
๐ Technology team building @Doximity's Clinical AI Suite ๐ ๐ฉบ
Distributed
Joined November 2015
- Tweets258
- Following44
- Followers607
- Likes20
Pinned Tweet
We are proud to share that Doximity Ask outperformed OpenEvidence, GPT-5.6 Sol, Claude Fable 5, and other leading frontier AI models.
This quoted post is unavailable.
Doximity Tech retweeted
โJust make the model biggerโ isnโt a strategy for clinical AI. Neither is asking physicians to trust an answer they canโt check. Great clinical AI is an architecture, not a model. Today weโre adding a new layer: ๐จ๐ฝ๐ง๐ผ๐๐ฎ๐๐ฒ.
๐๐ผ๐
๐ถ๐บ๐ถ๐๐ ๐ฎ๐ป๐ฑ ๐ช๐ผ๐น๐๐ฒ๐ฟ๐ ๐๐น๐๐๐ฒ๐ฟ ๐๐ฒ๐ฎ๐น๐๐ต ๐ฎ๐ฟ๐ฒ ๐ฝ๐ฎ๐ฟ๐๐ป๐ฒ๐ฟ๐ถ๐ป๐ด ๐๐ผ ๐ฏ๐ฟ๐ถ๐ป๐ด ๐จ๐ฝ๐ง๐ผ๐๐ฎ๐๐ฒ ๐ถ๐ป๐๐ผ ๐๐ผ๐
๐ถ๐บ๐ถ๐๐ ๐๐๐ธ. In the coming weeks, clinicians at health systems that use both will see an UpToDate answer right alongside Askโs AI-generated answer.
Not a logo on a landing page. The actual answer, on the same screen, for clinicians to compare.
It joins the rest of the stack Ask is built on:
ย ย โข Evidence retrieval from peer-reviewed literature and guidelines
ย ย โข Peer-reviewed drug monographs for dosing and interactions
ย ย โข Physician review through ๐ฃ๐ฒ๐ฒ๐ฟ๐๐ต๐ฒ๐ฐ๐ธโข
ย ย โข And now ๐จ๐ฝ๐ง๐ผ๐๐ฎ๐๐ฒ, reviewed by 7,600+ expert clinicians, editors, and peer reviewers
๐ ๐ผ๐ฟ๐ฒ ๐ฒ๐
๐ฝ๐ฒ๐ฟ๐ ๐ฟ๐ฒ๐๐ถ๐ฒ๐ ๐ฏ๐ฒ๐ต๐ถ๐ป๐ฑ ๐ฒ๐๐ฒ๐ฟ๐ ๐ฎ๐ป๐๐๐ฒ๐ฟ. Link below.
Clinical AI should be shaped by the realities of practicing medicine.
Last week, more than 130 healthcare leaders joined us for Doximityโs Customer Showcase, bringing together perspectives from across the industry for conversations about where medicine is headed.
Our panelists shared how their practices are evolving, where AI is already fitting into clinical work, and what they hope these tools can make possible for physicians and patients in the years ahead.
We also heard from @mistyonpointe, whose journey to becoming the first African American woman promoted to principal dancer at American Ballet Theatre was a powerful reminder of what it takes to challenge convention, create change, and open doors for those who come next.
Thank you to everyone who joined us and brought their perspectives to the conversation. The future of clinical AI wonโt be shaped by technology alone. It will be shaped by the people who understand medicine, ask hard questions, and are willing to imagine what comes next.
Doximity Tech retweeted
Today, @Doximity's AI team is open-sourcing mvcc, a compiler and runtime that build and run CUDA C++ on Apple silicon GPUs. Point nvcc or CMake at it and your CUDA code runs on a Mac, with no NVIDIA toolkit, driver or GPU involved. Apache 2.0.
github.com/doximity/mvcc
Doximity Tech retweeted
Another epic Rails World in the books. Nearly 1100 people! Austin did not disappoint, neither did the content.
Amazing work @AmandaBPerino and team as well as all the speakers who put in a ton of time to make this event a delight.
I really enjoyed @robbyrussell, @flavorjones, and @ryan_sherlock talks.
As expected @dhh did not mince words, and the message of optimism for the future echoed loudly around the entire venue.
See you all in Lisbon ๐ต๐น next year.
Migrating a production iOS surface to SwiftUI is easy to describe and harder to execute. Here's how the Doximityโs iOS team approached the migration incrementally, kept feature development moving, and reduced risk along the way.
technology.doximity.com/artiโฆ
#SwiftUI #iOSDev #Engineering
Batch snapshots workedโuntil โfresh enoughโ stopped being enough.
Hereโs how we moved toward near-real-time data, the tradeoffs we faced, and what we learned along the way.
technology.doximity.com/artiโฆ
Doximity Tech retweeted
The loudest claims in Medical AI may sound impressive: perfect scores, zero hallucinations, systems that always know when to decline. Proving those claims is much harder.
