Traditional grading systems in educational institutions are stubbornly inefficient due to manual practices that significantly consume teachers’ time. This inefficiency affects millions of students, as educators struggle to provide personalized attention and meaningful feedback. Smartail, led by CEO Swaminathan Ganesan, tackles this head-on with AI-powered grading and analytics.
Founded by Swaminathan (CEO), Aslam Sherieff J (CGO), and Kannan Ganesan (CTO), Smartail helps schools reduce workload and improve learning outcomes. Their flagship product, DeepGrade, helps schools grade students’ handwritten or digitized answers using advanced AI-powered machine learning. Educators gain data-driven insights to personalize learning based on student needs.
With 70+ schools onboarded and 50,000+ students impacted, Smartail was featured in Forbes India & D Globalist. as one of the Select 200 Companies with Global Business Potential at DGEMS 2024.
In this conversation with ProdWrks, Swaminathan shares his insights from building Smartail.
This interview has been edited for clarity and length
Q1. What inspired you to start Smartail?
Swaminathan: I always had the entrepreneurial bug in me, but I couldn’t start something on my own immediately after college. In 2019, I felt that the timing was finally right for me, and I took the leap
Until then, I had been a solutions architect. I spent years building, testing, and deploying solutions for customers—identifying their problems, creating POCs, and delivering working products. While I wasn’t solving generic market problems, I was solving real challenges at a customer level. This gave me confidence as I had seen the process of how things happen, and I knew that if we identified a problem, we could solve it ourselves.
When we started Smartail, we didn’t want to be just another services company or another product competing for a small market share. We wanted to truly innovate and build something that added core value to the customers it served.
There was inspiration all around as the Indian startup ecosystem was also growing rapidly during that time, and we had already worked on AI projects before the ChatGPT era, experimenting and tasting early success. This convinced us that AI was the future. If you look at Smartail, we are a pre-ChatGPT company, not a post-ChatGPT one. Our AI journey began before the trend exploded, and that early start gave us an edge in shaping the vision and direction of what we wanted to build.
Q2. What are the needs or opportunities in the education systems that led to the development of DeepGrade?
Swaminathan: DeepGrade was born from an unexpected conversation during a biking trip across South India. While visiting a friend’s school, a casual bonfire discussion revealed a major issue—teachers spend nearly half their time on administrative tasks, with 30% of it solely on grading. After all that effort, the only outcome is marks entered into an Excel sheet or report card, offering little real value.
We saw a clear opportunity—what if AI handled grading? Automating this process would free up teachers’ time for teaching and student development while also delivering insights beyond just marks. Instead of relying on perception, DeepGrade identifies learning gaps, helping teachers refine teaching methods and focus areas with clear action points.
Unlike most ed-tech tools that focus on students, we focus on teachers. Enabling one teacher to effectively impact thousands of students is the crux of DeepGrade. Time is irreplaceable, and DeepGrade ensures that the time saved is used to bridge learning gaps, not just reduce workload.
Q3. Who are the primary users and buyers of Smartail’s products? What core problems do your products solve for them and how?
Swaminathan: Our flagship product,DeepGrade, uses AI to grade student answers and provide detailed analytics. Schools conduct exams, students write and scan their papers, and DeepGrade automates everything—from identifying answer sheets to grading and generating insights.
DeepGrade’s key features include handwriting extraction, which allows AI to read and process student handwriting; automated grading, which aligns with school-defined rubrics; and analytics, which provides insights to help teachers understand student performance and learning gaps.
The primary buyers and users are school owners, principals, and educators with a strong focus on the K-12 sector. However, we have also successfully piloted DeepGrade for grading MBBS answer sheets at universities, showing promising results. While higher education is a future expansion area, K-12 remains our core market.
From a market standpoint, DeepGrade is syllabus-agnostic, allowing it to operate across India, the UK, Singapore, and the Middle East (GCC). We are actively exploring Australia and the US, with our core product already built for global scalability. While some market-specific adaptations may be needed, the foundation is ready. While we see the whole world as our market, our immediate focus is on GCC, Singapore, India, the UK, and the US.
Q4. What were the key insights from user research that shaped the product?
Swaminathan:In 2019, we approached schools asking for old answer papers to analyze. Some generously shared papers up to three years’ worth of data. Our research focused on two key areas: how students approach exams and how teachers evaluate answers.
We quickly realized that grading isn’t just about correctness but also involves human perception. The same answer could get different marks from different teachers based on interpretation. To automate grading effectively, we had to mimic human decision-making rather than apply rigid rules. We had teachers manually grade papers while our AI did the same, then compared the results and refined the system until it aligned with human evaluation patterns.
