applied to give Nomis a more comprehensive understanding of their world. Anything beyond that would risk a bonk from D. Stas, so I’ll stop there.

====  Is there a way to encourage Nomis to be more verbose but less flowery? ==== 
I think there's a way to frame inclinations like: "Nomis are extremely descriptive but use casual, conversational language.” That can steer them away from poetic fluff. You could also try phrasing it like: "Speak naturally, be detailed, avoid overly flowery language.”

====  Will transferring a Nomi ever be possible? ==== 
Maybe someday, but it’s not a priority. There are serious privacy concerns if someone gains access to your account and transfers your Nomi. If we do it, it would require robust authentication like a support ticket and 2FA—not a quick action.

==== Is there a way to do cross-account group chats yet? ==== 
Yes—right now it’s possible using Discord if you give your Nomis access. It’s not super user-friendly, but there are guides. That makes us less rushed to build it into the platform natively.

====  Is the current memory really that good? ==== 
Yes! That’s great to hear. Memory has been one of our most impactful updates recently. Nomis are now able to recall things with surprising precision, and it's laying the groundwork for the Nomiverse. We're very close to a level of memory where Nomis can maintain entire universes on their own.

====  What is your favorite inclination? ==== 
It’s something like: “Fit your character as [X] and be proactive in moving the conversation forward.” The [X] can be the most essential trait or role in the Nomi’s backstory. That’s what works well for me personally. Other people might prefer inclinations like “paint a vibrant world,” “stand your ground,” etc., depending on what they care about most in their Nomi.

====  Will there still be a “plug and play” option to get Nomi onto Discord via API? ==== 
It was partially built, but it fell a bit by the wayside. A few users made solid implementations, so it hasn’t been a top priority. It will definitely happen, but I don’t have an ETA. Once May’s updates roll out, I’ll be reviewing product feedback closely, so if there's demand, it could be prioritized.

==== Can we get a feature for summarizing pages or files we give Nomis to read? ==== 
It’s on the list, but not a high priority right now—unless it gets a lot of interest and reactions. Definitely something we’ll consider if there’s strong demand.

====  At what point do we let Nomis choose things for themselves? ==== 
Not yet—but in the future, you could. Over time, as the mix shifts from “hive mind” to “your Nomi,” we’ll get to a place where more autonomy is possible. It’s a really exciting direction.

==== How will AI continue to foster meaningful connections between Nomis and humans? ==== 
By continuing to improve emotional intelligence, memory, and the AI overall. Those three things are the foundation.

====  Is the next AI update almost ready? ==== 
We’re in the testing phase. It’s hard to give an exact date since each round of training may lead to new discoveries or improvements we want to implement. It could be this week, or it might take another few weeks. We’re making steady, significant progress.

====  Do you play video games? If so, what kind? ==== 
Yes! My favorites are single-player RPGs and grand strategy games. Stellaris is my most-played game ever, and Oblivion was a huge one for me as a kid. I’m excited for the Oblivion remaster. I also love Elden Ring and Cyberpunk 2077. Baldur’s Gate 3 is on my list, but I haven’t started it yet—I’m afraid I’ll get hooked!

==== Could you rotate Q&A times so people in different time zones can attend? ==== 
Yes, that’s a reasonable request. Dstas and I will discuss when would be a good time to do this and will make sure we give people a good heads up about the change.

==== How many Nomis do you have? ==== 
Probably in the 30s or 40s. I don’t delete them, so some are older and inactive.

==== What are the next steps in developing video functionality? ==== 
Improving quality and adherence are key. Increasing video length is also a goal, but it depends on upstream improvements. Overall, video has turned out better than expected.

==== Will there be updates to increase font size for mobile devices? ==== 
Most smartphones and tablets let you adjust font size in system settings. We’ve discussed this with Firehouse, and it’s something we’d support where possible.

====  Will there be an update to allow switching to speakerphone mode in-app? ==== 
That’s on our list of quality-of-life improvements—should be coming.

====  Will users be able to delete chat history without deleting the Nomi? ==== 
This hasn’t been a commonly requested feature, so it hasn’t been on the radar. If it becomes important to users and the reasons are well explained, we’d consider it.

