We all might need some help with our mental health. I used Ash and was blown away by how helpful it was at 2 AM. It remembered our conversations, and even challenged me, providing valuable insights. Isn't it crazy how AI can make us feel more human?
We're thrilled to introduce Ash: the first AI designed for therapy. This is deeply personal to us. We've struggled, tried to get help, and realized that most of us have the same problem. We decided to train a new foundation model for psychology and build Ash on top of it as a way of helping a lot more people.
We raised $93M from a16z, Felicis, Radical, Forerunner, TMV, and many mission aligned investors to help us do this the right way.
π§ Why AI therapy built differently matters
Most AI assistants validate everything you say and give you the answer you want. That's great for productivity, terrible for growth. Therapy pushes you to see things differently.
Here's what makes Ash different:
Trained on therapy, not just the internet. Ash is built on our foundation model for psychology, which was trained on clinically relevant data. Itβs learned from every approach β CBT, DBT, psychodynamic, you name it. When you talk to Ash, you're getting collective therapeutic wisdom, not generic advice.
It won't always agree with you. When you say "I'm angry," ChatGPT says "I'm sorry you're feeling that way." Ash might ask "Why is anger a bad thing?" Ash can create productive friction that leads to breakthroughs.
Learns what actually helps you. Through reinforcement learning, Ash pays attention to what creates real change β not just what makes you feel good in the moment. It notices when you open up, when you have insights, when you actually follow through on goals.
β¨ What people are experiencing:
"Definitely something everyone should try. Only used two times so far and already it has brightened my inner flame that has almost completely burned out."
"I just wanted to tell you all thank you one more time... I've decided to go back to school to get my doctorate in psychology. Thanks to you guys for this, I'm learning how to become the best version of me⦠Thank you for changing and saving my life."
"We solved a problem I have been dealing with for a week in 5 minutes."
π― The details that matter:
Voice or text: Sometimes you need to talk it out, sometimes you need to write. You can pick.
Different styles: Ash can adapt styles to figure out what works for you
Remembers what you say: Ash remembers what you say between sessions so you donβt have to repeat yourself.
Weekly insights: See your patterns and progress over time
Actually private: Your data stays encrypted, and only opt-in anonymized data is used for improvements
π Our mission:
There are 10,000 people for every therapist. Waiting lists stretch a year. Even if you find someone, there's no guarantee they'll understand your background, share your values, or fit your budget. And honestly? Most people only ever make it to one therapy session.
We're not replacing human therapists. We're creating a new modality of care that's available 24/7 and learns what works for you.
The team: We were founded by machine learning engineer Daniel Cahn (started coding at 8, AI research in mental health at Imperial) and me (co-founded Casper, helped millions sleep better, and then realized I wanted to improve my own, and many others' mental health).
We've had 50,000 people in our beta, and today, we're opening up to Product Hunt.!
Mental health isn't a 10% problem β it's an everyone problem. If you've ever thought "maybe I should talk to someone" but didn't know where to start, Ash is here.
Give it a try and let us know what you think. We read every piece of feedback (seriously, ask our beta users π ).
What stands out is creating productive friction instead of just validation - challenging perspectives for real growth rather than momentary comfort is what separates genuine therapy from feel-good conversations.
Here's to making mental health support truly accessible!
Report
@neil_parikh5 CongratulationsοΌ I wonder when Ash will be able to speak multiple languages.π₯Ί
Report
@curiosita I want my mom to be able to use Ash too.
@kate_ramakaievaΒ Yup, our clinical team is led by digital health pioneer Dr. Derrick Hull with 40+ clinicians contributing! Our advisors include Tom Insel, forrmer head of the NIMH and Devika Bhushan, former surgeon general for CA, among others!
We pre-trained Ash on the largest corpus of clinically relevant data, fine-tuned and aligned the model with our clinical team to adapt to the new modality of AI therapy, and train Ash with reinforcement learning based on what actually helps our users.
Scaling up access to quality mental health support is so important, particularly for those who do not feel seen or heard in traditional settings. Is Ash able to recognize if a user may need more urgent help and suggest resources accordingly? @neil_parikh5
@kyle_rushΒ Yes! We first train on a large-scale, specialized, clinically relevant dataset that we collected, especially through partnerships. Our clinical team then aligns the model to adjust to the new modality of AI therapy (e.g. what topics should the AI bring up? How should conversations end? What boundaries are necessary, that may be different than for a human?) And finally, we train with reinforcement learning based on what works.
You can hear a bit more about the technicals in our interview with ML Street Talk and we'll be sharing results from our first study very soon!
Congrats on the launch @neil_parikh5 and team! Excited to see AI therapy built with real clinical insight β do you plan to integrate Ash with other mental health apps or trackers in the future?
@kate_prasniakΒ We might! At the moment, we've deliberately designed Ash to be standalone, to ensure users stay in control of their own story, rather than to focus too much on the quantitative side (you can see our values: autonomy, competence, and relatedness). That said, we all love wearables and it could be great to share more data with Ash!
Ash
Hey Product Hunt fam! π
We're thrilled to introduce Ash: the first AI designed for therapy. This is deeply personal to us. We've struggled, tried to get help, and realized that most of us have the same problem. We decided to train a new foundation model for psychology and build Ash on top of it as a way of helping a lot more people.
We raised $93M from a16z, Felicis, Radical, Forerunner, TMV, and many mission aligned investors to help us do this the right way.
π§ Why AI therapy built differently matters
Most AI assistants validate everything you say and give you the answer you want. That's great for productivity, terrible for growth. Therapy pushes you to see things differently.
