Can’t believe Elon showed up to our Grok Bot event today

I'm finally realizing how important focus and specialization is. It's remarkable. You can get so far niching down on something, versus casting a wide net and not specializing on anything. Niching requires commitment and saying "no" to a lot of things, so it's actually hard to do.

Let’s bring culture back to tech!!

Publishing my event livestreams to YouTube now 👀

Throwback to Montreal Grok Bot Build Day earlier this month

This Saturday’s @modern_lyceum AI Café theme was all about ...VOICE AGENTS 🗣️🤖 It was in collaboration with an inspiring and ambitious startup that deploys one of the most effective business use cases of voice agents: Clerk AI (they're a San Francisco-based but just opened an office in Montreal) What we covered:  🔹 The Timeline Development of Voice Agents: From Siri (2011) to Speech-to-Speech Models Today (2026) - Presented by me  🔹 An Overview of Current Voice AI Startup Landscape and Build Tutorial - Presented by @_williambowen, Lead Forward Deployed Engineer at Clerk AI 🔹 The Complexities and Remaining Open Problems of Voice AI in the Application of Language Learning - by @nsphin, cofounder of Conversaflex 🔹 Using AI to Keep Seniors and Loved Ones Safe from Scam Callers - by Richard Gotlieb, cofounder of OnGuardAI 🔹 Voice AI Applied to Language Learning - by Marie Membrado, co-founder of LinguoSphère Also recorded the session, I'll share the video soon! Keep an eye out 😄

X’s algorithm is crazy

IF YOU WORK AT @cohere OR KNOW SOMEONE THAT DOES, please reach out to me -would love to do an event together 😄

POV: you’re hosting a Canadian tech event

AI is still way too slow at making slide decks

As you get older, stretching quickly turns from optional to mandatory

I don't want to leave my AI bubble, it's scary out there

There is so much happening in voice AI right now: - Grok Voice 3.0 launched at $0.08/min with < 1 sec latency (best speech-to-speech model) - OpenAI added prompt to voice for GPT Live (can literally create ANY voice you describe) - Gemini 3.8 Live can call tools while it keeps talking (e.g. send email) - Microsoft streaming transcription starts in about 100 ms (agents can start processing before you finish talking) - ElevenLabs can now detect synthetic voices on live calls (businesses can spot AI callers and cloned voices) If you are in Montreal, I'm doing an event dedicated to voice agents at the Clerk AI office. Sign up here: luma.com/ietfqhq0

Hosting an event all about AI Voice Agents in Montreal Saturday, October 24th. Will be fun, come through!

Meta bench-maxxing again?

Anthropic code merges per day in 2026: - May: 2.5k - September: 30k They say coding is largely solved and now the bottleneck is software infrastructure that was previously built for "human scale".

We're a single MCP/Plugin/Connector away from LinkedIn combusting

I put the GPT-Realtime voice model on my Apple Watch

I much prefer a total weekly usage limit than annoying daily hourly windows, another great product decision by the grok bot team 👏