Sabitlenmiş Tweet
Gaurav
1.4K posts
Gaurav
@AI_303030
Building Omna- universal semantic layer between enterprise data and AI. https://t.co/OOWn08dTIf https://t.co/Xhw6D25UjX https://t.co/36Yj6fyDSY
Manhattan, NY Katılım Ocak 2024
724 Takip Edilen314 Takipçiler
This is exactly what we're seeing. The "AI for everything" era is really the "software for everyone" era. Every SMB becomes a software company.
Codie Sanchez@Codie_Sanchez
You don’t need to build the next J.A.R.V.I.S. You need to build useful tools to make businesses more efficient (data from @replit):
English
Welcome to YC, @kamil_m_ali, @asimhirji, and @plutisinc — and congrats on the launch!
Plutis is building APIs to enable effortless discovery for companies' end users through semantic search.
Check it out: ycombinator.com/launches/I9N-p…
English
@levelsio Use function calling and let it store important info to persistent memory
storeMemory($text)
Use semantic search to automatically find relevant memories for a given query and augment prompt with those.
Also allow AI to retrieve manually:
retrieveMemories($query)
English
Collected feedback from ~10 beta users and will improve that today
Not sure how to improve its memory though, I submit the convo history with every request but sometimes it forgets stuff like your name (!)
@levelsio@levelsio
Telegram is such an easy way to launch a startup My cbt.chat is just a link to Telegram t.me/cbtdotchat And that is a channel I made, people who follow that link in Telegram auto sub to the channel Once the app/bot is done, I simply msg the channel and instantly I reach ~1000 people who can start using it, of which % will pay
English
Supabase has Vector Buckets, a vector storage in addition to pgvector for storing embeddings!
Vector buckets allow you to store up to 50M embeddings in Supabase Storage for semantic searches!
You can use it in combination with pgvector for scalable semantic search!
English
4/ search will get super powerful. Semantic search will be the norm but some learned “router” layer will understand when the query is a question, action, or explicit keyword search.
English
We just released "Large Language Models with Semantic Search”, built with @cohere, and taught by @JayAlammar and @SerranoAcademy. Search is a key part of many applications. Say, you need to retrieve documents or products in response to a user query; how can LLMs help? You’ll learn about (i) Embeddings, to retrieve a collection of documents loosely related to a query, and (ii) LLM assisted re-ranking, to rank them precisely according to relevance. You’ll also go through code showing how to tie all this together to build a complete search system for retrieving relevant Wikipedia articles. Please check it out!
deeplearning.ai/short-courses/…
Palo Alto, CA 🇺🇸 English
Supabase Edge Functions have a built-in AI feature to generate embeddings!
This allows you to convert string data into embeddings and perform vector searches against them without having any other dependencies!
Great for creating semantic search all within the Supabase stack!
English
2️⃣ Semantic
Semantic means the meaning layer
Not the syntax
Not the label
Not the shape of the thing
The meaning
Semantic search means the system understands intent, not just matching words
Once you own this word, you start seeing the difference between storing information and making it usable
English
You are a COMMANDER🎖️
Usage limits are not the real bottleneck
It is your ability to command language
To bend it toward what you actually mean
You have gaps in your vocabulary, and lazy prompts make them visible
So the Agent guesses
And guessed output almost always drifts toward mediocre
Here are some words from my wordbank⬇️
English
@JeffBezos Two small Christmas wishes:
• Enable copy/paste in the Kindle web app.
• Make the Kindle desktop app good. (Imagine how great it could be -- semantic search via embeddings, for example.)
English
Build a semantic search tool across internal data sources with @gpt_index!
A top health tech startup is looking for a developer to create an internal People Search tool 🔎 Listed for $7,200
@MorganBreitmeye/internal-tool-for-pe" target="_blank" rel="nofollow noopener">replit.com/bounties/@Morg…
English
Gaurav retweetledi
2 of the most underrated @openclaw features (and why I could not imagine my life without it):
1) semantic search
It’s kinda like Ctrl + F for every conversation you’ve ever had with clawdbot.
And semantic search means it searches your conversations based on MEANING, not just specific terms.
So if I ask clawdbot about “that product idea we discussed last month” it can find the exact idea based on that tiny amount of context.
2) persistent memory across platforms
If I’m talking to clawdbot in discord I can continue the conversation on any other platform (WhatsApp, slack, telegram etc) and it doesn’t skip a beat.
Picks up right where we left off and remembers everything we talked about.
If you’re not using clawdbot you are missing out BIG time.
Dan Peguine@danpeguine
why @openclaw is nuts: 1. your context and skills live on YOUR computer. not a walled garden and you can switch models with one command 2. it's open source so you're not paying anyone for insane value 3. it has a growing community building skills for it - this is compounding rn 4. world class developers like @steipete, @theguti 5. only 19 days old and is constantly improving, shipping speed is insane 6. "personal AI assistant" undersells it. it's a company assistant, family assistant, team tool. 7. it's accessible via any of your favorite platforms (whatsapp, telegram, signal, discord, slack...) 8. it's proactive AF: cron jobs, reminders, background tasks, it reaches out to you 9. its memory is amazing, context persists 24/7 what else?
English
QME
@rex_woodbury @NotionHQ @airtable @Guru_HQ @getcommande @usefyi 100% on context. Especially with better natural language processing, I'm very big on semantic enterprise search.
Also think many destination apps may instead evolve to become critical features a la @krispHQ
English
The proliferation of work apps has become unsustainable. It's beginning to actually erode worker productivity.
If the last decade was about unbundling Microsoft Office and Google's G Suite, the next decade will be about a "rebundling" of work.
A few thoughts on this trend👇👇👇
English
@hahnbeelee for coding docs, and coding agents sure. but i would prefer semantic search for general & enterprise use cases. hybrid search still outperforms grep.
English
semantic search is dead. file-based context retrieval is far better.
let us know what you think and if there's more you want to see from our AI assistant
Mintlify@mintlify
The assistant just got a lot smarter It's now much faster and more precise What's new: → Grep retrieval (more precise than semantic search) → Multi-modal (ask questions about images) → OpenAPI search (ask about a specific API endpoint)
English
We're still in the stone age of writing.
GPT-3 and other language+knowledge APIs need to be applied to document writing tools (like Notion and Roam) yesterday.
e.g.: better autocompletion, tokenization (e.g.: acronym expansion), suggested document links, semantic search.
San Francisco, CA 🇺🇸 English
@dhonos253 @Austen - semantic search (retrieval when text doesn’t match, natural language understanding)
- exhaustive spellcheck
- fast autocomplete
- ranking docs well (much much harder than simple page rank)
- freshness (breaking news content)
- de-duplicating identical content
- porn filtering
English
I swear Google’s search results are not only worse than they used to be but rapidly declining
English