Illumea: Building Intelligent Chatbots for Deeper Business Insights

I was working on an interactive educational bot to help children understand math when a friend of mine asked me to evaluate chatbots for a customer. While doing my research, I found that while most chatbots were feature-rich, providing analytics and integrations with multiple systems, not many really provided deeper intelligence that could help businesses uncover actionable insights from conversations. Filling the intelligence gap requires mapping business objectives to the data along with various inputs like market and competitor intelligence. These insights could flow into their operations - sales, marketing, etc.—as a strategic report at the end of every month. Also, for many larger customers, chatbots become an integrate-and-forget system, and as knowledge bases grow or due to lack of maintenance, the responses are fetched from stale data. Just flagging stale data doesn't help here; actions need to be drawn, communicated, and tracked to ensure that information gets updated.

Worked, Since I was new to the market, a lot of word-of-mouth and referrals via friends helped me get into conversations and early feedback. The objective was to build trust and start conversations about the product. On one side, I was able to generate interest and convert. On the other side, it also helped in validating the idea with some high-placed industry folks who represented my target customers. Flopped, Social media strategy didn't work. I am not a heavy poster on social media, which is probably why my posts on LinkedIn didn't work as expected. While I got a lot of clicks from many connections who were mostly friends, conversations didn't happen as the reach and proper targeting for these posts wasn't enough. Also, I didn't try other channels like YouTube, Reddit, or X, so that is going to take some time to build and pick up as well.

Till a few weeks ago, customers (and prospects) were a bit confused about the pricing, which caused regular friction and probably loss of sales. When I started, I fixed the prices thinking as a developer from a component level, splitting pricing into various costs, plans, and buckets. Customers had to select various components to build the plan and get an estimate. They had to wait until the final proposal to know actual pricing. While I thought this made customization easier, it backfired since customers wanted to directly know how much to pay. The reason for this confusion was mainly due to a spread-out infra setup, multiple AI costing mappings, and, more importantly, covering for currency rate fluctuations. Once I hit v1.0, I knew this wouldn't help scale, so I moved to new infrastructure at better rates which helped me simplify into having just one monthly retainer that covered everything.

When you take a look at the chatbot market, it looks like it is very overcrowded with even teenagers building chatbots and hosting on hobby Vercel+Supabase plans. The truth is that this market is growing and evolving, with 2030 estimates indicating this as a $30 billion market. On the other side, when I looked at the number of medium to large websites on the internet, I noticed that there is still a lot of scope to sell. The penetration into many industries is less, and in many industries, they are not aware of the value chatbots would give them. They look at it as a fancy search functionality which might not be needed, or they see it as an idle investment since the number of visitors who might ask a chatbot something might be far lesser. While they are difficult to convince, we need to change the pitch strategy to a more educative approach or go to them with insights that could help you have informed conversations about how the chatbot would fit.

For the next 2-3 months, my primary focus is on doing better at marketing and sales. The plan is to onboard at least 6-10 growth or scale plan customers in the next couple of months. As the sales pick up, I need to integrate the Intelligence report (which is in beta currently) into the main dashboards and monthly reporting. The next big feature is integration with Voice AI, for which the planning is done, but we are still working on architectures, costs, and other nitty-gritties.

We have just started marketing and talking to prospects. Onboarded one customer and probably another 2-3 by the end of August.

Out of my 20+ years of overall experience, I have spent the last 12 years designing and building data intelligence applications at Akamai Technologies. The reports generated from applications help customers understand how their subscriptions are working for them, with insights on how to improve or take actions. On one side, it helped me take the right technical decisions to build the product, and on the other side, it helped me understand how to work with businesses and define value better. Also, my mixed skillset in creative and tech helps me in building a good tech product with a great user experience.

I didn't do enough research on existing solutions and jumped a bit too fast into building since I had a good set of features to implement already lined up. I only stuck to the top 4/5 chatbots and slowly discovered more during my conversations or while planning marketing activities. This made my feature list grow rapidly and exponentially in the beginning, making it hard to keep up with deadlines. Good market study and targeting are important to scope out what is relevant before you jump into building since the market is evolving rapidly and directions are changing towards voice, for example.