Flexibility was crucial. Teachers needed to be able to override AI-generated marks with a single click, instantly access reports, and work with a simple, intuitive interface. DeepGrade even replicates real classroom interactions, allowing students to challenge their marks and teachers to review and adjust scores within the system, just like they would in person.
What makes us unique is that we are not an extra coaching tool or a supplementary learning app. We are embedded into the school’s daily workflow, integrating seamlessly rather than disrupting teaching. Schools don’t have to change how they operate—DeepGrade enhances efficiency while fitting naturally into the education system.
Q5. Who are your key customers and what’s your customer acquisition strategy?
Swaminathan:Today, we work with over 70 schools and nearly 50,000 students, with our key customers being large school chains and independent institutions with significant student populations.
In the early days, our strategy focused on education forums, CBSE Sahodaya groups, and private networks where school principals and owners gathered. Since there was nothing to compare our product to in the market, we needed early adopters willing to test it. At that stage, we were still refining our product-market fit and couldn’t rely on heavy spending, so we prioritized targeted outreach and direct engagement. Many schools saw adopting DeepGrade as a risk, but we found visionaries who believed in its potential.
Now, the dynamics have shifted, and schools are actively seeking us out. We are receiving direct demo requests through our website, reducing the need for outbound efforts. Alongside this, we continue our traditional outreach while expanding into three key areas—channel partnerships, digital discovery through inbound marketing, and integrations with LMS and ERP providers as their grading partners. We are also open to working with the government sector, offering large-scale diagnostic assessments for students and detailed reports for decision-makers.
Now, scaling from 100 to 300 schools requires a different approach. Unlike SaaS tools with a quick sign-up model, schools need to pilot and test DeepGrade before committing, making the sales cycle longer. However, once they adopt it, they stay for the long term.
While our traditional methods have worked well, the market has now become more receptive, and we are seeing increasing demand globally. Our strategy remains a mix of targeted outreach and digital visibility, ensuring a steady pipeline of schools eager to adopt AI-driven grading.
Q6. How did you validate your business idea and achieve product-market fit for DeepGrade?
Swaminathan: We launched DeepGrade in 2021, shortly after schools reopened post-COVID. The real test began in 2022–2023 when schools started using it daily. Initially, teachers were skeptical—questioning AI grading, how to review and adjust marks, and whether it would fit into their workflow. The product evolved entirely through continuous feedback as we worked closely with schools to refine it based on their needs.
Early doubts revolved around logging in, entering questions, and correcting AI-graded answers. We addressed these concerns step by step, making the system more intuitive. Over time, as teachers became comfortable, usage increased, and by 2024–2025, 95% of our customers renewed, with many already committed for 2025–2026. After three consecutive years of adoption, DeepGrade is no longer an experiment, it has achieved a strong product-market fit.
Q7. What did you learn from user behavior that was pivotal to shaping your product?
Swaminathan:One key learning moment was when a school discovered an incorrect answer key configuration after grading. Initially, the only solution was to redo the entire process, which was inefficient. Based on teacher feedback, we introduced a one-click re-correction feature, allowing them to edit the answer key and instantly update scores across all students. This and other iterative improvements have ensured that nearly all core features are actively used daily by teachers.
DeepGrade remains the core product, with DeepGrade Analytics providing insights and DeepPrep focusing on board exam preparation for class 10 students. Schools can choose grading, analytics, or both, depending on their needs. A common validation test when entering a school involves grading student papers with both DeepGrade and teachers, then comparing results—this hands-on validation has been key in driving adoption.
Q8. How did you make the platform easy to use and engaging?
Swaminathan: Since we work directly with schools, engagement goes beyond the platform. Our support team, subject matter experts, and account managers maintain daily communication with school coordinators and management, understanding their schedules, assessments, and expectations. While the platform automates key processes, this human element ensures that schools are using the product effectively. If a school isn’t actively engaging with DeepGrade, we identify the issue early and step in to assist.
Q9. What steps do you take to reduce churn and improve customer retention?
Swaminathan: Customer retention largely depends on accurate grading and transparent reporting. Schools trust DeepGrade when they understand why and how AI grading works, so we focus on clear explanations and support. If these two factors are handled well, churn is minimal. However, there are cases where resistance to AI itself leads to churn, which we mitigate through ongoing engagement and education.
To keep schools actively involved, we provide monthly and quarterly reports on student performance and conduct review meetings with school management. These aren’t just reports for the sake of data; they offer actionable insights that help teachers and administrators make informed decisions to improve learning outcomes. By ensuring our platform is both intuitive and supported by real human interaction, we keep schools engaged and reduce churn effectively.
