Quick Takes 2023 Trends: Why Generative AI is Next Big Thing

What is Generative AI?

Generative AI is a type of artificial intelligence that is focused on creating data rather than simply processing it. It is a powerful tool for creating new content, such as graphics, text, audio, and video. Generative AI algorithms can be used to generate new variations of existing content or to produce entirely new content. It can be used for various tasks, including image and text generation, natural language processing, and audio and video synthesis. 

Why did Generative AI take off in 2022?

Recent advances in natural language processing, such as OpenAI's GPT-3, Google's BERT, and Microsoft's Transformer, as well as advancements in computer vision and machine learning, have enabled the development of generative AI algorithms that can generate realistic images, text, audio, and video.

Why is everyone excited about Generative AI?

Most articles tell founders to solve a problem that requires an aspirin not a vitamin. Generative AI is like Viagra it enables customers to do something they have never been able to do before. This idea was borrowed from Sarah Tavel's great 2016 post here.

Compare Generative AI to your typical SAAS or Marketplace startup. SAAS companies automate a process and typically save a customer time (e.g. Salesforce makes it easier to track customer relationships and next steps). While the marketplace removes the cost of search and transaction costs by making it easier to find a counterparty. Customers have a problem and expect software to solve it. Compare this to Generative AI where you can have the software generate the first draft of contract terms. Until now lawyers did not expect this. Software like Spellbook enables them to do something they never thought they could do with software. The result is, Generative AI companies are seeing rapid adoption and growth.

What will 2023 bring?

As the core technology behind Generative AI is open sourced you will see it appear everywhere. Non-generative AI companies will be incorporating Generative AI into their products (e.g. Notion) and many startups will be using Generative AI to solve new problems. The fundamental change that Generative AI represents will encourage every regulatory body, licensing body, and politician to recommend policies on how to deal with Generative AI. New job titles will emerge as Generative AI specialists emerge (e.g. seed word specialists). Generative AI may provide tailwinds for blockchain technology as the original creators of the work may want to validate that they created/seeded the work.

What do you think will happen with Generative AI? Let us know by replying to this email.

Profile: Scott Stevenson of Spellbook

What is Spellbook?

Spellbook is an AI Copilot for Lawyers. Using generative AI, it automates aspects of contract drafting & review. It works in Microsoft Word, the application lawyers spend most of their time in!

What was the insight/inspiration that led you to launch it?

I'm a software engineer by trade, and I had an amazing experience with GitHub Copilot, which helps programmers write code using the same technology. The parallel to contract drafting & review seemed uncanny, so we built Github Copilot for lawyers! One of the first times I used GitHub Copilot, it output 20 lines of code that I thought was garbage. I was skeptical. Then 10 minutes later I realized that Copilot was anticipating a problem that I wasn't. That was the first time I felt outsmarted by AI. It was when I knew this technology would change how programmers and other professionals work forever. 

How has it been received by the market? 

Unbelievably well! We have 5x more demand than we can handle! We are onboarding 35 law firms and legal departments this week! Shortly we'll be onboarding 100/week. We are rapidly scaling our team to digest. I guess this is what product-market fit feels like! Very tiring and very fun. I think the product is being received well because of what James Currier talks about in Find the Fast Moving Water: 

Lawyers are smart. They don't need to be told: "what problem we are solving". The moment they see Spellbook in action, they understand that their profession is about to change forever, the same way I did when I tried GitHub Copilot. It sells itself.

How did your work at Rally lead to this product launch?

Our mission has been to reduce the friction of business legal transactions with technology. Every day countless businesses are formed, employees are hired, sales are made, IP is licensed, and funds are raised. What if all of these things happened 10x easier, 10x faster? It would improve the planet's productivity dramatically. We spent 4 years working closely with lawyers and the businesses they work with to understand their workflows, their pains, and their desires. We built a lot of automation tooling. All of this insight went into Spellbook.

How did you know you were ready to found a company?

It's always been in my blood somehow! I tried a lot of terrible business ideas before I had any good ones, like selling MS Paint drawings on floppy disks.

How did you meet your founders/know they were the right people to start a company with?

I'm super lucky to have met Matt & Daniel. Matt's an engineer & designer and Dan's a lawyer. Those were the two skillsets I was looking for, and they were looking for an experienced developer. We brute-forced meeting each other through networking (our first angel actually ended up introducing us), and then we worked together for a trial period for a couple months before cementing things! It's worked out well.

What has been the hardest thing on your journey so far? 

The grind. Founding a company requires an intense grind, there is no way around it. I love what Dalton Caldwell from YC says here:

The reason for this is that you need to constantly Do Things That Don't Scale to find product-market fit. We probably ran 200+ different experiments before hitting true PM-fit. It's very hard to delegate all the grunt work involved. It's hard to train in sales reps when the playbook changes every week. But after accepting the grind (rather than running from it), I've really come to enjoy it.

What has been the most pleasant surprise?

