Today, we’re introducing Flows by Allia; a new way for mental health practices to automate repetitive work, directly inside the EHR.
Most practice software is designed to record what happened.
We want it to help determine what happens next.
Running a mental health practice involves hundreds of small operational decisions: what needs to happen, when it should happen, and what to do when it doesn’t.
Each decision is simple. Managing all of them, across every patient, is not.
That’s why we built Flows.
Flows lets a practice turn its own operating logic inside the EHR; defining the triggers, conditions, actions, and timing once, then allowing Allia to carry that process forward.
This is about more than automating a few administrative tasks. It’s a step toward an Allia that understands how each practice works and actively supports the operational layer behind care.
We’re excited to see what practices build with it.
Watch it in action below.
We’re building Allia to make high-quality behavioral healthcare dramatically more accessible and ultimately build the care platform people can rely on for their mental health.
We’re looking for a few really strong marketers in SF:
- Someone who can help define how different stakeholders understand Allia and build the marketing function around it.
- Someone who knows how to take different products across different ICPs to market
- Someone who knows consumer side growth deeply; acquisition, lifecycle, conversion, retention, and brand.
If this sounds like you, or you know someone great, DM me.
Jev is amazing, but what is very funny is twitter discovering classifiers and calibration. Good time to rebrand AlexNet as a “structured output instant VLM” and raise some money!
Everyone, repeat after me.
- Jev isn't a magical universal classifier. It is a very good general-purpose decision model for a particular class of problems: turning unstructured state into choices, scores and probabilities. For a stable domain where you have good labeled data, you can still train a task-specific classifier that beats a generalist on that task and potentially run it extremely cheaply.
- Using pretrained language models for classification is not new. We've been fine-tuning pretrained transformers for downstream classification since BERT, and doing zero-shot classification from natural-language labels. We've done the same ourselves for psychiatric trait classification and dimensional scoring using Llama/Qwen-family models as bases.
- The innovation lies more so in GTM here than anything. It's very interesting to see how something quite simple can also become a model class of a kind, solve a real use case for a specific type of audience and be really good at it. so kudos to @typesafeai and i am very curious about how it evolves from here.
(I don't know if anyone remembers but there was a specific "model" a couple months ago that was 60x faster than openAI and had a massive "waitlist" too.)
We’re launching the Allia Partner API.
Partners can now connect their products to Allia’s provider network and bring care access directly into their own experience.
Whether you’re a primary care organization looking for a trusted behavioral health referral partner, a health system connecting patients to Allia, or an employer or health plan expanding access to care; we’d love to work with you.
With the API, you can:
- Find providers and check availability
- Create patient records and manage care teams
- Book, reschedule, and cancel appointments
- Give patients access to virtual visits through telehealth join links
From finding a provider to booking a visit, you can connect patients to care within the products they already use.
Interested in building with Allia? Let’s connect.
An absolute delight to have partners like @pioneer_fund as we build an ecosystem that gets patients better while aligning incentives for everyone involved!
🧠 Allia Health (YC S26) is building the first full-stack, AI-native clinical group for mental health. 8,000+ clinicians already run their practice on it! Allia brings those independent practices into a single national group without asking any of them to give up their independence.
Allia gives practices the coordination, infrastructure, and payer contracts of a connected health system. Ops run in the background as care is delivered to the patient, which is the most important part! If a patient needs more care than they're getting from their therapist, a referral moves through the group and gets tracked to completion instead of getting lost in the void. Over 600 clinicians across 70 locations and in 32 states are already in the group.
Amie Leighton dropped out of Oxford to build this with cofounder @sarooshkhan98, who leads engineering with agents that read across the whole group. Congrats Amie and Saroosh, @pioneer_fund is looking forward to a great future together!
We've been on the lookout for founders that could grow their startup to a billion dollars in revenue.
I think we found it.
Amie Leighton and Saroosh Khan are building Allia Health.
Allia Health is building the first AI-native medical group for mental health. It's spent the past year giving out it's AI-native EHR system for free, now used daily by 8000 mental health clinicians.
600 of those clinicians will transition into their medical group in Q4 this year. Anticipated, annualized revenue will go from zero to 50M. And more will transition next year.
Here's the memo: https://t.co/4Kdz1BxB8j
If you have intros to medical group strategic angels or venture funds, please send them my way. It is a coveted round right now.
One note, spend just 5 minutes listening to Amie talk about her experiences and and why she and Saroosh are so motivated to make it easier to help clinicians get to better mental health outcomes with your patients.
You'll be so optimistic about the future they're building and blown away by their grit.
Sooooooo honored to be on their team.
@suchenzang so we train a model to have deep understanding and dense representations, only to use an exponentially smaller model to understand those representations (or attempt to) and then what exactly?
Good old PCA.
@rabib_ai Hmu. My cofounder has lived with ADHD and navigated the maze of medications at the same time so there's a lot she can help you w.r.t. lived experience!