In an era where convenience is defined by speed and precision, the convergence of geospatial intelligence and artificial intelligence is quietly revolutionizing everything from e-commerce deliveries to national defense. In this exclusive conversation, Brajesh Shrivastava, Founder and Director of Deduce Technologies, shares how his company powers real-time mapping, traffic intelligence, and autonomous mobility across 37 countries. From supporting India’s quick commerce boom to collaborating with DRDO during Operation Sindoor, Shrivastava explains why Deduce prefers to stay in the background while enabling some of the world’s largest organizations to stay ahead on the map.
TAM: How is the convergence of AI and geospatial intelligence redefining last-mile delivery efficiency for sectors like eCommerce and food logistics in India?
Brajesh Shrivastava: What’s happening in the industry right now is truly transformational. We are witnessing significant changes, especially in how people handle everyday essentials. It has become a part of daily life. For example, if someone runs out of rice, they simply open an app like Zepto or Blinkit, place an order, and within five to seven minutes, it arrives at their doorstep. It almost feels like magic.
But behind this convenience is a powerful engine driven by geospatial technology. Just like we once used to say “Intel Inside” or “Nvidia Inside,” today, almost everything that gets delivered quickly relies heavily on geospatial data. Without maps and geospatial intelligence, we wouldn’t have the kind of e-commerce or quick commerce experiences that we enjoy today. Geospatial data is now at the core of these services.
When it comes to how AI and geospatial technology work together, there is a lot happening in real time. For instance, consider the current weather conditions. It might sound simple to plan a route from one location to another to make a delivery, but depending on real-time factors like waterlogging, the plan needs to change instantly. That is where AI plays a critical role.
We are now using real-time satellite imagery, traffic probe data that comes from vehicles on the road, and other inputs to understand the environment and make intelligent decisions. One of the unique products we have developed is our Gated Community solution. This product covers over 200,000 gated locations in India and is also available in about 15 other countries. It helps delivery partners identify how to enter a particular gated community, which gates are allowed for deliveries, and which are restricted to residents.
These are complex logistical problems that need real-time solutions. Gate A might be open to residents and delivery personnel, while Gate B could be reserved only for delivery partners. Our systems use AI to determine this and ensure the delivery happens efficiently and accurately.
As consumers, we don’t usually see all this complexity. But in the background, there is a constant stream of AI processing multiple data sources to produce highly optimized routes, accurate ETAs, and smooth deliveries. This is where our company name, Deduce, comes from. It reflects our focus on intelligent deduction through geospatial data.
All of this leads to outcomes like receiving a precise notification saying your delivery will arrive in seven minutes, and it does. This is the big shift we are seeing, and it mostly happens behind the scenes. We often refer to ourselves as the backend team because we build and manage the systems that power these experiences without being visible on the surface.
TAM: How do you aggregate the data required to provide real-time insights? Do you have any partnerships in place?
Brajesh Shrivastava: Real-time data is essentially a combination of several sources, including satellite images. We are now receiving data through Bhuvan, one of the key satellite services offered by ISRO. It provides high-quality real-time updates on road information and satellite imagery.
In addition to that, we also have our own fleet equipped with vehicle tracking systems. Following the unfortunate Nirbhaya case in Delhi, the Indian government made it mandatory for all commercial vehicles to have built-in tracking systems. This allows people to know where their vehicles are at any given time. We are one of the leaders in this space and provide a platform that is now integrated with public transport systems, as well as private buses and taxis. This gives us a valuable stream of data known as traffic probes.
Probes are essentially digital pins that mark the position of a vehicle every few seconds. This data is incredibly valuable. Apart from that, we also collaborate with various partners to collect additional information. Without infringing on privacy, every mobile phone that is turned on, whether iOS or Android, sends location signals. By analyzing the movement patterns of these phones, we can estimate how quickly a device is moving. Using AI, we filter out irrelevant data and retain only the signals that are useful for traffic analysis, such as those from motorcycles or cars.
Our system also differentiates between two-wheelers and four-wheelers. This allows us to understand not just where a vehicle is but what type it is and how fast it is likely moving based on its category. This input is extremely useful when planning delivery routes or predicting travel time.
Another challenge we often deal with, especially in India, is accounting for unusual driving behavior. For instance, some vehicles may take U-turns where they shouldn’t or drive on the wrong side of the road. These irregularities need to be removed from the dataset so the AI can deliver accurate insights.
All of this real-time data is then processed and provided to our partners. We have long-term relationships with several companies, some lasting over six years. Every month, we deliver billions of such data points to our clients. That’s the scale at which we are operating.
TAM: Can you share specific use cases where location intelligence has directly improved operational metrics at scale?
