By: Anna Khan & Veronica Orellana, CRV
As an entrepreneur, you have definitely seen all the media attention around AI ever since ChatGPT was announced — which led to an even faster flurry of new tools and LLMs, and eventually funding into the space.
But if you’re a software developer or founder, what’s even more important is how does this new wave of technology help you? What can you do this month, this quarter, or this year to use the benefits of AI to accelerate your company?
We’ve gotten this question from a lot of founders we work with, who may or may not be building an AI first company. They understand that AI is a foundational platform and technology that they will have to incorporate just like cloud or mobile was almost a decade ago. But they don’t know quite where to begin, or if it is worth derailing their existing roadmaps and plans.
Here are three ways you can tactically think about AI as a software company founder:
1.Pick your strongest, stickiest workflow — and accelerate it with AI. Lots of founders panicked with the release of AI, assuming that it meant they had to become an AI first company. But we believe that’s the wrong takeaway. The first interaction that catapulted AI into the zeitgeist was to help us a consumer get answers to questions quicker, more succinctly, and more easily. ChatGPT created a UI that the average customer was more adept (and comfortable) interacting with. In other words, ChatGPT brought AI to the masses, and out of the ivory tower.
There is sometimes fear around testing AI in a revenue-generating product. What if it has bugs? What if it annoys a customer? If you feel some hesitancy in a specific workflow — find a small risk use case until you’ve fine tuned the AI well enough to understand its limitations. This is why one of Zendesk’s first AI use cases was agent assistance for responding to customers. If the AI-generated content was accurate, then it would massively increase productivity. If it was inaccurate, there was no risk since agents still had control over what was being sent to customers. Zendesk focused on a high frequency use case that was internally facing which lowered the initial risk. This could be a helpful first step in testing AI within your product.
Encord, a CRV portfolio company, based in London — has also used AI to accelerate their strongest workflow — labeling datasets. Annotation methods typically require scaled workforces that often have to sit offshore for cost reasons. The Encord team believes this constrains quality and does not work for use cases where data privacy is important. For several years, Encord has been building AI-assisted labeling powered by micro models. This has led to many inbound customers in the government, healthcare, and deep tech sectors that saw Encord as critical to their workflow. Encord was comfortable testing AI in a high impact workflow given their deep experience in the field.
Similarly, Northspyre, a vertical SaaS company for real estate owners and investors uses AI to accelerate a critical workflow in their product. Northspyre’s product offers proactive warnings for things like budget line overruns or contracts that are overspent. You’re sent an alert before overspending occurs or unexpected change orders take place.
2. Expand your product through AI: Looking to expand your product and launch a new revenue stream? Never before has it been possible to launch a new sub-product as easily as today. Whether it’s supercharging search through NLP, automating text, images, voice or videos, AI has compressed product roadmaps by years. One of the advantages startups with strong customer bases have is distribution. We’ve seen several examples both within our portfolio, and outside, of startups releasing AI features to their customer base and seeing very high attach rates that they’re charging a premium for.
And guess what? It doesn’t have to be perfect. Because AI has shortened development times — you can test 2–3 new sub-products at any given time to see which one is more sticky. This has created what we like to call the AVP, an “AI Viable Product” vs. a “Minimal Viable Product.” What can you push out that’s accelerated by AI to test with your customers?
Storyboard, another CRV portfolio company, is building the future of enterprise audio. Storyboard was able to accelerate the development of their AI powered voice messaging tool for frontline employees to better communicate with their managers via voice. For example, if a pharma rep or technician needed help remembering the steps to perform a task, Storyline could connect them to a manager via the app. Rather than a typically long voice call, AI would help truncate the steps needed to complete the tasks in real time. The employee could also refer to the summarized voice notes for future tasks — a much better use case than needing to access a company handbook or process automation bulletin while on the road.
3. Deepen your MOAT by leveraging your unique dataset: Data continues to be one of the strongest differentiating factors for creating value with AI. Since everyone has access to the same LLMs, fine tuning the model with industry or customer specific data is key to achieving a better “wow” factor. Whether it’s industry specific data that you’ve purchased, data that you’ve gotten through your customers or partnerships, or data that’s been created based on interactions your customers have had on your product — honing in on what’s unique to you should influence how you use AI to deepen your advantage and create an even better customer experience.
Our portfolio company Ciro, is a great example. Ciro is a sales platform for companies targeting vertical industries and local businesses. The team realized that one of the most time consuming activities their customers experienced was writing personalized emails and drafting outbound call scripts. Since Ciro has the best dataset for SMB companies in specific vertical industries, they’re able to leverage that to generate emails and call scripts with a click of a button. The example below shows the kind of information it can pull about a potential customer — that can help draft a personalized note to the business — a surefire way to get the attention of that business. They used AI as a shortcut for their customers, but the underlying data and workflows are what keep the product sticky.
We hope these three clear suggestions and examples offer you some help in navigating the rapid development of AI. First, pick your strongest, stickiest workflow. Second, test a new revenue stream through AI or as we like to call it, an “AVP”, and lastly, deepen your moat by leveraging your unique dataset. If you are a startup that has tried some of these paths to incorporate AI, we would love to speak with you!
