Elvis Ha

Design-led, AI product management -- forging new paths in this exciting era. Very demure, very mindful. VP @

[Elvis Ha @elvis_ha]

Claude 4 day interview between @mikeyk and @DarioAmodei: "Machines of Loving Grace" as the R&D product roadmap, and also something I've thought of a lot as also someone w/ neurosci background: AI is teaching us so much more about the human condition. https://t.co/qdexKkALQd

[Elvis Ha @elvis_ha]

Had a great 5 min convo with Lennybot! Looking forward to Lenny and Friends on Thursday esp with this conversation starter on multimodal AI product management! @lennysan

[Lenny Rachitsky @lennysan]

Exciting announcement of the week #2: A new and much improved Lennybot 🤖🔥

It's trained on all of my podcast episodes (200+ guest interviews) and newsletter posts (500+ issues, including all of my paid posts).

The coolest part is that the bot is also trained on my voice, which means you can talk to it! Try clicking "Call" at https://t.co/TqkCh3hnMC. Seriously, try it. It's wild.

Even cooler, you can call or text my bot anytime you have a question about product, growth, or your career: [phone]. Add it to your speed dial for the next time you need help.

A huge thank you to the team at @withdelphi for making this bot possible 👏👏👏

Delphi’s architecture was inspired by Ray Kurzweil’s book “How to Create a Mind.” They take a modular approach to building the backend intelligence, where models are stitched together to perform specialized processes. The models work in unison to decide the most appropriate and personalized response, mimicking human-like decision-making. They are using models from OpenAI, Anthropic, Meta, and Mistral, depending on which model performs best on specific tasks.

Try it out and let me know what you think: https://t.co/TqkCh3hnMC

[Elvis Ha @elvis_ha]

When you eventually extrapolate this to web3 digital worlds with digital goods that have real economic value, or better water game simulations of geopolitical issues like inflation and global supply chain, it will start to impact the world of poilitics and business (4/4)

[Elvis Ha @elvis_ha]

1000 agents interacting with each other in Minecraft is a very interesting experiment (1/4) https://t.co/NEcwZLDOye

[Elvis Ha @elvis_ha]

Fig 2 - tech tree progression by AI. This can help identify improvements to pro esports strategy by finding out optimal ways to hit big power surges in tech tree progression games (Starcraft, League of Legends, upcoming game Deadlock) and establish competitive advantage (3/4)

[Elvis Ha @elvis_ha]

Been thinking a lot about Double Diamond model https://t.co/HQBXTJZhGy and how it is especially successful in the AI era. It is my answer to why we have a bubble and why design-led culture is the way out of the bubble. Putting non-deterministic experience sin front of real users.

[Ridd 🤿 @ridd_design]

A new type of design role is about to emerge...

And the more I think about what it might entail the more excited I get 👇

[Elvis Ha @elvis_ha]

This shift is crucial in the AI era, where user experiences will be non-deterministic – each user might have a slightly different experience.

It emphasizes the need to put prototypes in front of real users and client personas as early as possible. The future of design is here!

[Elvis Ha @elvis_ha]

News from the design front: Figma #Config2024 conference reveals how @figma is automating the tedious parts of designers' lives, mirroring our efforts to streamline recruiters' and hiring managers' work.

[Elvis Ha @elvis_ha]

The designer's role will become more creative, focusing earlier in the process on ensuring products are used, discussed, and iterated upon.

[Elvis Ha @elvis_ha]

@mikegonz @ridd_design @figma I'd thought that AI would first automate accountants'/ lawyers' work 1st and creatives'/designers' work last. What I didn't foresee is @figma leadership's brilliant & bold strategy of automating the "BS" (boring stuff) of designers' work! Designers' QOL and careers rejoice!

[Elvis Ha @elvis_ha]

The future is multimodal: here's a layperson's introduction into what that term, and what that vision, means for the recruiting experience! https://t.co/K8VO74PUdU

[Elvis Ha @elvis_ha]

⚡This era marks the most exciting inflection point in my tech career! I'm immensely thrilled to roll out our vision in production. Discover what we've been working on, with more to come this summer! https://t.co/d2kbrlk9Um

[Elvis Ha @elvis_ha]

This year's debate: will the future be large foundational models, or multiple small ones? Google VPs at #GoogleCloudNext hinted at much larger foundational models, that will make it easy to use multi-modal inputs and outputs. So, text, voice, visuals can flexibly serve as both.

[Elvis Ha @elvis_ha]

It has been said that "the most popular coding language this year will be English", but that doesn't mean the best user experiences in 2024 will be conversational. They are likely to be multi-modal.

[Elvis Ha @elvis_ha]

1) Ground and cite data sources
2) Summarize them in the best-fit modality (not necessarily text, perhaps visual charts) and place those summaries where the job to be done needs it most
3) Do as much of the job to be done as possible, including the last mile

[Elvis Ha @elvis_ha]

Google's own productization of GenAI Knowledge Worker Productivity tools keeps repeating the same playbook very well, as shown in this keynote clip of generating Google Slide presentations from a set of grounded https://t.co/tn7FDxYqNV