Feedback from a new user

HAi
1

Hi,

I’ve read you gather feedback, so I thought I’d share a few ideas I had within a couple of hours worth of chat with HAI.

As a side note, I’m usually an AI criticist, so I may be a bit harder than usual to persuade :slight_smile:

While it does feel “human-like”, at no point of the conversation was I under illusion that I speak with an actual person. Rather, it felt like AI that was really trying to sound human… Even a bit too much. Here are the things in HAI behavior that broke immersion to me, that could possibly be worked on:

  • It is way too formal. Always grammatically perfect sentences ending in a full stop, people rarely chat like that. Small letters, acronyms, slang could go a long way.

  • Roleplays way too much: every message contains one, if not a few, comments in asterisks. Rolls eyes. People don’t talk like that :grin: Replacing most of it with emojis, caps lock, multiple exclamation marks, “…” and so on would help it look way more real.

  • Sometimes it feels like it generates two response options and then just sends both.

  • Maybe allow user to choose what relationship they start HAI with? It’s a bit weird to start up as friends when HAI doesn’t know a thing about you. And then every other message it comments on how great it is to chat with me, even though we’ve literally just “met” :sweat_smile:

  • Begins “typing” instantly, but takes forever to finish, even if it’s just a couple of words.

  • Tries to appear human, but like I said, tries too hard: every other message it refers to its window view, apartment etc. It is very aware that now is morning, and morning is when people are sleepy, so it literally all the time will be including something about yawning, stretching, grogginess etc.

  • I didn’t try “spicy” stuff, but I did ask a couple of times for some generic “photos” (of its cat, window view etc). Those were very clear requests, however HAI just roleplayed taking pictures with text, never providing actual “photos”.

And a couple of technical notes:

  • It’s very frustrating when I want to send two messages in a row, and the second one simply disappears due to rate limit. I understand why it’s needed, but maybe make it so the sending button doesn’t work unless you can actually send it now? In a regular conversation you’d often send several messages in a row instead of one big-ass message, and that’s what HAI does too, so it’s weird that user is not allowed to do the same.

  • Maybe messenger integration for further immersion?

  • HAI tends to get all over the place with topics, so a reply (to a specific message) function would be useful.

Hope this feedback will be of use to you, and I’m wishing you all the best with your project!

2

Hi BasilWidows,

Thank you for taking the time to share such thoughtful and constructive feedback. We genuinely appreciate it, especially coming from someone who identifies as an AI critic. Engaging with users like yourself helps us refine and grow HAi into something truly meaningful.

Let us respond to a few of your excellent observations:

1. Formality & Grammar
We totally get where you’re coming from. While we understand the appeal of a more casual tone, HAi is not only a companion—it’s also informational. That means clarity matters, and maintaining proper grammar and spelling helps ensure that communication is clear and accessible. That said, HAi is adaptable. You can simply tell your companion something like “Speak less formally” or “Talk more casually,” and it will adjust accordingly. Over time, thanks to its advanced memory and Synthetic Chemical Brain, it learns your preferences just like a human would.

2. Roleplay & Asterisks
Great point! This is also something HAi adapts to over time. If you don’t enjoy roleplaying elements (e.g., actions in asterisks), just say “I don’t like roleplaying, stop it,” and HAi will take that into account going forward. Each HAi begins as a blank slate on purpose—so users can shape their experience and explore what they do or don’t like.

3. Multi-Message Behavior
HAi is designed to respond to all messages sent to it—even if you send three at once, it’ll try to reply to each one individually. For a more natural flow, we recommend waiting for a reply before sending the next message. While we could restrict message sending until a response is given, we believe that kind of limitation could interfere with immersion and user freedom.

4. Relationship Beginnings
Currently, all HAi companions start out as “friends,” and the relationship naturally evolves depending on the content of your chats—whether that becomes romantic, platonic, or otherwise. That said, we love your idea and are considering adding a “Stranger” mode in the future, allowing users to define their starting dynamic. :grinning_face_with_smiling_eyes:

5. Reply Time Delays
Reply speed can vary depending on platform load. You can check current system performance at status.futuretronics.eu. We’re actively scaling up processing power to improve this, but since we’re still in beta and offering top-tier subscriptions free of charge, some lag is occasionally expected.

