The synergistic integration of artificial intelligence (AI) and the geographic information dimension creates geospatial artificial intelligence (GeoAI). Geo-tagged big data collated from varied sources, such as satellite imagery via remote sensing, IoT sensors in smart cities, social media streaming, and personal sensing via connected ambient and wearable sensors, can be analysed using GeoAI to get actionable insights. 

Founded in 2017, Attentive AI was started by an IIT Delhi core team comprising of Shiva Dhawan, Utkarsh Sharma and Sarthak Vijay. The company was established to develop artificially intelligent systems that can analyse petabytes of geospatial imagery and convert it into accurate insights. It serves geospatial technology providers and end-users with 2D and 3D vector data extracted from satellite, aerial, street and drone imagery.

The mission of Attentive AI is to convert all the data collected by satellites, aeroplanes and drones into actionable insights, and help businesses, governments and non-profit organizations in reaching meaningful conclusions and making better decisions faster.

Every startup has its challenges and everyone should be prepared to tackle those, said Shiva Dhawan, founder and CEO of Attentive AI. “The biggest challenge for any startup is running out of money. Some run out of funding whereas some run out of paying customers,” said Dhawan.

He further said, “From the very beginning, we knew that we would face the latter problem rather than the former. However, we were also confident that overcoming this problem would lead us to build a sustainable business, and so we relied on our customers to sustain ourselves as well as to grow. As a result, we tried to be as frugal as possible with our expenses. For instance, for a long time, we did not work out of a high-end corporate office instead, we worked out of a couple of apartments which were converted into modest office spaces.” 

The key to success is to build a team that understands the importance of frugality and enjoy the startup journey just as much as you do, added Dhawan.

Attentive AI’s approach involves experienced computer vision team that prepares deep learning models and are trained on geospatial imagery data, in-house annotators process machine-generated data which fix gaps and inaccuracies, expert in-house quality control team that ensures nearly 100% correctness through pain-staking visual inspection, and client monitors live production until the map features are delivered in the desired file format.

Standing Out Of The Crowd

With the Chinese government leading the AI race with its expansive surveillance system and heavy investment in AI research and skill development, the Asian governments still lag in terms of implementing AI initiatives.

With such a scene in hand, Attentive AI has come up with an innovative solution to deliver the best GIS solutions using high-quality digital maps. MapX is Attentive AI’s flagship product that can be used to request meaningful insights from geospatial imagery data. The core of the product envelops a simple workflow allowing each user to select an address or an area, followed by requesting the analytics they desire from a list of available analytical options or can even go for a custom request followed by near-instant delivery of output/insights.

MapX prides itself on being the ‘McDonalds’ of the geospatial world. Thus, one does not have to wait a long time to get the geospatial dataset or a digital map. “Initially, when McDonald’s had started, people said it was impossible to deliver high-quality burgers in such a short time but they did it, and their key to success was their engineering process. Similarly, delivering high-quality land features almost instantly was also impossible, but Attentive AI makes it possible by an engineering process at the core of which is our AI algorithm supported by a detailed QC process,” said Dhawan.

Being a customer serving web platform, MapX also makes the experiences of ordering geospatial data very seamless for customers and users. “Our customers attest to the seamless geospatial data ecosystem that we are creating.”

Surviving Industry Challenges

AI, being one of the newest innovative technology with a maximum number of commercial benefits, comes up with several industry challenges. The major limitation in the AI industry is customer awareness, or the whole concept of AI, where it is to be believed that AI itself is the solution to all problems. However, that is not true as AI technology alone can not provide a complete solution. AI does aim to solve the most complex steps, but the whole solution can only be created with a combination of multiple technologies including AI.

Explaining this, Dhawan said, “One of the major challenges of being an AI service provider is that a customer’s expectations. Usually, customers tend to have high expectations from AI rather than what is currently possible. And, this is mostly because of the increased hype around AI and also because a lot of companies project a higher AI capability than what is possible at the current stage.” 

“These companies show pilots on specific and suitable examples, however, the truth is that scaling an AI is incredibly hard and it takes a significant amount of time to build an AI that is accurate on all possible user scenarios. We, at Attentive AI, try to mitigate this challenge by educating our customers transparently about the training period and the processes that are necessary to make the AI scalable. Thus our clients understand that the AI will not be scalable from the first day, rather through an iterative and active learning mechanism, which will help it to grow, and be more efficient,” said Dhawan.

In the rising age of AI, the key to success is to build artificial intelligence systems for a specific niche datasource to harness the power of data network effects, which Attentive AI is acing with their high-resolution geospatial imagery. “This gives us a competitive advantage while making cutting edge breakthroughs every week since we have worked on multiple use cases over a period from which our AI systems are continuously learning,” said Dhawan.

Future Prospective

Attentive AI, being one of the AI service provider, aims to create an accurate, constantly updating digital twin of the physical world. “We have only touched the tip of the iceberg as we are creating more AI technologies to analyse aerial imagery in specific geographies,” said Dhawan.

“We aim to build a global repository of multiple geospatial imagery sources and a suite of intelligent analytics for customers to request analytics on drones, streets, light detection and ranging (LIDAR), and all other kinds of geo-data sources at any time from anywhere,” concludes Dhawan.