- Ben's Bites
- Posts
- A vision language model
A vision language model
PLUS: create AI-enabled apps with no-code
Hey folks, this is Ben’s Bites - your AI copilot. Today we’ve got big companies like HubSpot showing their AI cards with new tools, Microsoft continue to roll out their AI functionality across their product suite and two new models from Google; a vision language model and a language model (tad confusing).
Let’s get to it.
🤌 Our Picks
Highlights if you've only got 2 minutes
Hubspot's big entry in the AI game is called ChatSpot. This is one of the two best name plays I've seen from established brands integrating AI in their products. The other one is Zapbots from Zapier (not biased 😉)
There are two categories of use cases for ChatSpot that Hubspot is highlighting:
a) eliminating manual work like data entry and follow-up emails
b) understanding all the data with natural language prompts
With the amount of marketing and sales data enterprises have with Hubspot, these are game-changing features. In addition to this, they have released a Content Assistant to generate blog ideas and outlines.
Check the announcement here: Introducing HubSpot's new AI tools.
Microsoft has also added AI functionality to a bunch of apps in its business ecosystem, primarily targeting sales and marketing users. The addition to the Dynamics 365 is called Copilot, and similar to Github’s Copilot making coding easy for developers, it aims to make selling easy for salespeople. According to Bloomberg, Nadella sees this as a step towards making one Biz App workflow instead of separate CRMs, ERPs, etc.
Power platform from Microsoft also got the AI upgrade with conversation boosters in Power Virtual Agents and GPT model in Power Automate. The Power platform is a set of no-code solutions and new AI features embed text generation in those solutions.
The next update from Microsoft is on March 16th for “workspace productivity,” which probably means Office 365.
As always, where others are adding AI to products, Google is releasing more papers on new models. (when will they ship products?!!)
First one is PALM-E, a general-purpose vision language model trained with 562 billion parameters. This combines the visual question-answer capacity of language models with robotics to perform actions in the real world. This thread from one of the authors gives an overview of the model.
In November last year, Google announced its plans to make a language model with automatic speech recognition that supports 1000 languages. They are making steps towards it with the Universal Speech Model (USM). The Verge also reported it here: Google’s one step closer to building its 1,000-language AI model.
When: March 15th, 2023
What is it about: Given the size of computer vision datasets, identifying what to label, ensuring high label quality, and curating the right datasets at scale are challenges that can make or break your Computer Vision application in production.
In this webinar, we'll show you how to save time, money and sanity by building efficient data-centric workflows for Computer Vision applications.
Who is it for: Whether you're already working on a computer vision project or just getting started, this webinar is for you!
Price: It's FREE
🛠️ Cool Tools
Product launches, updates and demos
Wikipedia GPT - Talk to Wikipedia using chatGPT. (link)
Ask YC from Transcriber - Question answer based on YC’s youtube channel transcriptions. (link)
Real-time job market analysis. (link)
Adrenaline - AI assistant that can help debug your code. (link)
Changelog by Olvy - Inform your users about product updates without spending hours every week writing releases. (link)
Berri - Build production-ready ChatGPT apps in under 2 minutes by easily connecting your data to an LLM. (link)
Chat thing - Turn any Notion workspace into an AI chatbot. (link)
Naval GPT - AI-powered search & chat for Naval Ravikant's Twitter thread "how to get rich." (link)
Roasted - Get roasted by AI generated jokes. (link)
Summ - Intelligent question-answering and search for user interviews. (link)
Opinionate - Watch ChatGPT debate itself on a given topic. (link)
Syncly - AI customer feedback analytics that drives retention. (link)
Diffuse Bio - Generative AI for protein design. (link)
Sherloq - Collaborate and manage your SQL data with the power of Gen AI. (link)
Hubble - No-code platform for creating AI-enabled applications with no engineers. (link)
🤓 Miscellaneous
News, podcasts, videos, blogs etc
Beyond the screen - Fully automated podcast from text to production. (link)
ARTEMIS - Advanced robotic technology for enhanced mobility and improved stability. (link)
Developers are turning to GitHub Copilot. One startup VP says it helped him save 10% time. (link)
Berri AI, a YC company, uses Replit in its development to bring LLM products to market quickly. (link)
Amazon’s big dreams for Alexa fall short. (link, without paywall here)
GPT-3 will ignore tools when it disagrees with them. (link)
Political media’s next big challenge is navigating AI deep fakes. (link)
Open banking startup Abound nabs $601M to supercharge its AI-based consumer lending platform. (link)
Sceptical investors worry whether advances in AI will make money. (link)
Generative AI and the future of creative jobs. (link, without paywall here)
🎓 Learn
How-to’s and resources
👋 Too many links?! I created a database for all links mentioned in these emails. Refer 1 friend using this link and I'll send over the link database.
🔬 Research
Published research papers
StyO - Stylize your face in only one-shot. (link)
Prismer - A vision-language model with an ensemble of experts. (link)
HiCLIP - Contrastive language-image pretraining with hierarchy-aware attention. (link)
FoundationTTS from Microsoft - Text-to-speech for ASR customization with generative language model. (link)
TrojText - Test-time invisible textual trojan insertion. (link)
PixMIM - Rethinking pixel reconstruction in masked image modelling. (link)
Learning humanoid locomotion with transformers. (link)
Nerflets - Local radiance fields for efficient structure-aware representation of 3D scenes from 2D supervision. (link)
Taming Stable Diffusion with human ranking feedback. (link)
📰 Unclassifieds
Short, sponsored links
Writers Brew - An AI writing assistant that works across apps & browsers. It can WRITE. IMPROVE. REPLY. SUMMARIZE. TRANSLATE. Check out here »
Rewind AI is hiring for Head of Marketing. Learn more »
Fable is hiring a Sr. ML Engineer for generative modelling. Learn more »
Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 37k subscribers?
Advertise with us | Job board | Join Community
🖼 AI Images of the day
Funny memes and pics from around the web
Gotta convert those leads
Send this to 1 AI-curious friend and receive my AI project tracker database! Use this link.
How was today's email? |
Reply