Get live statistics and analysis of Brandon Galang's profile on X / Twitter

Today is day #1 of a 365 daily video posting challenge. Inspired by seeing @SherryYanJiang take after @YoniSmolyar's example. I've always been a proponent to sharing your thoughts and perspectives on social media. A sentimental part of me has loved to journal and document the passage of time. From 20-25, I actually maintained a daily written journal and a "1 second every day" video after watching the 1SE Ted Talk. In my early 20s, it was watching @garyvee's messaging around putting yourself out there to have a chance at being noticed by others and developing the skill of presentation. I've tried more than once to start a blog and YouTube channel, but never stuck with it. A few years ago, the interest was reinforced by reading @naval's "How to Get Rich" manifesto, where he speaks about the different forms of leverage: - Human - Financial (capital) - Media (social!) - Code In the last 6 months, posting content has led to me learning so much about the AI/LLM/agent space and chance re-connections from old friends, colleagues, and peers who happened to read my stuff. You'll be seeing my face a lot more; every day for the next year at least. Not gonna lie, I re-recorded this at least 5 times. I told myself moving forward, I'll limit myself to one take per day. If you're interested in joining me or have tried to do something similar before, would leave to hear about it in the comments.

If you're using Claude Code on Pro or Max, I strongly recommend giving this a try. @MarkusOdenthal @EricBuess : am curious what your numbers look like.

7 months ago, there was barely a whisper of AI at Wonder. Now, the #wonder-ai-portal channel I kicked off has almost 100 people across Wonder, Blue Apron, GrubHub, and Tastemade. We're still figuring how to get folks access and how to prioritize new LLM initiatives, but we're leaps and bounds ahead of where we were at when I started seriously pushing for this. When Google dropped Gemini 2.0 in December, I was drowning in tech debt and a critical legacy tool was buckling under the load of a newly prioritized Growth project. It felt like my back was against the wall and I didn't see any way out from it. Gemini 2.0 hinted at a future of "intelligence too cheap to meter" as others have said. I saw a future where all of our operational challenges could be deftly tackled with the flexible reasoning of SOTA LLMs. Since then I've been obsessed with figuring out how to get this deployed internally. I became "that AI guy." While the prioritization of these features and getting budget allocation for tooling remains a challenge, this vision is more or less playing out. Manual data cleansing, classification, and RCA on context sparse, few-shot examples is practically solved with SOTA models like Gemini 2.5 Pro. OpenAI's recently announced o3 price drop continues this trend. Why build rigid, deterministic business logic when your business changes so rapidly when an LLM can do it cheaper, faster, and with higher quality? Why waste iteration cycles defining the rules when reasoning models can run through your operational data and TELL you what the problems are? If you're in the operations space, getting up to speed on AI should be your #1 priority. If you haven't kept up to date with the recent development of SOTA reasoning models, you're doing yourself a disservice. They are trivializing a large number of previously impossible to solve, long-tail problems. They are helping to make the most difficult challenges workable. I haven't been this fired up in a long time. How does your experience compare? Are you seeing this in your space, or is it just me?

Today is day #1 of a 365 daily video posting challenge. Inspired by seeing @SherryYanJiang take after @YoniSmolyar's example. I've always been a proponent to sharing your thoughts and perspectives on social media. A sentimental part of me has loved to journal and document the passage of time. From 20-25, I actually maintained a daily written journal and a "1 second every day" video after watching the 1SE Ted Talk. In my early 20s, it was watching @garyvee's messaging around putting yourself out there to have a chance at being noticed by others and developing the skill of presentation. I've tried more than once to start a blog and YouTube channel, but never stuck with it. A few years ago, the interest was reinforced by reading @naval's "How to Get Rich" manifesto, where he speaks about the different forms of leverage: - Human - Financial (capital) - Media (social!) - Code In the last 6 months, posting content has led to me learning so much about the AI/LLM/agent space and chance re-connections from old friends, colleagues, and peers who happened to read my stuff. You'll be seeing my face a lot more; every day for the next year at least. Not gonna lie, I re-recorded this at least 5 times. I told myself moving forward, I'll limit myself to one take per day. If you're interested in joining me or have tried to do something similar before, would leave to hear about it in the comments.

If you're using Claude Code on Pro or Max, I strongly recommend giving this a try. @MarkusOdenthal @EricBuess : am curious what your numbers look like.

7 months ago, there was barely a whisper of AI at Wonder. Now, the #wonder-ai-portal channel I kicked off has almost 100 people across Wonder, Blue Apron, GrubHub, and Tastemade. We're still figuring how to get folks access and how to prioritize new LLM initiatives, but we're leaps and bounds ahead of where we were at when I started seriously pushing for this. When Google dropped Gemini 2.0 in December, I was drowning in tech debt and a critical legacy tool was buckling under the load of a newly prioritized Growth project. It felt like my back was against the wall and I didn't see any way out from it. Gemini 2.0 hinted at a future of "intelligence too cheap to meter" as others have said. I saw a future where all of our operational challenges could be deftly tackled with the flexible reasoning of SOTA LLMs. Since then I've been obsessed with figuring out how to get this deployed internally. I became "that AI guy." While the prioritization of these features and getting budget allocation for tooling remains a challenge, this vision is more or less playing out. Manual data cleansing, classification, and RCA on context sparse, few-shot examples is practically solved with SOTA models like Gemini 2.5 Pro. OpenAI's recently announced o3 price drop continues this trend. Why build rigid, deterministic business logic when your business changes so rapidly when an LLM can do it cheaper, faster, and with higher quality? Why waste iteration cycles defining the rules when reasoning models can run through your operational data and TELL you what the problems are? If you're in the operations space, getting up to speed on AI should be your #1 priority. If you haven't kept up to date with the recent development of SOTA reasoning models, you're doing yourself a disservice. They are trivializing a large number of previously impossible to solve, long-tail problems. They are helping to make the most difficult challenges workable. I haven't been this fired up in a long time. How does your experience compare? Are you seeing this in your space, or is it just me?