An interview with Haarishkumar Bhaskar, AI Product Developer and technology entrepreneur, on AI product engineering, mobility intelligence and the shift from reactive systems to predictive decision-making.
Artificial intelligence is increasingly being incorporated into mobility platforms that depend on real-time data, operational coordination and timely decision-making. As transportation systems generate more information about users, drivers, routes, timing and behaviour, the challenge is not simply collecting that data but interpreting it in ways that can support safer and more efficient operations.
Haarishkumar Bhaskar brings experience from both AI product development and mobility technology. Through his work at Raido, a company developing digital transportation solutions, including its Raido School platform, he has developed a practical perspective on how AI can be integrated into products used in real-world transportation environments. Raido School publicly provides features including real-time ride tracking, driver information, trip history and notifications for school transportation. (Google Play)
The discussion explores how mobility data can be used to identify emerging risks, support operational teams and improve decision-making. It also considers the product engineering required to move AI beyond an experimental feature and into a capability that can be understood, monitored and used responsibly.
We spoke with Haarishkumar Bhaskar about how he approaches AI product development, why risk intelligence is different from conventional analytics, and what it takes to move AI from an experimental feature into a useful product capability.
Q: Your work sits at the intersection of AI and mobility. What attracted you to this area?
Haarishkumar Bhaskar:
What interests me about mobility is that it is a real-world environment. The product isn’t operating only inside a screen. Every ride generates signals about users, drivers, routes, timing, behaviour and operational events.
That creates an opportunity for AI to do more than simply automate a task. It can help identify patterns that aren’t immediately obvious to an operations team and turn those patterns into actionable intelligence.
This is also why the area is relevant beyond transportation. Mobility is a useful example of how AI can be applied in environments where decisions affect physical operations, customer experience and safety. My interest has therefore been in building AI capabilities that solve actual product and operational problems rather than adding AI simply because it is a current technology trend.
Q: What does AI product development mean to you?
Haarishkumar Bhaskar:
I think there is an important difference between building an AI model and building an AI product.
A model can produce a prediction, but a product has to answer a much bigger question: What happens because of that prediction?
When I approach an AI feature, I think about the complete flow—what data is available, what signals are meaningful, how the system should process them, how the output should be presented, and what action a product or operations team can take from it.
That means product engineering, data thinking and user experience have to work together. This distinction is important because many AI projects fail to create practical value when the model is developed separately from the workflow in which its output will be used.
Q: One of the areas you’ve worked on is AI-based fraud and behavioural risk detection. What problem were you trying to solve?
Haarishkumar Bhaskar:
Mobility platforms generate large amounts of behavioural and transactional data. Some unusual patterns may be harmless, while others can indicate potentially suspicious behaviour.
The challenge is not simply finding something that looks unusual. The challenge is creating a system that can distinguish meaningful patterns from normal variation.
My approach has been to think of risk as a combination of signals rather than a single event. User behaviour, driver activity, ride patterns and transaction-related signals can be considered together to produce a more useful risk picture.
The objective is to help the platform identify cases that deserve attention instead of forcing an operations team to manually examine every event. This is relevant because the volume and speed of mobility data can make purely manual review inefficient, while an automated system can help prioritise human attention.
Q: How is that different from a traditional dashboard?
Haarishkumar Bhaskar:
A traditional dashboard mainly tells you what has already happened.
AI can potentially help answer a different question: What deserves attention next? That is where I see the value of risk intelligence.
Instead of presenting hundreds of disconnected metrics, an intelligence layer can identify anomalies, behavioural patterns or risk indicators and bring the most relevant information forward.
The final decision should still remain with the appropriate human or operations team. AI should improve the quality and speed of decision-making rather than remove accountability.
This distinction is important for mobility because operational teams often need to respond to changing circumstances quickly. A system that only reports historical information may be useful for analysis, but a system that helps prioritise emerging issues can have greater practical value.
Q: Safety is particularly important in school transportation. How do you think about AI in that environment?
