AI for Science @Microsoft • 2x Founder • @ForbesUnder30
San Francisco, CA
Joined October 2013
- Tweets1.2K
- Following488
- Followers1.2K
- Likes2.4K
Pinned Tweet
As model companies absorb more product surface area, the strongest startups may be the ones that can prove an AI output is correct.
That’s the verification moat. I wrote about why it could be AI’s most durable advantage.
Nirman Dave retweeted
We now have state-of-the-art pre-trained base models in the 3T-parameter weight class, built entirely in-house at MAI.
Post-training results coming soon -- stay tuned!
Lots of details in our technical blog post.
microsoft.ai/news/frontier-p…
LFG
We're excited to announce our collaboration with @microsoft to bring Heaviside and AI-powered electromagnetic design to the RF, photonics, and semiconductor ecosystem on Microsoft Discovery.
Every electromagnetic design problem reduces to the same question: how will this design actually behave once built? Heaviside was built to solve the verification problem, predicting the full 3D electromagnetic field response in milliseconds and within a decibel of commercial accuracy.
We are thrilled to be collaborating with the incredible team at Microsoft Discovery. Learn more here: arenaphysica.com/articles/he….
Nirman Dave retweeted
Introducing Microsoft-Decision-1, our new model for fast decision-making.
It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality.
We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.
Nirman Dave retweeted
We're launching the most accurate real time transcription model in the world... #1 !!!
55% faster and 60% cheaper than ElevenLabs.
Come build agents on our platform!
Microsoft AI has released MAI-Transcribe-2-Streaming, taking the #1 spot for Final Transcript accuracy and First Partial Transcript accuracy on AA-WER Streaming with 2.5% WER at 0.13s after end of speech
MAI-Transcribe-2-Streaming is @MicrosoftAI's new streaming Speech to Text model, joining the non-streaming MAI-Transcribe-2. It leads streaming Final Transcript WER at 2.5%, ahead of the previous #1, SpaceXAI's Grok Voice Transcribe 2.0 at 2.7%, and returns that transcript in 0.13s rather than 0.49s. It is available for streaming transcription at $0.54 per hour of audio, at the higher end of pricing among the leading streaming models.
Key takeaways
➤ Final Transcript: MAI-Transcribe-2-Streaming achieves 2.5% WER at 0.13s after end of speech, ranking #1 of 38 models. It is more accurate and faster than Grok Voice Transcribe 2.0 at 2.7% and 0.49s, Muse Voice Transcribe at 3.1% and 0.16s, and ElevenLabs Scribe v2 Realtime at 3.6% and 0.14s. It is also more accurate, though slightly slower, than Cartesia Ink Preview (external endpoints) at 3.1% and 0.11s
➤ First Partial Transcript: The model achieves 2.5% WER at 0.12s, ahead of Grok Voice Transcribe 2.0 at 3.4% and 0.49s, and ahead of Muse Voice Transcribe and ElevenLabs Scribe v2 Realtime on accuracy, both at 3.6%, and slightly faster than both at 0.12s versus 0.13s. It is more accurate but slower than Cartesia Ink-2 (external endpoints) at 4.0% and 0.07s
➤ Price: MAI-Transcribe-2-Streaming costs $0.54 per hour for streaming, or $9.00 per 1,000 minutes. This puts it level with Gemini 3.5 Transcribe Live at $9, above the $6.50 charged for ElevenLabs Scribe v2 Realtime and Deepgram Flux, more than twice Cartesia Ink-2 at $4, and three times Muse Voice Transcribe at $3. Non-streaming transcription costs $0.10 per hour, or $1.67 per 1,000 minutes
Congrats to the @MicrosoftAI team on the launch!
See more details below
Once you see the enterprise world, every startup in SF looks like a baby.
Serious answer: yes. Before joining Microsoft, I knew maybe five people who used Copilot. I lived in a San Francisco tech bubble, and all of us worked at companies with fewer than 5,000 employees.
Microsoft 365 Copilot now has over 30 million paid seats and is growing fast. It's also a significant market penetration if you look at the number of seats / knowledge workers in Enterprises
Working at MSFT has made me understand why. Large companies need AI that works with their existing data and meets their privacy and compliance requirements, often across multiple countries. Add governments and regulated industries, and it gets even more complex. Copilot can handle that.
