What a week.
The emotional swing from the low of the SVB crisis last weekend and Jason Calacanis’s caps lock, to the optimistic future offered by GPT-4 being released, to Balaji now betting $1 million on hyperinflation happening within 90 days… was a lot to process.
As founders we typically want to focus on the problems (so we can fix them), but for this week’s issue I’m focusing on the good things.
There’s a ton of Twitter threads about GPT-4, but most are full of examples that aren’t super actionable.
So this week’s post is all about how founders can specifically use GPT-4 to save time or money. 👇
🤝 A Message from Deel
Ready to take the next step in your hiring journey? You’ve found the global talent but need to onboard them. Deel can help.
Deel’s simplified a whole planet’s worth of information. It’s time you got your hands on our international compliance handbook where you’ll learn about:
Attracting global talent
Labor laws to consider when hiring
Processing international payroll on time
Staying compliant with employment & tax laws abroad
With 150+ countries right at your fingertips, growing your team with Deel is easier than ever. Download our international hiring guide now.
Want to get in front of 23,000+ founders? Sponsor Houck’s Newsletter
9 Ways Founders Can Leverage GPT-4
Write Code
GPT-4 can help both experienced and entry level engineers. A new engineer becomes capable and a 10x engineer becomes 100x:
I don’t think people understand how fast AI is changing the game right now.
The team is already using GPT-4 to build entire components in 10min that otherwise would have taken us days.
We don’t even have experience in the language these components need to be written in! https://t.co/txvlmcDFqR
Tools like warp.dev are saving hours or even days of dev time pretty regularly and simultaneously dramatically lowering the barrier to entry for someone who wants to learn engineering.
Today I used GPT-4 to make "Wolverine" - it gives your python scripts regenerative healing abilities!
Run your scripts with it and when they crash, GPT-4 edits them and explains what went wrong. Even if you have many bugs it'll repeatedly rerun until everything is fixed https://t.co/gN0X7pA2M2
For non-technical founders, it may now be possible to create simple scripts to automate early processes you’re doing even if you don’t have a technical member of your team.
Manage Your Inbox
To consistently stay at inbox 0, use Zapier + OpenAI to automatically draft email responses for you. Then, fully drafted emails will be waiting in your drafts throughout the day.
Simply send each new email you receive to OpenAI as a prompt. I like to add a step to send a text (or Discord message) to myself a notification, just in case the email is from someone I need to respond to immediately.
You can improve this further by:
Adding additional steps to generate subject lines
Adding a step to check the length of the prompt before you send it to OpenAI (this will reduce errors)
Build a custom webhook that teaches the model how to automatically respond to certain types of emails on its own, and what questions to ask you if it doesn’t know what to say back
Currently, Zapier only works with GPT-3 but I imagine that will be changing soon.
Keep Up With Industry News
You probably don’t have time to keep up with all of the relevant industry news in your space — you’re busy building your startup.
Obsessing over competition is a distraction, but ignoring it is a blindspot.
GPT-4 can accept up to 32,000 tokens per prompt. This means you can send it a really long piece of text for it to process, without issue.
Rather than spending hours reading articles, industry newsletters, and white papers you can simply send each one to GPT-4 and have it summarize it. You’ll be able to digest 10x the information per hour of reading.
Here’s an example. This is how GPT-4 summarizes my post from last week just by giving it a simple prompt to “summarize this and extract the main takeaways”:
My newsletter only takes ~5 minutes to read, but GPT-4 gives you the learnings in 10 seconds. If you only have 15-30 minutes to read each day between meetings and work, it can be a huge benefit.
There are also tools that let you do this with videos on YouTube as well.
Join Meetings
This sounds simple but can be very powerful.
Use Otter.ai to automatically transcribe your meetings into text. Then, feed that text into GPT-4 at the end of the meeting. You can easily ask it to do some simple tasks:
Organize and summarize meeting notes
Identify action items and owners
A simple prompt you can use for this:
Here are notes from a recent meeting. Attendees were x, y, z. Please summarize the notes with bullet points and, separately, extract action items and who each action item’s owner is.
