News * Competition * About other projects

[Twenkid]

13.12.2022: Microsoft seem to develop their "Jack-of-all-trades": https://www.reuters.com/technology/microsoft-mulls-building-super-app-information-2022-12-06/ "Dec 6 (Reuters) - Microsoft Corp (MSFT.O) recently considered building a "super app" that could include shopping, messaging, news and web search services among others, The Information reported on Tuesday, citing people familiar with the matter" https://www.theinformation.com/articles/microsoft-eyes-super-app-to-break-apple-and-googles-hold-on-mobile-search

4.1.2023: Microsoft plan to include ChatGPT into Bing

[Twenkid]

27.12.2022: Vectara and *PineCone

• Vectara is a startup which is implementing ideas which I also had/have in mind, once I started playing with Clip-retrieval interface. However I think there could be all kinds of search methods which could be applied: both keyword based, extended with query expansion, synonyms, similarity and colocation, relevance, using resources such as WordNet and Word Sketches etc., and "rule-based", either automatic and manual etc.... And... a generic vector comparison which has produced latent vectors by an end-to-end process without such complex branching on different variants. The enumerated ideas above could be put into vectors as well etc. The articles also seem to talk about hybrid approaches.

https://vectara.com/ What is Vectara Vectara is neural search as a service. The platform provides a complete ML search pipeline from extraction and indexing to retrieval, re-ranking and calibration. Every element of the platform is API-addressable. Developers can embed the most advanced NLP models for app and site search in minutes. https://venturebeat.com/ai/vectaras-ai-based-neural-search-as-a-service-challenges-keyword-based-searches/

• PineCone- vector database service, in April announced +$28 mill. dollars investments, in 2021 reported 10 millions etc. Vector embeddings into 32-20K params, compression, indexing --> smaller index. Smooth update of the indices after update of the data. Filtering by metadata. Similarity ...

https://www.pinecone.io/ Long-term Memory for AI The Pinecone vector database makes it easy to build high-performance vector search applications. Developer-friendly, fully managed, and easily scalable without infrastructure hassles.

[Twenkid]

NeevaAI

https://neeva.com/blog/introducing-neevaai http://neeva.com

A new search engine, released this year, employing LLM. Ad free, they claim they do not track their users. Free and Premium membership are offered. The service supports search within connected Gmail accounts etc.

The "AI" advanced features seem to be available only in USA for now though. I'm testing it during the last few days and it did the job well so far as a normal google-like search engine.

[Twenkid]

LAION-AI/Open-Assistant

An open answer to ChatGPT, aiming at retrieval enhanced operation, querrying search engines etc. One of the developers is the youtuber Yanick Klicher.

https://docs.google.com/presentation/d/1n7IrAOVOqwdYgiYrXc8Sj0He8krn5MVZO_iLkCjTtu0/edit#slide=id.g1c276c7e915_4_5 "OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so. It can be extended and personalized easily and is developed as free, open-source software." " So they are doing some of the goals of the "Sacred Computer" projects, undeveloped, these are AGI-directed functions. In my nomenclature that is sort of "intelligent OS" and a real intelligent assistant, unlike the basic dialog bots which just answer questions. An AGI system learns to interact with and to map its representations to any systems.

https://github.com/LAION-AI/Open-Assistant Open Assistant is a project meant to give everyone access to a great chat based large language model. We believe that by doing this we will create a revolution in innovation in language. In the same way that stable-diffusion helped the world make art and images in new ways we hope Open Assistant can help improve the world by improving language itself.

[Twenkid]

PERSONAL.AI / LUTHER.AI

https://www.personal.ai/

That's in tune also with my inhouse unpublished project called ACS, Assistant C#, or Research Assistant.

AI that’s personally yours Supercharge your memory and relationships with your own, personal AI messenger.

Supercharge your digital mind Imagine if you instantly could have the answer to anything you once knew, or could recall every detail about your conversations without endless scrolling or searching.

Your personal AI is the digital library of you—a treasure trove of your life’s information. Whether it's future dinner plans with friends or work meeting recaps, everything is indexed automatically and is entirely discoverable—simply by chatting.

[Twenkid]

GPT4, ChatGPT (GPT3.5) etc.

...

[Twenkid]

Cohere

https://www.youtube.com/watch?v=ooBt_di8DLs https://cohere.ai

[Twenkid]

LMQL

A domain-specific language to sample LLM etc.: https://lmql.ai/

[Twenkid]

SuperAGI
Open-source framework to build, manage and run useful Autonomous AI Agents

https://github.com/TransformerOptimus/SuperAGI

[Twenkid]

YouAI

"Powerful AIs for All Kinds of Things"

https://youai.ai/ Self-digitization etc., Delegation of action to assistants* etc. https://youtu.be/feoISCIbz-s

[Twenkid]

Rewind AI

https://www.rewind.ai/

"A truly personalized AI Rewind is a personalized AI powered by everything you’ve seen, said, or heard. Your colleagues will wonder how you do it all."

"Perfect memory" - recording user's experience and full-sensory-search.

This is related to Sacred Computer's ACS/Assistantt/Intelligent OS/Research Accelerator prototypes and ideas.

Steven Wolfram has mentioned that he's logging his activity as well.

[Twenkid]

BioSeek

A Sofia, Bulgaria based search engine specialized in Life Sciences, medical research etc.

https://www.bioseek.eu/bg/bul/research/home

https://www.facebook.com/bioseek/

[Twenkid]

Verses AI

Verses AI - the company that develops multiagent systems, inspired by and implementing the Free Energy Principle/Active Inference by Karl Friston, releases a system called "Genus". https://www.youtube.com/watch?v=mIUcU5c-vEs

FEP/AIF actually repeat, expand with volume, prove and continue the work on some of the main principles of my own Theory of Universe and Mind - the classical works were published between late 2001 and early 2004.

Look forward for "Universe and Mind 6" and the book about the Prophets of the Thinking machines, and check the github page for comparisons.

• PS. Funnily their demo features a drone, LOL (see EZ Drone experiments). According to Yahoo finance etc. they are evaluated at about $100M, there's info about some funding round of $3M CAD.

[Twenkid]

AnythingLLM

https://www.marktechpost.com/2023/06/15/meet-anythingllm-a-full-stack-application-that-transforms-your-content-into-rich-data-for-enhanced-large-language-models-llms-interactions/

MIT License, ... how users interact with documents or any piece of content using an LLM. Interested users can clone the project from this Github repository and can proceed to set up their application.

https://github.com/Mintplex-Labs/anything-llm