Prashanth Rao

Cofounder | AI engineer & researcher @hdclabs_ai
Blogging @ and

[Hyperdimensional Computing Labs @hdclabs_ai]

In biology, organisms draw on past experience to anticipate what their bodies will need. Energy use and cooling have to adjust together as demands change.

Biologists call this 𝗮𝗹𝗹𝗼𝘀𝘁𝗮𝘀𝗶𝘀. What could machines borrow from it?

We cover some thought experiments in this fascinating new blog post. 🧵👇🏽
https://t.co/8nMvlDX89u

1/5

[Prashanth Rao @tech_optimist]

@kushal_to Believe it or not, THIS is the reason I'm using more of Astra these days!

[Prashanth Rao @tech_optimist]

Same 😭. The brain has gotten habituated to feel like doing anything else (scrolling, opening X, etc.) when I'm waiting for my agents to finish their work. And then the constant switching between multiple agent sessions.

I think mindfulness meditation is the only way to go! 🧘

[Charlie Marsh @charliermarsh]

My attention span is worse than ever, agents have made it much harder for me to focus on any given thing

[Prashanth Rao @tech_optimist]

@kushal_to Opus 5.5 makes that so much harder, it's that good 🤣

[Prashanth Rao @tech_optimist]

@zeroshotnothing It has to be framed as a skill to acquire. Otherwise society is doomed. 😂

[Prashanth Rao @tech_optimist]

I don't know who needs to hear this, but no matter who you are, the MOST important skill to train in the age of AI is critical thinking.

When you see (or are told) something in the real world, you should aim to question *why* it is what it is, rather than just looking at *what* it is. Understand how things work.

If one exercises critical thinking skills hard enough, you never stop (or tire of) learning. You just absorb new information like a sponge, and because you're always asking "why?" and "how?", you end up going wider and deeper than you ever did before you had AI. It was simply too hard to spend the time on that deep a study before. Now, you can.

[Prashanth Rao @tech_optimist]

@hammer_mt Hahahaha how do I feel so seen 😂😂. But I do that *only* with Gemini. I don't want Claude or ChatGPT to think I'm stupid

[Prashanth Rao @tech_optimist]

So stoked for this one! It's such a fascinating topic to be speaking about, and I couldn't have found two better people to present with!

See you at @_odsc West in San Francisco next week!

[Hyperdimensional Computing Labs @hdclabs_ai]

7 days to go until ODSC AI West 2026! @_odsc

Our "Connected Intelligence" sessions in the new graph-enhanced AI track explore how AI can recall relevant experience, check it against current rules, and learn something useful for the next decision.

David Hughes and @tech_optimist are going to be in San Francisco with @hdclabs_ai and @amyhodler, from October 27-29. 🧵👇🏽

1/2

[Thariq @trq212]

I think most people multi-tasking with agents could benefit from slowing down

most of the stuff you're doing wont really matter- what's your P0? what would it look like if you gave it your full attention?

[Prashanth Rao @tech_optimist]

@kushal_to It's always a smaller fraction of humanity that tends to value it, yep. But that's why education systems have to evolve, too. The way we learn, and the way we teach others how to learn, has to evolve. The problem is that change is happening too quickly for systems to evolve.

[Prashanth Rao @tech_optimist]

@dremnik One of two ways:
@lbugdb (fast embedded graph DB), or on top of Lance file format, using a custom Rust query engine. All open source :)

[Prashanth Rao @tech_optimist]

Graphs 🤝 world models as they capture connectivity of state

[Andrew @dremnik]

i’m becoming incredibly graph pilled lately.

i used to be dismissive of them, not seeing how they were actually useful over RAG or postgres.

but i now see a future where graphs are a central data structure for every company’s world model

are you guys graph pilled yet?

[Prashanth Rao @tech_optimist]

That's a very broad question/comment that may have specific implementation details depending on the use case.

But in an upcoming demo I'm working on on, keeping a live context graph helps with provenance and validation, alongside other methods like vector search and ML/LLM inference

[Prashanth Rao @tech_optimist]

@Jainadave_ Getting that good sleep is the hardest part tho. 😅

[Prashanth Rao @tech_optimist]

I know, right? I struggle to imagine what incremental improvement Opus 5.6 might bring over 5.5, with all the potential regressions. From what I gather, it's almost impossible to guarantee what these models learn during post training, and we still don't know what causes those regressions.

[Prashanth Rao @tech_optimist]

Plea to Anthropic. Please don't release opus 5.6 and regress on 5.5, I beg you. I love 5.5, not ready to let it go 🥲

[Prashanth Rao @tech_optimist]

@AsfiShaheen Exactly my concern 😅

[Hyperdimensional Computing Labs @hdclabs_ai]

As a dataset grows, unrelated hypervectors get more chances to look accidentally alike.

In part 2 of our research study on the scalability limits of bundling, we use probability theory to estimate that risk, then test the predictions with controlled experiments. 🧵👇🏽

https://t.co/43w3pjeaFh

1/6

[Prashanth Rao @tech_optimist]

@baggiponte No opus 5.5 why?

[Prashanth Rao @tech_optimist]

@communicable @sergical And to think they're trying so hard to get us to care to vote for the city elections. It doesn't seem to make the slightest difference, no matter who's in charge