Brajesh Shrivastava: One of our largest clients is a leading e-commerce giant, and we currently support their operations across approximately 12 countries, including India. One of the common challenges they face is attempting to deliver to a location but being unable to reach the exact address, which often leads to the delivery partner calling the customer for assistance.
Now, one of the key metrics that e-commerce companies track is the number of times delivery partners end up calling end customers. This is considered problematic for two reasons. First, it creates a poor customer experience. Second, it raises privacy concerns, as delivery partners gaining access to customers’ phone numbers can lead to potential misuse. There have been instances where such access posed risks to end consumers, and naturally, companies want to avoid such scenarios. Most e-commerce players, especially the large ones, prefer that delivery personnel do not call customers directly.
To address this, the e-commerce giant leverages our solution, particularly a product designed for last-mile delivery challenges, which we call Environ. This product is designed for both gated communities and open areas. It provides highly detailed location data. For example, in a given apartment complex, we offer insights such as the number of units, floor plans, and even elevator mapping.
We can identify if a building has two or three elevators, say A, B, and C, and provide instructions accordingly. For instance, Elevator B might serve only odd-numbered floors, Elevator C the even ones, while Elevator A goes to all floors but may be slower. This level of granularity helps the delivery partner reach the customer quickly and efficiently without needing to make a call.
Thanks to the accuracy and depth of our data, the e-commerce client saw a return on investment (ROI) within just six months. That’s particularly notable because, in many cases, companies expect ROI timelines of five years, which is already considered strong in the industry.
We work on a performance-based model. These clients don’t ask us about the cost of the product alone but they ask about the productivity gains. If we say the solution will deliver a 10x improvement and we charge an amount equal to 1x, they focus on the remaining 9x gain. That’s the value they see. So to sum up, this is a very successful example where ROI was not only achieved quickly but significantly exceeded expectations. It highlights how our intelligent last-mile solutions can transform operations and elevate the end-user experience.
TAM: What role does AI-led data annotation play in enhancing road safety and advancing autonomous mobility solutions in the Indian context?
Brajesh Shrivastava: Autonomous vehicles today are significantly more efficient and deliver much higher levels of precision compared to human drivers. The advancements in autonomous systems and mapping technologies have brought us to a point where vehicles can navigate complex environments more safely and consistently than humans in many scenarios.
A key area enabling this progress is data annotation, and we’re doing extensive work in this space. While we can’t name the clients we work with, these are some of the most cutting-edge companies operating both in India and internationally. We are not just talking about cars. We’re now seeing autonomous trucks capable of driving at highway speeds, close to 100 miles per hour, without any human intervention. These are no longer future concepts. They are real, deployed, and functioning on the ground today. Our role in this is significant, as we support these efforts by providing real-time data annotation services.
Data annotation, in our context, means the real-time extraction and classification of data using AI and machine learning tools. For example, using Python-based automation, we can identify and differentiate between objects such as human figures and static mannequins even when displayed on a billboard. Our systems can recognize which are real, which are static, and what actions are needed accordingly.
This level of annotation is critical for training autonomous systems. We’re not talking about traditional offline annotation where someone manually labels footage. Instead, much of this is being done in real time, enabling immediate feedback loops and decision-making capabilities for autonomous vehicles.
Companies involved in full self-driving (FSD) initiatives are already leveraging this kind of advanced annotation to improve their models. Alongside real-time annotation, we are also heavily involved in building machine learning data libraries, curated, labeled datasets that power AI model training at scale. These data libraries are essential to ensuring the accuracy, safety, and adaptability of autonomous driving technologies across different terrains and conditions.
In short, data annotation is not just a support function anymore. It is foundational to the safe and efficient rollout of autonomous mobility at scale and we’re proud to be contributing at the forefront of this transformation.
TAM: What are the key challenges in scaling AI-powered mapping solutions?
Brajesh Shrivastava: While autonomous vehicles are promising, they come with their own set of challenges. However, when you look at the broader picture, there’s potential for significant improvement in road safety, something that we hope governments will recognize and act upon. Some countries are already moving in that direction.
From a privacy standpoint, we follow strict global guidelines like GDPR, focusing on protecting personally identifiable information (PII). For example, we are currently working on a project in New York City where we are capturing street-level imagery across all five boroughs namely Brooklyn, Bronx, Manhattan, Staten Island, and Queens.
Given how stringent privacy laws are in the US, the cameras we use are equipped with software that automatically blurs faces and number plates even before the images are transmitted. This ensures there is no risk of personal data being shared inadvertently. Privacy is taken very seriously.
In fact, someone once shared a humorous example with me: sometimes the software is so cautious that it ends up blurring the face of an elephant, just because from a certain angle, it might resemble a human face. That is the level of precision and caution being applied.
