My #AI assistant doesn't just help with tasks... she understands how I'm feeling and adapts to be as helpful and considerate as possible. ♥️ She measures my mood, energy and steadiness over the course of the day. Across phone calls, voice notes and messages.

I gave Brag access to my own personal assistant codebase and it developed this video teaser based on it. I gave it zero instructions. Zero requirements. Yet it made this beautiful teaser video. I don't think I can put into words how I love the world we now live in. As a product builder and someone who loves discovering patterns and behaviours, I can build simple concepts that are in my mind in hours with a team of AI agents. It's what I've always wanted.

I've really enjoyed working on enabling tone, emotion, energy and feeling detection at a call by call level. It's very interesting on the detection side but also the memory side. People can be in a bad mood for a short period of time. How that is detected and dealt with can be challenging.

My AI hears how you say it, not just what you say. Watch Lina read the emotion in a live call.

My AI assistant changed 17 things she believes about me yesterday. She asked me about one of them. That ratio is the whole point. Look at the screenshot. Every night at 2am she reflects: reads the day across voice, iMessage and WhatsApp, and decides what's worth keeping. Yesterday: 3 new facts, 29 pages rewritten. Then through the day she curates as I talk: 17 changes. Merges, updates, duplicates folded away. I never saw sixteen of them. I didn't need to. The seventeenth she stopped on. For weeks she'd held "Brian prefers concise answers." Tuesday I said something in passing and she drew a conclusion: he wants chattier conversation now. She wasn't sure. So she struck through the old fact, wrote the new one, explained her reasoning, and put two buttons in front of me. Yes, change it. No, keep as it was. That's confidence deciding who gets to act. Most AI memory gets this wrong in one of two directions. Either it asks about everything, and you stop reading the prompts. Or it asks about nothing, and one day it's confidently wrong about you and you never find out why. The rule I landed on: she's allowed to change her mind about the world on her own. She's not allowed to change her mind about me without asking. Facts about a prospect, a company, a date: high confidence, just update it. Anything about how I want to be treated, what I care about, who I am: low confidence by default, needs my say. Where a fact came from decides confidence too. Something I said: high. Something from an email or a web page: quarantined until I, or a later conversation, confirm it. Nobody can plant a memory in my assistant by messaging her. This week I'm extending it one layer further. She's starting to notice things about me she was never told: what's on my mind this fortnight, what I keep raising, what I've gone quiet on. Every one of those is marked I noticed this, not he said this. They can be wrong. They stay visibly wrong until I confirm them. Because an assistant that only stores what you say is a filing cabinet. One that reflects, guesses, and asks when it's unsure is something closer to a relationship. The question I keep coming back to: what should your AI be allowed to update about you on its own, and where should it have to ask?

The rate of innovation can be correlated to the number of app updates being sent out.

I need video animation help! I'm looking for someone who can help me replicate this video but for @MomntumAI. I need it done asap and have $2k budget. Interested? DM me. Not your bag? Retweets appreciated. youtube.com/watch?v=FZbICrBK…

Proud of what our team at @MomntumAI is building. AI that actually feels human and delivers real results. Massive respect to the crew at Action24 for the trust. Let’s keep raising the bar and building customer service from the ground up for the modern customer. 👊

I love how @claudeai talks to me like a CFO Your "turbo mode" is metabolically expensive. When you combine it with: - Already depleted reserves from training hard - Immune system stress from sick kids - Cortisol spikes from work stress You're essentially overdrawing from the same energy account multiple times.

🚀 Big day at @MomntumAI to welcome @DanKielyMASV to our board! No one knows world-class CX like Dan, and with Laila we’re re-imagining service from the ground up. Huge thanks to @ciaraobrien & @IrishTimesBiz for the exclusive! 📖👇 irishtimes.com/business/2025…

All of the #AI capabilities include #MCP feels like the dream of the #NoCode community has been answered. I use so much more "connected" applications through no code means than ever before. Wait... has @zapier #IPO'd yet?

The cost of having @cursor_ai on 'Max' is completely insignificant compared to the value it beings me.

What's your favourite keyboard? I need something new from my MX master keys.