๐๐ป ๐ต๐ฒ๐ฎ๐น๐๐ต๐ฐ๐ฎ๐ฟ๐ฒ, ๐๐ต๐ฒ ๐๐๐ฎ๐ธ๐ฒ๐ ๐ฎ๐ฟ๐ฒ๐ป'๐ ๐๐ต๐ฒ๐ผ๐ฟ๐ฒ๐๐ถ๐ฐ๐ฎ๐น. ๐ฅ๐ฒ๐ฎ๐น ๐ฝ๐ฎ๐๐ถ๐ฒ๐ป๐๐ ๐๐ถ๐ ๐ฑ๐ผ๐๐ป๐๐๐ฟ๐ฒ๐ฎ๐บ ๐ผ๐ณ ๐ฒ๐๐ฒ๐ฟ๐ ๐ฎ๐ป๐๐๐ฒ๐ฟ. That's why we owe physicians more than impressive metrics, we owe them rigorous evidence and an honest account of what the system can and cannot do.
๐งตFirst in a series of articles about how Doximity Ask works under the hood.
Clinical AI reliability is paramount.
We share a look under the hood to see how Doximity Ask works to make answers useful in real-world care.
For clinical AI, the model is only one layer. You also need evidence retrieval, clinical workflow fit, privacy guarantees, continuous evaluation, and physician review through PeerCheckโข.
Thatโs the system behind Doximity Ask โ now part of a Clinical AI Suite purchased by 150 health systems, including 8 of the top 20 hospitals.
Hereโs how Doximity Ask works under the hood:
technology.doximity.com/artiโฆ
The bug looked harmless: a memoized Redis lookup. The profiler told a different story: 45,000 unnecessary fetches. Hereโs how one production profile turned a โslowโ Sidekiq job into a 6x faster one.
technology.doximity.com/artiโฆ
Moving an agent from demo to production isn't a model problem, it's a systems problem.
Context, retrieval, tool design, permissions, observability. The model is just one component in the chain.
technology.doximity.com/artiโฆ
Choosing the right data tool doesnโt mean picking just one. At Doximity, we combine LLMs, ML, and statistics to unlock more accurate, transparent, and impactful solutions.
technology.doximity.com/artiโฆ
Doximity Tech retweeted
At @austio36's #RailsWorld talk, he'll share how @doximity_tech replaced a slow, pricey analytics vendor with a custom-built Rails pipeline, scaling to 3,000 events/sec, cutting latency from hours to seconds, and saving hundreds of thousands annually. Ditch vendor lock-in!
rubyonrails.org/world/2025/dโฆ
Doximity Tech retweeted
At Doximity, weโre driven by purpose, collaboration, and the energy of a team that believes in making a difference for doctors.
๐ Whether working independently or coming together, we thrive on curiosity and innovation.
And when we meet in person? The energy is electric. โก
We are so proud to be on @rails!
Introducing Rails Case Studies. Because what better way to show the world that Rails is an excellent choice for building fast, flexible, and scalable applications than by telling the stories of companies like @doximity_tech, who have done exactly that? With Rails, we have a lot of stories to tell, and this is just the beginning. Read more here: rubyonrails.org/2024/12/12/iโฆ
Ever wondered what it's like to balance coding and team leadership? At Doximity, our data managers do just thatโworking with data, cultivating teams, and building strategies without the 80-hour workweeks. Weโre also hiring for this dream role!
technology.doximity.com/artiโฆ
We recently sponsored @tropical_rb and @railsgirlssp , fostering growth and diversity in tech.
๐ Highlights include keynotes from Rails core team members, upcoming Rails 8 features, and empowering women in software development.
technology.doximity.com/artiโฆ
๐ฑ๐ก Learn how our team is simplifying Android deeplink management using annotations and a dynamic routing system. A huge shoutout to our dedicated developers for continually innovating our app's navigation capabilities.
technology.doximity.com/artiโฆ
We thoroughly enjoyed teaming up with The Rails Foundation (@AmandaBPerino) to support the @railsgirlssp event this past weekend hosted at @LeWagonBrasil. What an inspiring group of women, who came to deepen their knowledge of @rails.
Special thank you to the Doximity mentors who helped with the event @davidbrusius, Julio Monteiro, Camila Tormena e @douglasandrade ๐๐ผ
Doximity Tech retweeted
Am still gathering my thoughts and notes (and energy ๐
) after 3 amazing days in Brazil at @tropical_rb and @railsgirlssp with @doximity_tech. When we talk about the Rails Renaissance, Brazil is already on the way. ๐ช






