Of course, there are ethical concerns too. One common example is the “trolley problem” scenario, if an autonomous vehicle must make a choice between two bad outcomes, how should it decide? Regulations may say one thing, but designing software to respond accordingly is a major ethical and technical challenge. For instance, if going off a cliff has a higher chance of passenger survival due to airbags, should the system make that choice? These decisions depend on regulation, and fortunately, I am not the one tasked with making those calls.
Another concern in the industry is the centralization of data. Increasingly, information is being consolidated by a few large tech companies. We work with three of the Fortune 10 companies, and some of our clients are part of the group often criticized for controlling too much of the world’s digital infrastructure.
From the outside, it may appear these companies hold too much power or data, but having worked closely with them, I can say that most have no intention of controlling or blocking others. That said, public perception remains a challenge.
So yes, there are multiple concerns such as privacy, ethics, regulation, and centralization of data. These need to be carefully considered as data capture and innovation continue to evolve.
TAM: With evolving regulatory and infrastructural landscapes, how are AI-geospatial platforms adapting to ensure reliability, accuracy, and real-time responsiveness in mission-critical deployments?
Brajesh Shrivastava: There’s a significant difference in how we build our content compared to what other companies are doing in India and globally. While I won’t name specific companies, many have become overly reliant on AI tools. There’s a growing belief across industries, whether legal, journalism, or mapping, that AI can handle everything. You’ve probably heard claims that AI will replace journalists. In our industry too, there’s a perception that we no longer need people on the ground for verification, which is a dangerous assumption.
There have been unfortunate incidents where this over-reliance on technology has led to serious consequences. This is where we differentiate ourselves. We have a team of about 1,800 people across eight countries, with operations in 37 countries overall. In addition to full-time staff, we also collaborate with a large network of contractors. We are present on the ground in countries like Spain and Malta, verifying key details such as speed limits and address data to ensure accuracy and reliability for our clients.
In India, where infrastructure is growing rapidly, one of our biggest ongoing challenges is collecting and maintaining accurate toll information. Government websites often provide outdated or incomplete data. To address this, our team directly contacts state governments and municipal bodies to confirm toll rates and related information. This allows us to offer the best possible customer experience with real-time, verified data.
On the regulatory side, we work closely with the Ministry of Science and the Department of Science and Technology, which currently oversee India’s geospatial and mapping policies. The introduction of the National Map Policy in 2021 was a major turning point. Before that, we had to work under the Ministry of Defense, which imposed strict restrictions. For instance, we were instructed not to map petrol pumps due to national security concerns, which were valid but often hindered progress in building business-friendly geospatial platforms.
We are now encouraged by the fact that the Department of Science and Technology, with guidance from the Prime Minister’s Office, is actively leading reforms. We are proud to be among the key companies contributing to policy discussions and offering industry insights to help shape the future of geospatial mapping in India.
Alongside our work with regulators, we also support infrastructure organizations such as the National Highway Authority of India by providing updated data on new infrastructure developments. Our goal is to keep the ecosystem well informed while continuing to support policymakers, businesses, and citizens through reliable, accurate geospatial data.
TAM: How does your approach offer a competitive edge when compared to larger players in the geospatial industry?
Brajesh Shrivastava: They are the ones facing what we call the B2C customers, firms that directly interact with end consumers. We, on the other hand, are not client-facing. We operate behind the scenes. Currently, about 95% of the mapping companies around the world use our data. Virtually every major player in the industry relies on us. You name a company, and chances are they are using our data in some way.
We are not focused on customer-facing operations because that is an entirely different business model. Instead, we supply the core data that powers these services. This is true not just in India but in approximately 37 countries globally. It’s a win-win relationship, they grow, and we grow along with them. That’s how our model works.
TAM: Final comments
Brajesh Shrivastava: There’s a lot more we do beyond just working on maps. For instance, one of our important areas of work, which is also featured on our website, involves collaborating with DRDO. Recently, during Operation Sindoor, we were proud to support the Indian government and our armed forces in the background. It was inspiring to see how effectively they responded to the situation, and we were honored to contribute.
Also read: How Operation Sindoor Was Fought With the Aid of Modern Technology
Our work with DRDO involves managing and visualizing large volumes of radar data, ranging from coastal radars to those used by our armed forces. All of this data needs to be accurately displayed on maps, and that requires a tremendous amount of backend work, which our firm is deeply involved in.
So our role isn’t limited to last-mile e-commerce or navigation. We’re also playing a part in strengthening national defense. It’s a matter of great pride for us, not because of the commercial value, but because it was critical for the country, and we are glad our efforts made a difference during a time of need.