Haarishkumar Bhaskar:
Safety has to be approached differently because the consequences of a poor decision can be significant.
For school transportation, I think the technology should focus on visibility, early identification of unusual situations and better operational awareness.
The foundation already exists in products such as Raido School through capabilities such as live tracking, driver information, trip history and notifications. (Google Play)
The next layer is intelligence: understanding patterns across those signals and identifying situations that may require attention.
This is relevant because school transportation involves multiple stakeholders, including schools, parents, drivers and operations teams. Each group needs timely and understandable information, and AI can potentially help connect those sources of information without replacing the people responsible for making decisions.
For me, the goal isn’t to claim that AI can guarantee safety. It is to use data and predictive techniques to give people better information earlier.
Q: What are some of the biggest challenges when building AI for mobility?
Haarishkumar Bhaskar:
Data quality is one of the biggest challenges.
Real-world systems rarely produce perfectly structured datasets. You have missing information, inconsistent behaviour, changing environments and new patterns that weren’t present when a system was initially designed.
Another challenge is explainability. If an AI system produces a risk indicator, the product team needs to understand what contributed to that result. Otherwise, it becomes difficult to trust and act on the output.
I therefore believe AI products need to be designed with monitoring, explainability and human oversight from the beginning rather than added later.
These challenges are particularly relevant in mobility because the data reflects changing real-world conditions. A system that performs well in one environment may need to be monitored and adapted as routes, users, behaviours and operational processes change.
Q: Do you think AI will eventually replace operational decision-making in mobility?
Haarishkumar Bhaskar:
I don’t see it that way.
I think AI will increasingly become an intelligence layer around human decision-making.
The system can process much more information than a person can manually review. It can surface patterns, prioritise risks and recommend possible actions. But decisions involving safety, customers and operational consequences still need appropriate human oversight.
The strongest AI products, in my view, are not necessarily the ones that automate everything. They are the ones that help people make better decisions.
That perspective is one reason this topic matters. The future of AI in mobility is unlikely to depend only on model performance; it will also depend on whether organisations can integrate AI responsibly into existing operational processes.
Q: You also work on AI products outside mobility. Has that changed the way you approach product development?
Haarishkumar Bhaskar:
Definitely.
Working across different AI products has made me think less about individual technologies and more about the underlying product problem.
Generative AI, predictive models, AI agents and automation can all be useful, but the technology should follow the problem.
I usually start with four questions:
What problem are we solving? What data or signals are available? What decision or action should the AI improve? And how will we know whether the feature is actually useful?
That framework helps prevent AI from becoming a superficial layer added to an existing product. It also makes it easier to evaluate whether a system is producing measurable value for users, operators or the organisation.
Q: Where do you see the next stage of AI-powered mobility going?
Haarishkumar Bhaskar:
I think we’re moving from connected mobility to intelligent mobility.
The first stage was digitising transportation—booking, tracking, payments and communication. The next stage is making those systems understand what is happening.
Eventually, mobility platforms will increasingly combine behavioural intelligence, predictive risk analysis, operational optimisation and AI-assisted decision-making.
The important part is that these systems should be designed around real-world outcomes. AI has enormous potential in mobility, but its value will ultimately depend on whether it can make transportation safer, more efficient and more responsive to the people using it.
That is why conversations about AI in mobility should focus not only on technical capability, but also on implementation, accountability and the practical benefits delivered to users and operational teams.
About Haarishkumar Bhaskar
Haarishkumar Bhaskar is an AI Product Developer and technology entrepreneur working across artificial intelligence, Generative AI, AI agents and digital product development. His work focuses on translating emerging AI technologies into practical product capabilities, with particular
interest in mobility intelligence, behavioural risk analysis and intelligent automation. His experience at the intersection of AI engineering and product development provides the basis for this interview’s focus on how emerging technologies can be applied to real-world mobility challenges.
This interview was published as an editorial discussion of Haarishkumar Bhaskar’s professional perspective and experience in AI product development. It should not be interpreted as a claim that any particular AI system guarantees safety, fraud prevention or operational outcomes.
