NuntiumAI

This Will Be A Household Visual for EM Investors within 3 Years
Nuntium Allows Qualitative Query Refinement Across All Unstructured Data, Making Human Research More Efficient

Inspiration

Raffay, being from Pakistan, and Adi, being from India, understand the difficulties of information accuracy in the EM markets, especially the information profiles that inform investable opportunities. The team is also painfully aware of the desperate need for investment in the rest of the world, and hopes that improved returns will ensure more capital goes to the right opportunities in EM markets. In order to accelerate that process, we built Nuntium AI.

What it does

Nuntium AI offers a set of tools to make EM investment research easier. We have 1) unique data pipelines, including open sourced data like GDELT and custom client-specific pipes, 2) pattern analysis tools that leverage vector embeddings to find common themes and sentiment analysis in text based data, and 3) an LLM powered front end that allows users to pull data from unstructured data sets with simple questions.

Nuntium will allow EM investors, advisors, and policy-makers to make better, data based decisions. Instead of trawling the web for information light sources, or trying to find specific information using human contacts in country, Nuntium allows any investor (large or small) to access huge swaths of available information and find the best evidence for and against their investment hypothesis.

Today, Nuntium does not offer investment advice or investable products direct to retail investors. However, as our algorithms and data products improve, that may change. But for now, we are focused on supporting people with strong investment backgrounds succeed in EM markets.

How we built it

Raffay and Adi originally wanted to automate the work of consultants, having both spent hours (cumulatively, weeks/months) putting together research reports for our managers/SMs that required substantial volumes of Googling followed by narrative construction.

However, the team quickly realized that "AI consulting" was a very crowded space and we lacked differentiation. More than that, we were getting traction in the EM space, and so we went where our clients told us to go.

During the pivot, Raffay took a step back and thought about the real potential of LLMs: Linguistic Analysis of text to level of nuance not possible before. Where is this nuance most lacking or rather where is the contextual relevance extremely important when analyzing text, foreign unstructured information which was inaccessible in prior NLP techniques because the mechanisms lacked contextual awareness. Thus, analyzing non US or even non English unstructured information is where a lot of the value of LLMs can be leveraged.

Adi, on the other hand, was an investment analyst on Goldman's Growth Equity Team (and his dad was a hedge fund founder at Vernier Capital). While this expertise has been helpful, it's taken a backseat to the tactical feedback from Nuntium's active users, who have been very proscriptive in what they want from the product and it's tools. Adi's experience is mostly used to facilitate conversations with these users: he can speak the same language these investors do.

At the end of the day, more importantly than the rest of this long answer, we know people need Nuntium because we are building what EM investors say they need.

Challenges we ran into

We currently are spreading via word of mouth. The EM investor community is not large, and, because of the historic lack of returns in the sector, many firms are hesitant to focus their technology budgets on EM markets. The high barrier of trust in the EM investment space makes our engagement and sales cycle very slow and very methodical. Worse still, each and every bug sets us back not just with the product itself but also with our user relationships, pushing us to make our core products as sticky and as reliable as possible.

Accomplishments that we're proud of

When a user was experiencing a technical problem, we fixed it in 12 hours from email notification to pushing to prod to closing the loop with the user. This wasn't "4 hours of work done in 12 hours", it was "12 hours of coffee and code". The best part of this entire story was the technical issue was so meaningless to the core product: it was a UI rendering issue that simply bothered this one user. But at the time it felt like the end of the world.

What we learned

The best B2B SaaS companies sell products that people need, not that people want. But need is a function of user dependency. Most users are perfectly happy to continue with the status quo processes, even with their inefficiencies and inaccuracies, because they have inertial incentives to "do what they've always done" However, if a SaaS product flips the script - if people cannot work without Nuntium, then we can position ourselves to take over the entire industry one analyst desk at a time.

What's next for NuntiumAI

Today, Nuntium does not offer investment advice or investable products direct to retail investors. However, as our algorithms and data products improve, that may change, allowing us to expand the scope of the pipeline connecting Western capital to EM markets. But for now, we are focused on supporting people with strong investment backgrounds and existing pools of AUM.