Santa Monica, CA
- Tweets 80
- Following 398
- Followers 306
- Likes 1,487
It feels really good to start talking about what we've been cooking at @usearini 👨🍳
🎧 New Podcast: lnkd.in/eWGXTvMC
📞 Here’s what you’ll learn:
✅ Why 20-30% of patient calls go unanswered—and what that costs your practice
✅ How practices are booking hundreds of extra appointments each month with Arini
✅ front desk vs. AI answer calls
7
749
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
Our co-founder and CTO @rami_rustom is giving short talks throughout the day about the current state of AI and its impact on dentistry.
Join us at booth 5212 at the CDS Midwinter conference!
What's AI? How does it actually work? What's currently possible with AI....
1
7
919
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
Too much noise around AI, we cut through that at conferences by showing impact and scale:
* Real-time map w/ live calls
* Testimonials on loop: reducing headcount by ~17% for DSOs
* Live demo
* Dentists who use Arini were around to discuss their journey firsthand
1
3
747
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
Hear the story behind Arini at @DeepgramAI's AI Minds podcast. @abduljamjoom (Arini's CEO) discusses voice ai, healthcare, dental tech, and more!
deepgram.com/learn/podcast/a…
1
1
601
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
🎉 Arini partners with Cartesia to revolutionize dental practices! 🦷🚀
Arini is transforming dental care with their innovative AI receptionist that answers calls 24/7, schedules appointments, and boosts revenue for dental practices.
We're honored to be their main voice provider, powering patient interactions that recover thousands in revenue per call.
Here's why Arini chose Cartesia:
⚡️ Ultra-low latency — Sub-90ms response time keeps patients engaged.
🔊 Realistic voices — Lifelike speech reduces hang-ups.
🔒 HIPAA compliant — Secure handling of patient data.
🌐 Multilingual support — Serving English and Spanish speakers.
👩💻 Developer-friendly — Smooth and efficient API integration.
We're excited to support @abduljamjoom and the exceptional @usearini team in their mission to transform dental practice management.
Read the full story here 👇
cartesia.ai/blog/arini
1
2
20
2,451
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
🚀 @useArini! AI receptionist for dentists🦷
🌐 arini.ai
⭐️ Schedules appointments 24/7
☎️ Answers every call. Drives revenue.
📈 "Increase your production. Cut your costs."
Congrats @abduljamjoom & @rami_rustom!
tryfondo.com/blog/arini-laun…
3
7
1,209
Excited to share what @rami_rustom and I have been working on! Tell your dentists to book a demo with us
arini.ai
YC W24's @useArini is an AI receptionist for dentists that answers phone calls and schedules appointments 24/7, lightening the load for the front desk and helping practices increase revenue.
ycombinator.com/launches/KUK…
Congrats on the launch, @abduljamjoom and @rami_rustom!
1
3
17
2,767
RT @amasad: Western elites — left & right — want you to believe that the Israeli-Palestinian conflict is complicated or fundamentally insol…
55
1
...you should "probably" always stream LLM responses...
in an entertainment chatbot setting, synthetically adding 1s of latency for LLM responses decreased mean conversation length by 3%, while 2s of latency decreased it 6%
faster response are critical for engagement. i bet we'd find a similar stat if we compare streaming vs not streaming responses, where streaming gives the impression of a faster response.
can someone with a large user base please run this study? cc: @chai_research 🙏
arxiv: arxiv.org/abs/2303.06135
2
958
"Rewarding Chatbots for Real-World Engagement..." fine-tunes chatbots to increase Mean Conversation Length (MCL) as a metric for user engagement (~70% improvement).
this makes me wonder what repercussions this has on model safety...
arxiv: arxiv.org/abs/2303.06135
2
4
934
I feel quantifying whether a conversation was "safe" for the user is quit tricky -- it's hard to detect unsafe messages if they don't include profanity for example, but are psychologically manipulating.
1
1
412
Checkout my latest blog post about context priming, a method to personalize LLMs
open.substack.com/pub/wortha…
1
1
3
1,136
In the next few weeks, I'll explore a few approaches to build a generalizable memory management and retrieval system aimed to make agents more reliable. @langchain and @llama_index have done amazing work to layout a lot of the foundations that enable building such a system.
1
1
371
@DrJimFan and team have shown some fascinating results in the VOYAGER paper on how a skill library (memory management and retrieval) along with context priming allows for SOTA Minecraft agents. This inspires my work. More on this soon...
What if we set GPT-4 free in Minecraft? ⛏️
I’m excited to announce Voyager, the first lifelong learning agent that plays Minecraft purely in-context. Voyager continuously improves itself by writing, refining, committing, and retrieving *code* from a skill library.
GPT-4 unlocks a new paradigm: “training” is code execution rather than gradient descent. “Trained model” is a codebase of skills that Voyager iteratively composes, rather than matrices of floats. We are pushing no-gradient architecture to its limit.
Voyager rapidly becomes a seasoned explorer. In Minecraft, it obtains 3.3× more unique items, travels 2.3× longer distances, and unlocks key tech tree milestones up to 15.3× faster than prior methods.
We open-source everything. Let generalist agents emerge in Minecraft! Welcome you all to try today: voyager.minedojo.org/
Paper: arxiv.org/abs/2305.16291
Code: github.com/MineDojo/Voyager
Deep dive with me: 🧵
1
686
Useful post about the future of LLM personalization by @rami_rustom!
Just read a book by @stewartbrand about the MIT @medialab and it’s crazy how 40 years ago they were building voice-controlled personalized AI agents
Got me thinking about personalization w/ LLMs, so I’m writing a series of blog posts. Here’s part I:
worthahavana.substack.com/p/…
2
446
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
Have you ever wondered how game designers create such massive open worlds?
Wouldn't these environments take decades to build by hand?
Believe it or not, there's actually a way to make these worlds build themselves.
Another long 🧵
56
1,664
15,142
4,666,988
Abdul Jamjoom | عبدالرحمن جمجوم retweeted
Software engineer @rami_rustom walks through how we built a collaborative AutoGPT AI assistant prototype in Threads using @langchain and @replit.
Check out more below ⤵️
threads.com/thread/344726562…
1
5
12
1,802