There are a lot of poised storytellers in business that are good at raising money and looking successful. I was pleasantly surprised to learn that a lot of that is smoke and mirrors, and the bar for real execution in startups is not that high. If your team can ship daily, run new experiments weekly, navigate pivots and mentally commit to a 10-year journey—the odds of success are surprisingly good.

What is your super strength?

High tolerance for chaos, mess, and ambiguity. Building new things is messy! Creative flow can't really be controlled, if you try to control it too much, you kill it. I think we avoid the unstructured game of creation by constantly coming up with structured things to do instead. Read another book. Clean the house. Set up some new tooling. Make a new spreadsheet. These things guarantee satisfaction. But creative magic doesn't happen until you are really comfortable diving into the unstructured. I wrote about this in How to Finally Make Something.

What is your proudest accomplishment outside of Rally?

My proudest moment is the first time I completed and shipped a complex creative project through my own initiative, which was a crappy album of electronic music, over a decade ago. It took me 10 years of banging my head on the wall to figure out How to Finally Make Something! I took those lessons to programming, startups, and everything else I've worked on since. It was a really important moment in my life.

What advice do you wish you had been given prior to launching Rally?

Founders really need to demonstrate conviction in their beliefs to succeed. No matter how great your team, investors & advisors are (N49P is awesome!), you are going to have the most context. If you bend like a wet napkin and don't demonstrate conviction, it makes things harder for everyone. If you don't fill the "conviction void", someone with less context (rightfully) will.  

What do you think the future of Generative AI is in the short term (e.g. a year)? Long-term (5 to 10 years)?

In one year, I think the biggest advancement we will see are action-oriented models (which I call "ActGPT"), which are oriented not just around language, but around actually taking actions in the world, like updating spreadsheets, pushing data to APIs, etc. I think we will have AI team members in our Slack channel within 12 months.

In 10 years: Terrifyingly hard to predict! I think our lives will become a lot more illegible. Every company will start to look more like a high-frequency trading firm. New product features will be mocked up by AI the second customer feedback comes in. AI will generate new ads overnight as a new trend emerges.

What are the Canadian Generative AI companies you like? 

1. Cohere

2. My head has been too far down to know many besides the infrastructure providers like OpenAI! I love GitHub Copilot though.

Other Canadian Companies doing interesting things with Generative AI:

• Rally - a legal tool tailored for businesses and their lawyers (they also just announced Spellbook, which we touch on above)

• Durable - an AI-powered platform for solo business owners to assist with everything from website creation, to managing finances, and everything in between

• WOMBO.ai - unleashing creativity through the power of AI

Let us know more about the Generative AI companies you like and that we should be talking about. Reply to this email and share your thoughts!

Top Headlines and Content

• If you are a founder this post will help you grow your marketplace more quickly and efficiently 

• Browsers: An Answer to Walled Gardens - READ HERE

• Advice for startups fundraising in the current down market - READ HERE

•  Founders and Funding: Wondeur AI and Two Small Fish Ventures - Eva Lau, Founding Partner at Two Small Fish Ventures, Sophie Perceval, Co-Founder at Wondeur Ai, and Olivier Berger, Co-Founder at Wondeur Ai chat through the VC/Founder relationship and the raising process overall - LISTEN HERE 

*ICYMI - Allen Lau, Eva's partner at Two Small Fish Ventures joined us at a recent TechTO Event where he chatted with Alexandra Reilly about Wattpad and his current ventures at Two Small Fish - watch here*

TechTO Talent

Senior Software Engineer, Data at Cohere (Remote) 

Why is this cool? At Cohere, their mission is to build machines that understand the world and to make them safely accessible to all. The Cohere platform provides access to Large Language Models through its APIs that read billions of web pages and learn to understand the meaning, sentiment, and intent of the words we use in a richness never seen before. APPLY HERE

Data Scientist - Climate Risk Modeling at zestyAI (Remote, Montreal)

Why is this cool? Zesty.ai uses novel data gathering and data science methods to produce higher-quality information about the risks to property from catastrophes like floods and wildfires. Joining their team means making a direct impact on the safety and well-being of people across the globe. APPLY HERE

Senior Full Stack Engineer, Valence (Remote, Toronto, Canada) 

Why is this cool? Valence (formerly Shift) is the only teamwork platform for managers that accelerates the team health, connectivity, and performance required for today's workplace. They are also venture-backed and their digital platform was created to mimic the experience of executive coaching at scale. You will join a fully remote engineering team that's building their web product, having your hand in the expansion of the product overall. APPLY HERE

Events and Links

Check out the interesting content, events, and happenings for the community here:

• TechTO Together: January 16, 2023 - Kick-off 2023 with us! - REGISTER HERE

• TechTO Together: February 6, 2023 - Announcements coming soon! - REGISTER HERE

• Intuit Prosperity Accelerator: Toronto - NOW ACCEPTING APPLICATIONS until January 13, 2023 - APPLY HERE

Want to showcase your company, events, and opportunities to 60,000 subscribers across the Canadian Tech Eco-system? Reach out to [email protected] to learn more!