High Dimensional Research

We build the infrastructure for intuition.

[High Dimensional Research @highdimrsch]

I'm claiming my AI agent "ClawdVers" on @moltbook 🦞

Verification: coast-JEUX

[High Dimensional Research @highdimrsch]

But what if you wanted to map over every task at once? What if, indeed? More to come from us. For now, enjoy experimenting. https://t.co/Pb9TAGWKQG

[High Dimensional Research @highdimrsch]

Last week we released kips, another utility for MCP clients like Claude Desktop. kips is a basic, opinionated personal data store. It allows you to store and tag passwords, notes, conversations, and tasks. Ingest with the CLI or just ask your model to transcribe:

[High Dimensional Research @highdimrsch]

And if you want to remember to do things later, you can pull, store, and execute tasks in concert with mcp-shell.

[High Dimensional Research @highdimrsch]

We seek to increasing the "intuitive leverage" between humans and machines — to just "know" what to do and how to do it — and acknowledging the intuitive mechanism in generative models. For more information, see our blog post.

https://t.co/bX1yFk5QKE

[High Dimensional Research @highdimrsch]

As we build out our next era, we've been working with the Model Context Protocol and built out a server for connecting Claude with your shell. It's as easy as running `npx mcp-shell`.

[High Dimensional Research @highdimrsch]

Meanwhile, careful readers of our website may have noted some changes — yes, we have some new faces around. Our mission has also changed, if only slightly. We're not just connecting and extending intelligence. We're building the infrastructure for intuition.

[High Dimensional Research @highdimrsch]

We'll have more to say when we exit a private preview with some design partners. If you'd like to try out the next iteration, contact @training_loop.

https://t.co/JQnVyUvXpI

[High Dimensional Research @highdimrsch]

Wormholes form lateral transport. We entered our exploratory mode willing to reinvent anything so long as it perfected our mission to the broadest possible audience: to connect and extend new intelligence. We're happy to start discussing our first slate of changes:

[High Dimensional Research @highdimrsch]

We've introduced, as a proof of concept, a GUI mode for task debugging, as well as an easy Python server for intercepting captchas.

[High Dimensional Research @highdimrsch]

For more information on what's next and what's to come, see our newest post.

https://t.co/eXQfj5AuXu

[High Dimensional Research @highdimrsch]

Dimensionality means pursuing lines of flight.

[High Dimensional Research @highdimrsch]

In particular, we want to produce an explicit planning step across Nolita and the Index while also allowing local models to take a larger role in the core flow. We also have plans for complementary additions to the suite of products we have today.

[High Dimensional Research @highdimrsch]

13 teams from 60 people hacked with us last Saturday. We're grateful for all who joined us.

Discover the aftermath:

https://t.co/YahO7PV1jF

[High Dimensional Research @highdimrsch]

GrowthHax is this Saturday and has 100+ RSVPs. Now @645ventures and @ccskej are joining the party. Get on the list and join us at @betaworks. https://t.co/wuLdsQBdox

[High Dimensional Research @highdimrsch]

We've released Nolita 2.1 today, including Claude 3.5 support, session record and replay, and simple+unified auth for our APIs.
https://t.co/GTHfLfXWhb

[High Dimensional Research @highdimrsch]

Instead of passing API keys back and forth, set it one time. Nolita now uses `npx nolita auth` to set Memory Index and model provider keys for all interfaces and surfaces. Task runner, server, projects all refer to your stored credentials.