So people use Copilot for real work, and yes, most people love it. I use it every day now: Copilot CLI for coding (with access to models from Gemini, OpenAI, xAI and Anthropic!) + Copilot Cowork for regular chat Etc. I’ve been happy with it.
I think the perception gap on this website comes from the San Francisco tech bubble. If you’re a tech bro like me, you mostly know people at smaller companies. You don’t see much of the enterprise world where Copilot is used.
Nirman Dave retweeted
Is AI actually speeding up drug discovery?
According to new McKinsey research, there are early signs that it is: all of these candidates were AI-enabled, with discovery time cut by ~15-80%.
(Incredibly bullish)
Nirman Dave retweeted
Any pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it's not worth pursuing.
We also need to accelerate and spread the benefits of AI, such that they are diffused broadly across countries, communities, and companies. This requires a frontier ecosystem in which both closed and open-source models can thrive.
And for firms, it’s imperative that they retain full control over their unique and tacit knowledge. Every organization should be able to build its own continuous learning loop/hill climbing machine, without becoming dependent on any one model provider, and have the ability to embed its own knowledge into models and weights they control.
So, in this context, we welcome the research, focus, and deliberate pacing needed to get alignment right as the design goal. We also welcome ideas like "embedded evaluators" and the broader efforts to develop the mechanisms to make this more than just talk.
The key is that this cannot be controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia.
This is the approach we are taking: broad access and choice at every layer of the AI stack; enterprise control of learning loops and models; and the “Code of Conduct” that underlies our own first party MAI models that we’ll publish tomorrow for public consultation.
Startups used to own urgency. Not anymore. No one can afford to be slow.
A month into @Microsoft and the pace here is nothing like what I expected from a big company.
We just hit 75.2% on Agent's Last Exam in Physical Sciences, the highest score of any publicly benchmarked agentic system!🔥
techcommunity.microsoft.com/…
Nirman Dave retweeted
Your entire life will change when you realize energy is the most attractive human trait. Not looks, wealth, or status. Energy. Start walking into rooms with genuine enthusiasm, curiosity, and interest. Watch what happens. You become a magnet for the highest quality people. Energy is contagious. Spread the kind you’d want to catch.
Nirman Dave retweeted
Don't join a startup just because you feel, the corporate life looks boring and you're not moving. You will be surprised, startups also have politics, manager drama and pointless meetings. And for a long time, you won’t know whether the company is early, growing or running out of money. Your job description will change every few months and it will confuse you. You may be hired for marketing but will be asked to do cold calling and reply to customer support and you've to leave your ego in your closet before getting to work. Nobody will give you complete instructions because nobody has complete answers. You need high agency and low entitlement which means you must pick up problems that don’t may not belong to you. Choose the founders more carefully than you choose the startup. Watch how they behave when things go wrong, because things will go wrong. Choose a honest founder, they may share genuine and useful stuff online. And not the ones doing video reels. Join them because you want to learn how to think, build, sell like them. The greatest upside of working at a startup might not be the equity but the person you become while earning it. A startup is not a smaller corporate. It is a half built company where your job is to help finish building it while everything looks like a struggle. Choosing the right people you are willing to struggle with is your decision.
What’s your advice? I asked @mecolalu when I started my co. “Always remember” he said, “nobody knows anything.”
Couldn’t be more true now than ever.
Excited to be speaking at @AIBoomi alongside some incredibly talented operators. See you there 👋
This jam is starting to look like everything we imagined it to be, and then some... ✨
Presenting the first batch of speakers joining AIBoomi Caravan ’26, bringing together some of the sharpest minds across AI, product, enterprise software, and company building.
Different journeys. Different vantage points. One shared willingness to come together and make the conversations count.
🔸 @Dthakker02 | @BatteryVentures
🔸 @iamnirman | @Microsoft
🔸 @smitahashim | Disruptive Ventures
🔸 @kamath_sutra | @smallest_AI
🔸 @tanmaigo | @PromptQL
🔸 @tr_vishwanath | @glean
💜 caravan26.aiboomi.org/
More names dropping soon.
#Caravan26 #Software #Startups #AI
Two weeks ago, I announced I was winding down my company. Today, I'm excited to share what's next.
I'm joining @Microsoft to build Discovery, its AI for Science platform.
AI is going to compress a century of scientific breakthroughs into a handful of years: new medicines, new materials, clean energy and more.
Microsoft is one of the few places with the ambition, talent and compute to make this possible.
Excited to shape this frontier, it's time to build.
