

Discover more from BrXnd Dispatch
You’re getting this email as a subscriber to the BrXnd Dispatch, a (roughly) bi-weekly email at the intersection of brands and AI. I am working on finalizing SF Conference dates and should have those in the next few weeks. As always, if you have any questions, you can reply to this email or send me a note.
Since I’ve spent the last few weeks sharing conference videos, I thought it would be good to break things up by returning to my real mission: to learn about AI in a hands-on way and share along the way. So here’s a few fun projects and other things I’ve been up to.
Calendar Bot
I meet with many different people throughout my week, sometimes back-to-back. That means I often don’t have a ton of time to Google them or check out their LinkedIn profile. So, for a while, I’ve had an idea to write a little app that would check my calendar, get some basic details about the person and the company, and use AI to compile a short briefing for the day.
About two weeks ago, I finally took the time to make it happen. Like most things, I built the prototype in Airtable (back in March, I wrote about the wonders of prototyping new AI tools there). The flow is actually pretty simple:
Use the Airtable built-in Automations to catch new Calendar events as well as updates and cancellations.
Insert new events and the people attending into events, people, and companies tables.
Use an API to get Google Results for each person and company and do a bit of scraping.
Use AI to take all that data and turn it into something readable.
Send it out daily based on what’s coming up on my calendar that day.
Like many of these kinds of tools, most of this is possible without AI, but the bit where you take a bunch of unstructured data (in this case, search results and page-scrape data) and compile it into something meaningful is pretty amazing.
Just today, I met with the AI music company Soundful, and here’s a bit of what the briefing included:
Is it perfect? Absolutely not. Is it about 70% as good as what I once got while running a company? Definitely. I’m not sure if I’ll ever make it something for more than just me, but if I do, I will be sure to let you all know.
Whisper Quick Action
If you haven’t played with OpenAI’s new open-source speech-to-text model Whisper, it’s amazing—by far the best transcription model I’ve ever used. While I still have an Otter subscription, I was finding myself frequently running stuff back through Whisper to get something better. To that end, it was getting annoying to keep having to run a curl command in the terminal to send the audio file to OpenAI (while the model is open source, OpenAI has a paid API that costs $0.006 per minute and is extremely fast (alternatively, you can also run it locally).
Anyway, I eventually ended up writing a little Automator Quick Action to streamline the process. Now, I can just right-click on any audio file on my computer and get a transcript in minutes.
You’ll need an API key from OpenAI to make this work, but if you’ve got one, it’s fairly straightforward:
Open Automator
Choose Quick Action
Follow the steps below
Here’s the two code blocks if that’s easier. First, the shell script that calls the OpenAI API (don’t forget to set up how the workflow receives data or to replace OPENAI_API_KEY with your key):
FILEPATH=@"$@"
OPENAI_APIKEY=OPENAI_API_KEY
curl https://api.openai.com/v1/audio/transcriptions \
-H "Authorization: Bearer $OPENAI_APIKEY" \
-H "Content-Type: multipart/form-data" \
-F file="$FILEPATH" \
-F model="whisper-1"
And here’s the Javascript that takes the output from the curl command and parses it for the text document:
function run(input, parameters) {
const parsed = JSON.parse(input)
return parsed.text;
}
It's nothing complicated, but super useful.
Some Quick Housekeeping
In addition to the San Francisco conference, I am working on a series of events for the fall in partnership with various brands. As part of that, I am curating a set of AI companies to come present to those companies and am looking to schedule as many product demos as possible. If you fit into this and are interested in getting in front of hundreds of hungry marketers, can you be in touch to schedule a demo? Also, if you aren’t listed in the BrXndscape: Marketing AI Landscape, be sure to request to be added.
AI @ Why Is This Interesting?
If you’re unfamiliar, one of my other projects is a daily newsletter called Why is this interesting? Just this week, there have been two AI-adjacent posts worth sharing. The first was about a new Arabic LLM out of the UAE that offers some interesting geopolitical implications. From Colin’s piece:
But there’s a deeper story here as well. As the role of AI and LLMs grow, the layer that sits on top—currently called alignment—which ensures that the model doesn’t go too far off script, offers quite a bit of editorial power to its creators. Try asking ChatGPT how to do something violent and it won’t give you an answer. Even a simple question about installing a light switch returns a heavy caveat: “If you're unfamiliar with electrical work or have any uncertainties, you should consult or hire a qualified electrician to complete the installation for you.” That almost definitely doesn’t come from the original training data, but is a layer on top that ensures ChatGPT doesn’t stray far from the pasture.
The second was about trying to play euchre with ChatGPT, a game where gab can be weaponized against your opponent. Fun stuff.
That’s it for this week. I hope you enjoyed it. As always, feel free to be in touch if there’s anything you want to talk about or I can help with.
— Noah