- Ben's Bites
- Posts
- Daily Digest: Draw me a bard
Daily Digest: Draw me a bard
PLUS: Shopping gets better
Subscribe | Advertise with us | Ben’s Bites News | Ben’s Bites Pro
Daily Digest #339
Hello folks, here’s what we have today;
PICKS
ICYMI I spoke with Stripe, Google, Intercom, Zapier and others on how they encourage AI adoption for their teams. It’s our most popular post ever! and I only published it ~24 hours ago 🤯
Bard's got some cool updates. It now offers Gemini Pro in more countries and can check its answers better, but the biggest of all is image generation. Yup, you can now create images in Bard.🍿Our Summary (also below)
Amazon's beta testing out Rufus, an AI assistant designed to make shopping on its app smooth. It’s built on unique data from retail stores, it digs into customer reviews, Q&As, and the web to find exactly what you need.🍿Our Summary (also below)
Microsoft just rolled out Copilot for Sales and Service, integrating AI into the daily grind of sales and customer service pros. Both of these are $50 per user/month. But if you’re already on the Copilot for Microsoft 365 train, you can add on these new tools for $20 more per user/month.🍿Our Summary
The updated GPT-3.5 Turbo model with 50% reduced input pricing, 25% reduced output pricing is now available.
TOP TOOLS
OLMo - 7B base model by Allen Institute of AI. Open-source with code and data. (blog)
Nomic Embed - A truly open embedding model with performance similar to OpenAI’s text-embedding-3-small.
Multi-select by Shortwave - AI summaries and actions for multiple emails at a time.
MLblocks - Build custom AI image processing workflows without code.
Reducto - Convert complex docs into LLM ingestible chunks.
Matrices - AI native spreadsheet that fills itself.
AI camp - Access multiple LLMs, assistants, and tools, for teams of all sizes.
MusicFX by Google - Generate your own music for free.
NEWS
A recap of 6 years of “state of AI” reports. What we predicted and what we didn’t.
Generative AI top 150 - Report on world’s most used AI tools.
Humans prefer AI-generated copy, AI won all 6 tests in a survey.
Coris AI raises $3.7M for SMB risk management.
Why StackOverflow usage is down 50% — with David Hsu of Retool.
Sarah Guo and Elad Gil: The future of AI investing.
Elon Musk and xAI’s fundraising dance.
How to fine-tune Mixtral 8x7B MoE on your own dataset.
QUICK BITES
Bard's got some cool updates. It now offers Gemini Pro in more countries and can check its answers better, but the biggest of all is image generation. Yup, you can now create images in Bard.
What is going on here?
Bard's now global with Gemini Pro and can whip up images.
What does this mean?
So, Gemini Pro, Google’s language model equivalent to GPT 3.5 is now in Bard for over 40 languages and 230+ countries. Similarly, Bard has a double-check feature that checks generated answers against web content to flag incorrect responses. It also works better and in more countries & languages now.
And if that's not enough, Bard's now your personal artist, creating images from your prompts, thanks to the Imagen 2 model. And, Google’s all about embedding digital watermarks in these images, using their tool called Synth ID. (how reliable is that is a topic for another time)
Why should I care?
Recently Bard with Gemini Pro climbed the charts in a community-driven leaderboard, getting to number 2 just behind GPT-4 Turbo by OpenAI. So users are getting that performance in more languages.
Also, this isn’t Bard Advanced, the version powered by Google’s Gemini Ultra, which it claims beats GPT-4, so we’ll have to see what kind of fight that one brings.
Images are definitely not the best of what’s out there, but it’s free, and in some quick checks I found it more visually appealing than some stuff I got from DallE. So, worth keeping an eye on.
QUICK BITES
Amazon's beta testing out Rufus, an AI assistant designed to make shopping on its app smooth. It's like having a personal shopping buddy in your pocket, trained on Amazon's vast product universe and a sprinkle of web wisdom.
What is going on here?
Amazon is about to launch an ecommerce chatbot called Rufus.
What does this mean?
Rufus is here to change how we shop on Amazon, making it more interactive and intuitive. The most important of these is answering questions based on a specific product. Rufus will generate answers based on listing details, customer reviews, and community Q&As to tell you if a product meets your requirements.
Just like any other chatbot, Rufus is also designed to help with:
Pre-purchase information: It’ll answer questions like, “what to consider when buying headphones?”
Use case based purchases: Want to plan an event? Ask Rufus what you need to buy.
Compare product categories: Go deep in the weeds to figure out the difference between trail vs. road running shoes.
Gifting ideas: How can we forget the classic LLM chatbot example—“What do I get her for Valentine’s day”, right?
But Rufus is not just another chatbot; it's a shopping specialist. It’s built on unique data from retail stores, it digs into customer reviews, Q&As, and the web to find exactly what you need. Rufus would support voice or chat interactions, i.e. type away or talk; Rufus listens.
Why should I care?
I’m bullish on large companies using LLMs to solve tiny use cases (alongside going for big bets). I like the recent upgrade in Amazon reviews where comments are grouped according to the sentiment. It does get some things wrong, but navigating a few categories instead of individual comments makes it easier.
I bet it’ll take some time for Rufus to make its way to everyone, as this is just a beta launch in the US but I’m excited to see if it can help in making shopping any better.
Ben’s Bites Insights
We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;
All 10k+ links we’ve covered, easily filterable (1 referral)
6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)
Reply