🎁 Milo Gift Giver GPT
Major mental load alert: figuring out thoughtful somethings for all the people, then tracking, buying, wrapping .. all before the giving can happen.
This week’s little AI-powered helper? Help with all the gifting this holiday season - whether it’s Christmas, Hanukkah, a birthday and everything between.
We’re playing where we think Milo can help most - 1. being a brainstorm buddy to figure out the right kinds of gifts for every kid/teacher/host/partner/family member (as a bonus, Milo will suggest some product links to check out) and 2. keeping track of what you’re getting each person so you have it in one place (that isn’t your brain).
It’s still new and experimental - so let me know what you think!
Giving gifting some behind-the-scenes help
Like many, I love giving and getting presents. They don’t have to be expensive or fancy, but thoughtfully procured (or made!) gifts, given with love, have such impact.
But I definitely don’t have that gift-giving superpower some of my friends do - in picking that perfect something for every occasion. Always unique and thoughtful.
So I leave it to the last minute until I’ve run out of options and I find myself fighting the hoards at the mall or falling back to tired staples.
The mental load of gifting is significant, before any kind of giving can happen.
So this year, we wanted to see if we could help a bit.
🤓 Tell Milo who you’re looking to get a gift for, their interests and budget. Try everything from a sitter or a teacher to your kids/partner to a host gift. Milo will work with the interests/budget to come up with ideas.
🛍️ See suggestions. Once you feel good about the ideas, Milo will pull some actual items for you to review. The idea is less to get the item exactly right but to help you get a sense of your options that you can also get from your preferred places.
📝 Ask Milo to keep a running list. Tell Milo what you’ve decided (can be other things you’ve already figured out, or handmade things you’ll be giving like baked goods). Milo will remember what you’re getting for whom.
Some tips/tricks:
👯♀️ Checkin’ it twice - under the hood, this little helper is more complex than coming up with some Elf ideas. It’s hard to not only search all the possible items but also find the right kind of brand/item/site. So think of this tool more as a brainstorming buddy.
😅 Be patient - researching and finding options takes a minute or two so expect a delay when pulling results. Try to keep things simple (eg one person at a time vs parallel requests) to keep it snappy.
👍🏽 Give feedback - these helpers are little experiments to see how we can use AI in everyday ways - but it requires feedback - on all of the good/bad/ugly. So if you find it helpful, give it a 👍🏽 and if you find either the quality lacking or it’s just not the help you’re looking for, give it a 👎🏽.
Give it a whirl - just sign up and text “Hi Milo” to the Holiday Helper line.
Do you have something you want solved?
We’re taking ideas from parents - tell us what to build and we’ll build it for you. If it has mental load written all over it, pop it onto our community build-list:
We hope we can help lighten even a little of the load this holiday season 💕
xo,
-avni + milo team
🔦 Technical spotlight
This week, a more technically complex helper.
How do you teach an LLM (like GPT4) to talk to the internet and provide gift options?
🤓 LLM Basics
LLM stands for Large Language Model, which is a type of AI trained to understand and generate human language.
These models, like ChatGPT, can answer questions, write content and more, based on the vast amount of text they've been trained on.
️These models don’t have the most recent information so they’re not great at being able to talk about specific prices, delivery dates etc.
Think of LLMs more like a brain - full of knowledge. When you ask a question, it will give you an answer but it’s hard to pinpoint exactly “how it knows” what it knows. On the other hand, when you ask Google the same question - it goes and “finds the best source/file” to supply you the answer. But it’s less good at generalizing an answer and extending into the fuzzy boundaries of what might be around that question.
🔎 RAG - Retrieval-Augmented Generation:
RAG is a technique that combines the generative power of LLMs with the ability to retrieve real-time data from external sources (like the internet).
It first retrieves relevant information from the internet (in our case, your fav shopping sites), and then uses this information to generate a more informed response
🎁 Putting them together:
In practice, when a user asks Milo to fetch gift ideas, the LLM uses RAG to fetch up-to-date information from online sources.
This might involve searching for product listings, reviews, price comparisons, or availability from various shopping sites.
Milo then looks at all this information, considers them against your preferences and then returns a list of recommendations.
So: there are lots of tricky bits to put together - giving instructions to “think about” gift-giving as a parent would, in a simple, succinct way. Focus search on high quality sites and items, and return the “best” options.. all in a time frame that feels snappy 😅.
Still - we think it can help to brainstorm and lighten some of that load so we’re excited to have you try it and see how it works!