Here's what makes Ash different:
Trained on therapy, not just the internet. Ash is built on our foundation model for psychology, which was trained on clinically relevant data. Itβs learned from every approach β CBT, DBT, psychodynamic, you name it. When you talk to Ash, you're getting collective therapeutic wisdom, not generic advice.
It won't always agree with you. When you say "I'm angry," ChatGPT says "I'm sorry you're feeling that way." Ash might ask "Why is anger a bad thing?" Ash can create productive friction that leads to breakthroughs.
Learns what actually helps you. Through reinforcement learning, Ash pays attention to what creates real change β not just what makes you feel good in the moment. It notices when you open up, when you have insights, when you actually follow through on goals.
β¨ What people are experiencing:
"Definitely something everyone should try. Only used two times so far and already it has brightened my inner flame that has almost completely burned out."
"I just wanted to tell you all thank you one more time... I've decided to go back to school to get my doctorate in psychology. Thanks to you guys for this, I'm learning how to become the best version of me⦠Thank you for changing and saving my life."
"We solved a problem I have been dealing with for a week in 5 minutes."
π― The details that matter:
Voice or text: Sometimes you need to talk it out, sometimes you need to write. You can pick.
Different styles: Ash can adapt styles to figure out what works for you
Remembers what you say: Ash remembers what you say between sessions so you donβt have to repeat yourself.
Weekly insights: See your patterns and progress over time
Actually private: Your data stays encrypted, and only opt-in anonymized data is used for improvements
π Our mission:
There are 10,000 people for every therapist. Waiting lists stretch a year. Even if you find someone, there's no guarantee they'll understand your background, share your values, or fit your budget. And honestly? Most people only ever make it to one therapy session.
We're not replacing human therapists. We're creating a new modality of care that's available 24/7 and learns what works for you.
The team: We were founded by machine learning engineer Daniel Cahn (started coding at 8, AI research in mental health at Imperial) and me (co-founded Casper, helped millions sleep better, and then realized I wanted to improve my own, and many others' mental health).
We've had 50,000 people in our beta, and today, we're opening up to Product Hunt.!
Mental health isn't a 10% problem β it's an everyone problem. If you've ever thought "maybe I should talk to someone" but didn't know where to start, Ash is here.
Give it a try and let us know what you think. We read every piece of feedback (seriously, ask our beta users π ).
Here to answer any questions!
π Neil & Daniel
Visla
@neil_parikh5Β Very cool, definitely going to look into it.
@neil_parikh5Β Incredible work! π
What stands out is creating productive friction instead of just validation - challenging perspectives for real growth rather than momentary comfort is what separates genuine therapy from feel-good conversations.
Here's to making mental health support truly accessible!
FuseBase
This is such important missionπ±
Did you collaborate with licensed therapists or psychologists to shape how Ash responds in sensitive or nuanced conversations?
Ash
@kate_ramakaievaΒ Yup, our clinical team is led by digital health pioneer Dr. Derrick Hull with 40+ clinicians contributing! Our advisors include Tom Insel, forrmer head of the NIMH and Devika Bhushan, former surgeon general for CA, among others!
FuseBase
Spark 2.0
Been using Ash for 4 weeks, and it's been transformational. I've spent my entire life building great products - I know one when I see it.
First of all, I genuinely believe the world needs this - especially now, in times of chaos, war, and uncertainty.
Hereβs what I believe is broken in therapy today:
Finding the right therapist is hard
Very few are actually great
Itβs expensive
And theyβre not available at 2 a.m.
Ash is on track to solve all of these. Massive respect to the team for doing the right thing - and doing it under solid clinical supervision.
What I love most about Ash:
It pushes and challenges me - unlike ChatGPT. (Ash is training its own foundation model, grounded in clinical data and real therapy transcripts)
It gives genuinely insightful summaries after each session
It remembers what we talked about weeks ago - and brings it up when it matters
The whole experience feels unlike anything else out there
Congrats to the team!
P.S. Itβs free right now - but Iβd happily pay $50/month for this.
Ash
Thanks Denys!
Ash
Hey Product Hunt! π Technical co-founder here.
We pre-trained Ash on the largest corpus of clinically relevant data, fine-tuned and aligned the model with our clinical team to adapt to the new modality of AI therapy, and train Ash with reinforcement learning based on what actually helps our users.
Super excited to share Ash with the world!
Spark 2.0
@daniel_cahnΒ πππ
Smoopit
Scaling up access to quality mental health support is so important, particularly for those who do not feel seen or heard in traditional settings. Is Ash able to recognize if a user may need more urgent help and suggest resources accordingly? @neil_parikh5
Astrid: Personal Shopping Agent
This is amazing. I can't wait to try it. It seems like you guys either trained your own model or fine-tuned a model. Can you say more about that?
Ash
@kyle_rushΒ Yes! We first train on a large-scale, specialized, clinically relevant dataset that we collected, especially through partnerships. Our clinical team then aligns the model to adjust to the new modality of AI therapy (e.g. what topics should the AI bring up? How should conversations end? What boundaries are necessary, that may be different than for a human?) And finally, we train with reinforcement learning based on what works.
You can hear a bit more about the technicals in our interview with ML Street Talk and we'll be sharing results from our first study very soon!
Migroot
Congrats on the launch @neil_parikh5 and team! Excited to see AI therapy built with real clinical insight β do you plan to integrate Ash with other mental health apps or trackers in the future?
Ash
@kate_prasniakΒ We might! At the moment, we've deliberately designed Ash to be standalone, to ensure users stay in control of their own story, rather than to focus too much on the quantitative side (you can see our values: autonomy, competence, and relatedness). That said, we all love wearables and it could be great to share more data with Ash!