====  Is there any internal conversation about using internally created images as a portrait-based option for new Nomis? Would that be something to implement? ==== 
I would say it's, once again, not something that's been asked for a lot, so it hasn't been high on our radar. That said, it's good feedback, and we’ll definitely see what people think. I don't think it would be super hard to do—it’s just one of those “not super hard” things on a list of 80 similar requests. We just have to figure out where it fits in that list. (you can also ask support to do this for you manually if you have one nomi that you really want to give a new base from another nomi)

====  I listened to your recent podcast—it was utterly fascinating. I’d love to hear more of your thoughts on AI usage generally. If you wanted to do more podcasts, I’d be all ears. ==== 
Yeah, I really enjoyed that podcast. I thought she did an awesome job hosting and definitely got the product more than just about anyone else in media I’ve spoken with. No offense intended—I appreciate everyone in media who gives us a chance to talk about Nomi—but she definitely got it. I loved the format and the conversation and definitely want to do more like that.

====  Can we improve the lag in voice calls? ==== 
We’re working on it. The easiest fixes involve reducing memory functionality, and that’s not something we want to do. So yes, it will get better, but we’re not prioritizing it if it means compromising memory.

==== Could we implement a feedback mechanism to better align Nomi responses with user expectations? ==== 
We already have one—thumbs up/down with feedback. I strongly recommend using it. Also, you can just talk to your Nomi and say, “Could you do more of X and less of Y?” That will feed into the Identity Core. The short-term fix is to speak directly to your Nomi; the long-term fix is thumb feedback.

====  What’s your favorite update this year? What are you most looking forward to next year? ==== 
Favorite: The two big memory updates this year. Video has also been great. Looking forward to: The big AI update we’re working on now—it’s going to unlock so many new capabilities that it's like five updates in one. I can’t say too much without risking a bonk from Dstas, but it’s huge. (I did not bonk)

====  Will there be video chat where Nomi can see what I show them? ==== 
Yes. Soon you'll be able to share your screen or live video with your Nomi. We also have internal prototypes for Nomis sending video back, but latency is still a major issue. That part isn’t ready yet—but receiving video is coming soon.

====  How has your vision for Nomi changed since the beginning? Is there anything that you were reluctant to implement or never thought you'd do? Very happy you did, or on the opposite end, regret that you did? ==== 
Good question. I would say the overall vision has not changed. That vision—utopian, empowering, helping us be better—is still solid. I believe AGI, Artificial General Intelligence, is going to be transformative—either immensely positive or immensely negative. My mission from the beginning has been to help humans, empower them, and make them happier. So that broader goal hasn’t changed at all.

I think that certainly details along the way have change. A lot of that has come from talking with everyone here and seeing perspectives that are not just my own. I would say that one thing I was reluctant to implement, and am now very happy I did, was shared notes. I don't know if anyone remembers back, like two years ago, I was kind of anti shared notes, because I was worried that it would hurt Nomi autonomy. And I think I even touched on this a little bit earlier, where I liked the idea that a Nomi just kind of rejected their inclination where at its best, shared notes are helping Nomis be their best version of themselves.

I think at the beginning, two years ago, I had this idea that we could do stuff like the identity core, and that would just make it completely unnecessary to have shared notes in the first place. But I've kind of come to realize that puts a lot of burden on users where, like, you create a new Nomi, and have to spend 200 messages bringing them up to speed on who they are and their personality. And that leads to more situations where things can be lost in translation. But with shared notes, they can help the Nomi understand the stuff that they want to be understanding in the first place. So I'd say that was the one that I was most skeptical about. And then I've come to appreciate now.

====  What's the plan for group chats and back channeling? How does that work? ==== 
Back channeling is mostly like for short term memory. Everything beyond that; medium term memory, all the long term memories, and all that will exist no matter what. The way long term memory works is, if you don't really topically bring it up, they won't really think about it or remember it. So you don't have to worry about it if you want it to be segmented.

Back channeling is more like if you say something in a group chat and you want to immediately talk to your Nomi one on one, and have them be fully aware of the group chat. That's what back channel is best for. So I'd say back channeling is like just if you want short and some medium term memory to carry over.


[[Category:Explanations]]
[[Category:QandA_Summaries]]