How Paralleldots Is Utilising AI To Impact The FMCG Industry

The Indian FMCG market is projected to grow up to $220 Billion by 2025. With rising consumer demand and increasing product categories, leaders are now supplementing their retail operation with an AI tech stack to simplify large-scale decision-making.

Outlook India caught up with Angam Parashar Co-Founder & CEO ParallelDots

Tell us about your company? How did you come up with the idea and when did you start?

ParallelDots is building an AI platform for retail with focus on computer vision. Our product ‘ShelfWatch’ helps FMCG companies and retailers measure and improve visibility and presence of their brands on retail shelves. We analyze pictures of retail shelves in real-time to extract meaningful information such as on-shelf availability of products, out-of-stock, pricing compliance, etc.

We started this company in 2017 as enterprise AI solutions firm. While working with clients in retail, we felt the need of a solution to analyze visual shelf data. So, in 2019 we started focusing exclusively on retail and FMCG industry by building solutions specific to their needs. Today ShelfWatch is deployed across 25+ countries with some of the largest FMCG companies and retailers. We have also grown our team to more than 100 people spread out across functions such as sales and marketing, customer success, data science, and product and engineering. We analyze 4M retail shelf images monthly. We have raised a total of $6M in venture capital from investors based out of US and UK.

What is your vision and mission for your ParallelDots?

Vision: Digitize offline retail globally. There are many parts that will lead to successful digitization of the offline retail market such as payments, inventory, shelf visibility, etc. – and we plan to touch all these aspects through a suite of products built over time.

Mission: Enable the transformation of retailers and FMCG companies using technology by driving operational efficiency across functions.

What were the difficulties and challenges you encountered while establishing ParallelDots, and how did you get beyond them?

We are a team of three founders – and none of us come from the FMCG/Retail background. This posed one of the bigger challenges for us to not only understand the industry in detail but also build trust among the industry stakeholders. Overtime we onboarded a team of industry experts from different countries to help us not just understand the global retail market but also bring in the necessary credibility.

We are all techies at ParallelDots. Since our clients are large FMCG companies and retailers – we all needed to come out of comfort zone of building technology to selling technology to large enterprises. Our first few sales were incredibly hard because of this. It required lots of patience and effort to turn the tech into product which results in business value for our clients.

We further learned how incredibly hard it was to sell AI-driven product to enterprises. It came with its own set of challenges including learning how to efficiently communicate with our clients. It also takes time to realize the impact of AI solutions for the enterprises it is more of a ‘crawl-walk-run’ journey, rather than outright disruption from Day 1. Often it is hard to achieve accuracy levels of 98% on Day 1. Building case studies and efficient and consultative communication helped us get over this challenge over time.

Why do you think that there is a need for your company to exist in today’s market? What differentiate you from your competitors?

With increased inflation, rising labor costs, and pressure from new-age D2C brands, FMCG companies and retailers are looking to gain operational efficiency in their retail execution practice. This operational efficiency gain should not only result in cost savings but give them an avenue to increase their sales. This is exactly where we come in. Not only we help our clients decrease their cost by reducing the time spent by their executives in the store, but we also help them increase their sales by helping them improve their retail execution KPIs in terms of reducing out-of-stock and increasing compliance.

We have built the strongest computer vision platform for visual shelf data with very high accuracy. Our platform is also robust enough to ingest new products and brand to detect them with high accuracy in a very short frame of time. Furthermore, we also differentiate ourselves on our customer success. With proven capability of deploying our product across countries and delivering clear ROI to our customers make us the best computer vision platform for this industry.

Can you throw some light on the emerging technology trends in the retail industry?

Retail world is changing. After the insane e-commerce growth during the pandemic, the offline retail is making a comeback with e-commerce growth slowing down. Digital transformation is of highest priority for large consumer good manufacturers and retailers for whom offline is still a disproportionately larger than online. According to a recent study, consumer goods manufacturers will spend $24B on digital transformation by 2030. A significant portion of this spend would go in adopting technologies such as AI/machine learning to drive efficiencies.

There is a larger push by the various ecosystem players to digitize the offline retail market. ONDC in India is one such example where government is building a platform to enable small merchants and retailer come online. This will help in further digitization of the offline world in markets dominated by traditional trade enabling even the smaller Kirana stores to sell online.

We are also seeing some early attempts to bring offline retail experience into Metaverse. However, it is too early to comment on whether it will gain traction within consumer or not.

What is the product roadmap in the years to come? What are your expansion plans?

Product Roadmap

o    Ability to train AI with less and less data. AI by nature is very data hungry. In retail world, where there are millions of unique SKUs with different packaging, it is important to have an AI engine that can digest new/unknown SKUs in a short time without compromising on accuracy. Very recently we launched our latest AI platform named ‘Data Flyweel’, which essentially allows us to train our AI models in a matter of hours by leveraging terabytes of data we have built overtime. It will further continue to evolve in the coming years.

o    Video recognition of retail shelf. In some cases, it is easier to capture the shelf through a video instead of images. While processing videos is heavier in terms of compute cost, it could give operational benefit while capturing the data. We are working to bring down the cost of processing retail shelf videos significantly.

o    Development of shelf camera. For categories which are fast-moving, it also makes sense to install a shelf camera to continuously monitor the stock on shelves.

 Expansion Plans.
o    Expanding our direct presence in markets like US and Europe in 2023
o    Going deeper in markets like India where we already have significant presence
o    Expanding our partners, who help us sell our product, in markets where we don’t have a direct presence.