Original article was published on Artificial Intelligence on Medium
Augment the knowledge worker
With solutions becoming cheaper and more available for end users, the main criteria driving end user adoption when choosing one product over another is which will make his/her job easier or more productive. This mindset aligns well with Intelligent Software — solutions that leverage artificial intelligence and machine learning to allow digital knowledge workers to be 10x more productive and effective in their job. By automating labor-intensive tasks as well as combining machine intelligence (the ability to mine vast amounts of data at scale and detect patterns that humans alone can’t find) with human intelligence (the ability to bring context recognition, business judgment, creativity, and common sense into the equation), Intelligent Software solutions enable employees to be orders of magnitude more efficient, faster, and productive.
A great example of an intelligent software company in the product-led growth era is Gong, a Series C startup backed by Sequoia, Norwest, Wing and others that provides a sales intelligence platform for sales teams to garner relevant and subtle insights from customer calls/demos to turn into actionable revenue opportunities. Gong’s product has allowed its customers to decrease onboarding time between 20–50%, increase win rates by at least 10%, and increase overall deal sizes by 35%.
Address the user pain point
In order for users to choose a product, it must effectively address a critical or common pain point. Many AI startups struggle with this because they get caught trying to build the most technologically advanced product and miss addressing a market need. An end user-focused approach when developing solutions allows companies to address critical customer pain points and demonstrate meaningful ROI in a short amount of time. This meaningful ROI in the context of intelligent software is usually in the form of step-function improvements in productivity and capability, leading to increased accuracy and time saved. This is most easily achievable by prioritizing industry relevance. By banking on depth rather than breadth, intelligent software startups can build highly specific AI applications for explicit market or industry use-cases, including medical diagnosis, mortgage lending, supply chain management, retail inventory optimization, and customer service.
Transcription solution, Verbit, is a good example. Verbit operates a natural language understanding platform that combines proprietary AI models with human-in-the-loop exception handling, to provision a smart transcription and captioning solution. Unlike Amazon, Nuance, IBM, and Google who all use a one-size-fits-all model for their transcription solutions, Verbit develops industry-specific models and further develops and refines customer-specific sub-models. This allows the solution to achieve higher accuracy levels and handle more complex use cases.
Release early and performance will follow
A common term used when speaking about intelligent software is the idea of a feedback loop, the process by which an AI model’s predicted outputs are reused to train new versions, subsequently increasing its overall accuracy. In other words, the more a product is used, the more effective the solution will become. When thinking about how a product-led growth solution is sold, this concept of a feedback loop becomes a key factor in terms of increasing both user retention as well as distribution. As users begin adopting the product, usually free of charge, they will not only be incentivized to continue, but also promote it to additional members of their team in order to increase product performance gains at a faster rate. The combination of increased usage, as well as a growing user base, allows the product effectiveness to improve at a rapid pace and create relatively defensible switching costs. Additionally, as more users within an enterprise begin adopting the solution, pricing increases and new revenue opportunities arise. Assuming that the product’s base level performance fulfills the criteria of its potential customers, the nature of intelligent software mitigates the risks of operating in a space with plenty of optionality and low switching costs.
Take Grammarly, for example. While they didn’t necessarily start as product-led, the moment Grammarly switched to a freemium model, they were able to spark rapid product adoption by removing onboarding friction and introducing new product features such as tone detection, syntax error, and insights. As more users sign up, Grammarly’s ML models expand beyond the training set and learn from user-generated data, allowing the solution to become more effective. Additionally, more users allow for more human feedback, which is crucial in order to identify idiosyncratic mistakes or misinterpretations. As users experience this, they will be more incentivized to use the product themselves as well as promote to others. This has been a main driver in daily active usage to reaching ~20MM+ in the past year, a 20x increase since 2015.
As product-led growth companies become more common in the enterprise ecosystem, Intelligent Software will follow. Solutions leveraging artificial intelligence and machine learning to create productivity and capability gains provide the immediate ROI necessary to capture users’ attention, while the feedback loop and continuous improvement create switching costs and distribution channels that will accelerate time to scale. We are seeing this first-hand in numerous industries from early-stage startups to multi-billion dollar public companies and will continue as the next wave of value is created in the enterprise space.