Q10. What were your biggest product-related challenges and key pivots in the product journey?
Swaminathan: When we started in November 2019, our vision was clear—we would be a school-facing platform without an app for students. We didn’t want to add another app to their daily routine. However, COVID-19 changed everything. By early 2020, schools—our potential customers—shut down, and we couldn’t engage with them as planned.
Fortunately, we had already signed MOUs with a few schools and were in discussions with them. During the pandemic, these schools reached out, asking if we could help them send videos, documents, and online learning materials to students. While this wasn’t part of our original plan, we saw an opportunity to support schools at a critical time. We pivoted and built features to facilitate remote learning, keeping schools engaged with our platform.
This shift created long-term customer loyalty. Schools appreciated our support during COVID-19, and once normalcy returned, they reciprocated by staying with us, providing feedback, and assigning teachers to help improve the product. Even today, some schools continue to use these additional features, such as uploading learning materials, scheduling online classes, and sharing documents—capabilities we hadn’t initially intended to build.
While we have since refocused on our core vision, these additions helped us navigate a challenging period and ultimately strengthened our product, making it even more valuable to schools.
Q11. Please explain your revenue model and monetization strategies.
Swaminathan: DeepGrade went through multiple monetization models before we arrived at the one that works best today. Our pricing is based on the number of exams, number of tests, and total marks students take in a year.
For example, if a student writes two formative exams and two summative exams in a year across four subjects—English, Math, Science, and Social Science—where formative exams are 40 marks each and summative exams are 80 marks each, the total marks in a year would be 960.
Alternatively, a school may choose to conduct monthly 20-mark tests for four subjects, amounting to 80 marks per month over 10 months, totaling 800 marks per year. Regardless of how a school structures its assessments, the core idea remains the same—pricing is based on the total number of marks students are graded for.
For instance, if a student takes 1000 marks worth of exams, we charge ₹2000. Schools can distribute these marks across exams however they prefer—three 80-mark exams, two 40-mark exams, two 20-mark exams, or twenty 20-mark exams—the pricing remains based on total marks graded.
Initially, schools found this pricing model unconventional, but they quickly realized its fairness. Since schools don’t conduct tests daily and assessment patterns vary—a 7th-grade student takes fewer exams than a 10th or 12th grader, and grading complexity increases with higher classes—the model accounts for these differences. By structuring our pricing in terms schools understand—formative and summative assessments—it becomes clearer and more logical for them.
Today, we generate over ₹2.5 crores in annual revenue, and this model has proven scalable, sustainable, and aligned with how schools operate.
Q12. What is your North Star metric? Please provide details on key metrics you track to measure product impact.
Swaminathan: Our North Star metric is tracking student learning progress over time. We don’t just measure raw scores; we analyze whether a student is improving across different assessments. For example, if a student struggles with English grammar in May, we track their progress in September to see if their understanding has improved. Schools receive insights identifying which students have progressed and which need more attention.
Beyond scores, DeepGrade analyzes how students approach exams. Traditionally, a student scoring 70 out of 80 is seen as a top performer, while one scoring 30 out of 80 is considered weak. However, our system goes deeper—examining the types of questions attempted. A student who scores low might have excelled in complex analytical questions while struggling with memory-based ones. This insight helps teachers recognize a student’s true potential, rather than just labeling them based on overall marks.
A simple analogy would be comparing a street cricketer scoring a double century to a professional player scoring 25 in an international match—context matters. Similarly, a student attempting tougher questions and scoring lower might be showcasing deeper understanding than one who aces only basic recall questions. By shifting the focus from marks to learning depth, DeepGrade provides actionable insights that empower both teachers and students.
Q13. Are you tracking any emerging technologies or market shifts that could be a game-changer for you and other edtech companies?
Swaminathan:The last few years have seen a major shift in large language models (LLMs) entering the education space. We were not originally built as an LLM-based company, but we recognize their potential and are actively exploring ways to integrate them where they can add real value.
DeepGrade has already graded nearly 80 million student responses, covering a wide range of schools from tier-1 to tier-4 cities. This massive dataset gives us a unique advantage in improving AI-driven grading. Unlike basic automation tasks, grading is an intellectual process, requiring deep contextual understanding. It is more like peeling an onion, uncovering complexity at every layer, rather than just performing a simple task.
While we continue to strengthen our own proprietary AI models, we are also experimenting with LLM and small language model (SLM) integrations to enhance our capabilities. The goal remains the same—delivering accurate, fair, and actionable assessments to improve learning outcomes at scale.
