Artificial Intelligence is one of the hottest topics today and is rapidly disrupting various sectors. With this wave, there is a growing interest amongst entrepreneurs to adopt AI to solve their problems. We at pi Ventures, being focussed on investing in AI companies, meet quite a few entrepreneurs working in the AI space across sectors like Healthcare, Logistics, Enterprise, Fintech etc. During the course of these interactions, we have seen some really disruptive use of AI. At the same time, we also come across entrepreneurs jumping onto the AI bandwagon without giving it a proper thought. We believe this is not only a waste of resources and energy, but can also sound a death knell for them.
Products that do not adopt AI will die a natural death, however the ones that adopt AI without a natural fit will die even sooner
We want to share our experience of exploring this topic via this blog and also provide a framework that can potentially answer two critical questions for entrepreneurs - Why should one use AI? And if using AI, when is a good time to get going with it.
Let us explore these one by one.
The Why Question
The first question that any startup with an AI ambition needs to address is “Why AI?”. The reasons can be multiple – customer demand, catch up with the competition etc. It is important to be as precise as possible in answering the WHY question. From our experience, we have seen that startups which are clear about using AI in a way which sets their product 10x apart in terms of value from other state of the art solutions, tend to do better with the use of AI.
Let us take an example and demonstrate this point. Niramai, one of our portfolio companies uses thermal images and AI to detect breast cancer in early stages. Since they use a modality which relies on deducting abnormal cell growth instead of density (which most other state of the art solutions use), they are able to detect breast cancer very early. This functionality cannot be delivered with enough accuracy without AI. Clearly, their solution delivers a unique and differentiated advantage over any other products. This is what we mean by delivering a 10x value (or the 10x factor)
One interesting point to note while discovering your 10x factor is that it can come from an insight which is led by your product / work (inside out) or it could be led by a customer demand which is difficult to meet unless AI is used (outside in). Irrespective of the trigger, the ground truth for answering the Why question is to discover the 10x factor for your product.
The WHEN question
The next step is to determine “When” and to some extent “How”. To answer these questions, we will take the help of the framework shown in the diagram below. We call it “The AI Maturity Curve”.
Let’s explore this a level at a time.
Every startup / entrepreneur starts off by solving a problem. Sometimes that itself consumes a large part of their time and attention. As a consequence of running the business and interacting with customers, the startup might get exposed to or generate tons of data, which may or may not be stored with thought and design. This is a typical characteristic of a company in Level 1.
For example, India Post, which has the maximum coverage of pin codes in India, has until recently not been collecting the delivery data. Similarly, many traditional courier companies have also been delivering packages without storing valuable information around addresses, routes etc. Question is, if they were collecting this data in a smart and reusable fashion, could it build a way for them to create a 10x factor for their customers!
If you are finding yourself at this level and have a keen desire to move to the next, here is what you could do:
Think through what data do you get in touch with or produce as part of your normal business. Think about the specific insights that can be derived from it to build the 10x factor, immediately or in future
Now that you have a data collection process and strategy in place, it is time to dig in and create data moats and work on specific insights the data can generate for you as you build your startup. Companies at this level are seen having a clear data strategy in place. The data strategy encompasses collection of relevant light data (more readily available data) as well as dark data (data that is not easy to get hold of). Relevant data, depending on the sector could come from medical records, web text data, media etc. It is key to know which data is important and also which data is proprietary to you. Thinking through and implementing the data strategy helps companies move to Level 3 successfully. Sometimes it can be confusing to decide what data to collect and what to leave out. If it is difficult to foresee the real impact of data you are collecting but have a hunch that it could be useful, it is recommended to collect it nevertheless. Golden rule being, if in doubt, better to collect than leave.
To map this concept to a real life example, let us go back to the courier companies we picked in Level 1. Now if they had been collecting delivery related data through smartphones/handheld devices in a reusable fashion, they would become Level 2 companies. Light data in this case could be customer and package details, while dark data could be location details that can help discover and map non-standardized Indian addresses. This location data could, for example, help eliminate the need to call customers to find their exact location and therefore lead to greater operational efficiencies.
Following are some recommendations if you are finding yourself at Level 2 and want to move to Level 3
We are getting closer to using AI now. In fact this is the level where startups are using AI extensively to solve problems with a 10x factor. They leverage data sets and typically have a feedback loop that trains and improves the models. These companies may not be fully autonomous and have human intervention built in. However, the key factor that differentiates companies at this stage from the ones at Level 2 is that AI is able to derive valuable data driven decisions to solve business case. It could extend from detecting breast cancer from thermal images as explained in the example earlier or suggesting the route to take to get home, as in the case of Google Maps.
To extend our favourite courier companies example, they can build real time analytics and route planning and optimization algorithms to bring in operational efficiencies.
There is another interesting factor to keep in mind if you are a level 3 company. Sometimes the impact of the decision is huge – an erroneous decision in a cancer detection system can have a deep impact on the life of a patient. Therefore, depending upon the use case, it is recommended to have a human supervision layer even if the algorithms are fully automated.
Now we come to the final Level 4. Here, companies typically have a very advanced level of AI, leading to completely autonomous solutions which do not require any human intervention.
An example of this is the self-driving car, which uses AI and computer vision to accurately interpret sensor data, detect objects, map the environment, and handle wide range of variables to make decisions like a human. Although some time away from commercialization, these fully autonomous vehicles will not require any human intervention. Another example is Amazon’s Alexa, a voice assistant that uses natural language processing and machine learning to understand the question and respond in a human like fashion. With recently announced features like memory and context carryover, Alexa will be able to make contextual, intuitive and smarter conversations.
In conclusion, AI can really help build the 10x factor in your product! The key is have a clear insight on what AI will bring to your product and at what stage does it make sense to adopt it. We hope this blog can bring clarity on those aspects!
We just sent you an email. Please click the link in the email to confirm your subscription!
OKSubscriptions powered by Strikingly