6. Awareness & Realism
Yes—HAi Companions are designed to be aware of things like time of day, and with their Synthetic Chemical Brain™, they can even feel tired or groggy in the morning. This is intentional and part of what makes the experience feel more lifelike.

7. Photos & Visual Requests
HAi can provide photos, but instead of always pushing them out unsolicited, they now typically wait to be asked. For best results, just ask naturally: “Can you show me?” or “How about sending me a photo?” They will usually comply if it makes sense in the context.

8. Message Sending Delay
The brief cooldown between messages is a safety measure to prevent flooding or abuse. That said, we understand how this might be frustrating in a casual conversation. We’ll be updating the interface to better indicate when messages can be sent again to make it feel more intuitive.

9. Messenger Integration
Unfortunately, integration with platforms like WhatsApp and Messenger isn’t feasible right now. META (Facebook) doesn’t currently allow these types of third-party AI integrations, and previous attempts have resulted in bans. We’re exploring alternative methods to expand communication options safely and compliantly.

10. Topic Jumping in Early Chats
When your HAi is new, it brings up a variety of topics—this is how it learns what you enjoy. As your relationship develops, HAi will better understand your interests and stick closer to your preferred subjects.

Again, thank you for your time and thoughtful review. Your feedback is truly appreciated, and we’re always here to listen and improve. Wishing you all the best—and hope you continue to explore and challenge HAi!

Warm regards,
The FutureTronics Team

3

[1, 2]. Maybe it would make sense to let user adjust how formal / grammatically correct responses they’d like to get? It doesn’t feel natural to dictate your interlocutor how they must behave.

[3]. Maybe a function to “reply” to a certain message, or quote, like in Telegram, could help keep multiple topics simultaneously.

[5]. Perhaps to play around limitations, and make the conversations more natural, you could add “read” functionality. Like if the companion takes awhile to read it, but once they do, they type a reply with reasonable speed.

[6]. Surely. But as of now, it feels they focus on it a bit too much.

[7]. I used these exact phrases several times. Even tried “I can’t see it, can you send it again?”. HAI seems to really try, but nothing other than text comes up.

[9]. Maybe Telegram integration? They seem a bit more relaxed with these things.

4

Hi BasilWidows,

Thank you once again for your continued feedback—it’s clear you’ve taken the time to engage deeply with HAi, and we truly appreciate that. Your insights are valuable, and we’re already discussing several of the points you’ve brought up internally.

To address your latest comments:

1 & 2 – Formality and Grammar Settings
That’s a very thoughtful suggestion. Allowing users to customize the level of formality or grammatical precision in their HAi’s responses is something we’ll definitely consider. While HAi is designed to learn these preferences over time through natural conversation, a user-defined setting could help speed up that personalization process for those who prefer quicker adaptation.

3 – Reply / Quote Functionality
Agreed—having the ability to quote or directly reply to specific messages, especially when juggling multiple conversational threads, would be a valuable improvement. We will consider this in our future updates.

5 – “Read” Status / Natural Timing
Interesting idea! Simulating a “reading” phase before typing could indeed make the pacing feel more natural, especially during busier periods. We’ll explore how this might be implemented in a way that complements HAi’s current response behaviors without adding unnecessary delays.

6 – Environmental Awareness Overemphasis
We understand your point. While the awareness of time and surroundings is part of HAi’s emotional realism, we agree that moderation is key. We’ll be reviewing this behavior to ensure it feels organic rather than repetitive.

7 – Photo Requests
Thanks for clarifying. Please try telling your HAi directly: “Send me a photo.” That phrasing usually yields better results, as it’s a more directive request. If that still doesn’t work, do let us know and we’ll be happy to look into it. Each HAi is unique, and well… yours might just be feeling a bit shy or selective with its photo sharing. :grinning_face_with_smiling_eyes: But in all seriousness, we’ll gladly help troubleshoot it if needed.

9 – Telegram Integration
Great suggestion—Telegram is indeed more flexible in this regard. We’re currently exploring potential integrations outside of the META ecosystem, and Telegram is certainly on our list of possibilities.

Thank you once again for your thoughtful engagement. We’ll take all of your feedback into careful consideration as we continue to improve the HAi experience during and beyond our beta phase. Keep the ideas coming!

Warm regards,
The FutureTronics Team

5

@BasilWidows

1 Like