AI is a demanding technology and when building an AI startup, you need to build for scale from day one.
This is my second time at the startup rodeo. So as we were starting up my thoughts were on minimum viable product and go-to-market strategy. But we were throwing AI into the mix and an AI startup is a whole different ball game, as I came to realize pretty soon.
AI needs a specific mindset, it demands that you think about scale really early, and it requires a high level of commitment and capability in engineering teams that can build for and handle this scale.
Now, AI is almost an essential weapon in a SaaS startup’s arsenal. Think of all that SaaS products offer, from in-app chatbots and personalized alerts to predictive analysis and automation of repetitive tasks. There is not a single area that SaaS covers that cannot be improved with AI’s prowess.
Related reading: SuperOps.ai is here to superpower your MSP
But, AI is not a quick fix. And, there are challenges very specific to building an AI startup, as I found out.
Before I get to the challenges, I must say that these will not all be applicable to everyone building an AI startup. The industry they are building for and their use cases will play a major part. However, I do believe every AI startup will face the following four challenges in some form or the other.
So, on to the challenges:
1. AI needs time
If a startup says they are building AI or deploying AI from day one, it is not true. AI needs time, it needs tender loving care, and most importantly it needs to be fed a lot of data. AI needs training. In the beginning, an AI-enabled tool or feature cannot do much, but as it crunches data 24 hours a day and for days in a row, it starts learning and improving. For this to happen, AI needs to consume a humongous amount of data. Associated with crunching massive quantities of data are other challenges like computational power and storage space, each adding to complexities.
However, time, which is a critical resource for startups, is what they need to invest the most.
2. Customization & relevance
Many of us get excited about AI, because we think we will be building a wow feature that will draw attention. But, that is not the most effective use of AI. We need to drill it in that AI is not a specific feature. AI needs to be linked to solving a pain point. AI is an enabler. It is like what cloud computing is. This is an underlying technology infrastructure on top of which features are built.
The pain point that we are trying to solve is what matters. AI needs to be customized and deployed in a relevant manner to solve that pain point - then it becomes a powerful tool. If crafted well, even the most basic AI functionality can create exponential impact for a problem you are setting out to solve.
3. Production greater than building
In the early days of starting up, most startups and the engineering teams are focused on building algorithms and how to get it to work. But this is the lesser of the challenge. Taking it to production is the greater challenge. Because, you start with an algorithm, but that does not stay constant. The point of AI is that as new data keeps flowing in, the model learns and evolves. How can production take into account constant evolution? You need to factor in this evolution in production and create an algorithm and model that keeps evolving, only then can Machine Learning be effective. Very few in the ecosystem have the experience of building a model that works for constant evolution and making such an AI product live and taking it to market.
4. Adoption & education
No matter what you build and how good the AI-enabled product or feature is, it will succeed only if it finds a market. There are still many misconceptions about AI. People worry that it will take their jobs away and make them redundant. Most are wary of AI, as they still don’t know how AI works. Educating customers and marketing correctly becomes critical. It requires a mindset change.
Connecting these challenges together is the idea of scale.
We think of scale primarily from a market or sales or revenue point of view. But that’s not just what scale is all about. It is also thinking about the massive scope of the “AI project” and all the variables and aspects it needs to encompass and everything we need to prepare for. It is about ensuring your engineering is AI-ready.
For AI startups, it is crucial to bring in seasoned technologists who can shepherd the scaling up of infrastructure, platforms, tools and all the other technology that is necessary for handling AI.
Enter our Head of Engineering
It is not that a sharp change in technology is happening for the first time today with AI. Far from it. It was a similar scenario at the dawn of the SaaS era. Ritesh Rathi, our Head of Engineering, has seen and driven the evolution of SaaS.
Rathi (everyone in the industry knows him as Rathi) has a rich experience of over 20 years building a SaaS cloud platform. He started at a time when AWS, Azure and such services were not available to the public. Rathi was part of the core team at Zoho that built this platform. He, along with the team, navigated the challenges of availability, scalability, provisioning, monitoring and built a robust platform. Today, this platform hosts more than 50 plus SaaS services, hosted across more than 10 data centers, handles petabytes of data and caters to millions of customers. Rathi was instrumental in designing and shaping this platform.
It would be safe to say that Rathi played an integral role in shaping SaaS as we know it today. The experience he has at scaling “data” is unmatched. He also has the advantage of experience, of knowing how things were in the beginning and the decisions and pitfalls that marked that journey. So, it is not a surprise that Rathi is among the most respected technologists in the industry today.
We are at the second chapter of the SaaS evolution. Just as he was there at its start, Rathi is here to shape the next stage of SaaS that will lead to predictable intelligence. His knowledge of and expertise in managing data will be of critical importance to scale AI.
We are at the doorsteps of the future and we need to bank on our past
It is true that everything we do today has to be done fast. The time for feature launches has shrunk, customers expect instant response and resolution, and market feedback cycles have contracted. Startups definitely need to put an emphasis on speed.
But that cannot be at the expense of scale. You cannot think about scale later; you need to build for scale right from the beginning. Which is why I am thrilled to have Rathi join us as we embark on this ambitious journey of transforming the future of IT with AI at its core.
We can’t wait to build the future.