Going one step further, you can add some custom functionality (Zapier has a guide on how to do this):
Automatically email the notes and action items to everyone who was invited to the meeting (even if they didn’t show up)
Create and tag Notion / Asana / Linear / etc tasks based on the action items, timelines, and owners discussed in the meeting
Where this gets really powerful, though, is if you hook GPT-4 up to Speechify or Murf and let it respond, verbally, during the meeting. If you give it a detailed prompt at the beginning of the meeting with full context about the meeting’s purpose, it can:
Propose ideas
Identify blindspots
Help resolve conflicts
Executive Coach
This week, a former member of the team at my last startup went viral on Twitter after posting that he was starting a company with a modest budget and letting GPT-4 make all the decisions:
I gave GPT-4 a budget of $100 and told it to make as much money as possible.
I'm acting as its human liaison, buying anything it says to.
Do you think it'll be able to make smart investments and build an online business?
Follow along 👀 https://t.co/zu4nvgibiK
This is the extreme end of the spectrum, but you can leverage GPT-4’s knowledge of the advice that leaders like Matt Mochary, Steve Jobs, or even Tony Robbins would give you when you’re dealing with the hard and ambiguous challenges founders work through every day.
Startups come with hard decisions that have tradeoffs rather than obvious solutions. Especially if you don’t have the budget for an executive coach, you can leverage the collective knowledge of many based on what they’ve previously shared online.
Just be sure to be extremely thorough with your prompts in these situations, since the decisions it will be helping you make are likely incredibly nuanced.
SEO
SEO is not simple, but it is straightforward: accumulate high quality backlinks to rank higher and drive more traffic.
Nerd Wallet famously built a massive business on the back of SEO-focused content.
Boutique firms can help startups do this too. There are many that create websites and build their reputation specifically so that they can provide valuable backlinks, but they are very expensive — even though the content is straightforward.
A tool like GPT-4 could easily churn out hundreds, if not thousands, of articles to help build the reputation of a site in a similar fashion but 100x more cheaply and quickly.
The irony, of course, is that tools like ChatGPT make SEO less relevant as a higher percentage of users use chat-based interfaces for search. But today it can still be a very defensible growth strategy.
Customer Success
Chatbots were supposed to solve customer service years ago. Turns out, they were just too early.
Intercom is already preparing their LLM-powered customer support agent, and they actually got beat to the punch by user.com. I see no reason these tools won’t be able to handle most customer concerns. You could even, again, hook it up to Speechify to answer phone calls as well.
Customer success is often a cost center that founders don’t invest enough in or cut too quickly, because the impact of a poor experience isn’t as obvious as issues with the product. This should make that choice unnecessary.
Legal Docs
Legal documents are extremely templatized, and for almost all types there are many examples that exist on the internet. That makes them a perfect use case for large language models.
Rather than paying your corporate counsel hundreds per hour to create some templates they likely already have on file, just ask GPT-4 to do it.
Be descriptive with your prompt and talk about the use cases for the template you’re creating.
Data Analysis
Copy/paste data from a Google Sheet along with a list of questions, and GPT-4 can return answers but also genuine insights and recommendations.
This can probably do the work of a junior analyst (without the Python learning curve).
Relatedly, an early player in this space that seems promising is Olli.ai.
I’ve been a fan of Kernal for a while (it’s basically Product Hunt for ideas) but just discovered they have a newsletter that shares new startup ideas each week.
There are honestly too many new AI tools coming out each day to keep up with right now. My go-to places to find them are Ben’s Bites, Built with AI, and Superhuman.
Whenever you’re ready, here are 3 ways for us to work together:
Grab time with me for a 1:1 session on fundraising, GTM and growth, hiring, finding PMF, or anything else
Promote your startup to 23,000+ founders by sponsoring this newsletter
My venture fund is actively investing in AI startups (and all other verticals). Share your deck here and we’ll take a look