Every few months, young Indian AI startup founders quietly board a flight to San Francisco. They raise money, sign a few contracts, soak in the energy of a market that pays fast, and then come back to India. The pattern is no longer anecdotal. It is a cycle.
To build AI for India, founders still feel they must first succeed abroad.
“Sadly, validation from outside has always led to acceptance inside [India],” said Apoorv Sood, global GTM head of smallest.ai, in a conversation with AIM.
The enterprise behaviour has not kept pace with India’s talent landscape. He wishes it would change.
smallest.ai was started by Sudarshan Kamath and Akshat Mandloi in Pune. The founders moved to Bengaluru for talent density and eventually moved to the US for capital and customers.
Sood said building entirely in India slows companies down.
“If you have to build a business in the US, you have to be in the US,” he said. “Access to the capital ecosystem is much more advanced and 5-10 times faster.” The company raised $8 million in its seed round in October, led by Sierra Ventures, with participation from several others.
Many of these startups are not leaving India. They are coming back with tech shaped in harder markets. Sood said that after years of superficial AI experimentation, India has begun to see a real shift. “My mother uses ChatGPT,” he said, laughing at how fast consumer comfort has grown.
Enterprise adoption, however, remains slower. Sood highlighted that labour costs in India are so low that employers prefer to hire workers rather than invest in advanced technology.
Building Global Tech from the West
Smallest.ai’s core technology is built for India’s conversational chaos. For instance, he points out that many people say “Infy” instead of “Infosys”.
The company is betting on speech technology built for accents, mumbles and informal speech patterns. Sood believes real Swadeshi tech comes from global excellence.
“The real Swadeshi Tech is somebody who’s building for the globe, keeping your own country in mind,” he said.
Amartya Jha, co-founder and CEO of CodeAnt, reached a similar conclusion but took a different approach. He and his co-founder permanently moved to the US, got backing from Y Combinator and raised $2 million in seed funding.
While engineering strength sits in India, sales strength sits in the US. “Our primary customer is the US, not India,” he said. Jha’s answer to the Swadeshi debate almost mirrors Sood’s.
CodeAnt, an AI-powered debugging and reviewing platform, competes with established large players like SonarQube and serves customers in both the Western and Indian markets.
For Jha, the problem is not the talent. It is the mindset. “Most Indians are risk-averse,” he said. As a result, the funding loops stall. Domestic investors often want to see a US firm write the first cheque before conviction builds.
Jha said he also noticed a behavioural shift once he relocated. “When we see that this guy is there outside India building something big, having better customers, we value our time better,” he said. When he took the same calls from India, decisions were delayed. After he moved to the US, the same clients closed in weeks.
He does not endorse the behaviour, but states the reality.
“Stop playing games,” he said. “If you believe the market is good, invest.” He argued that investors behave like a late-stage fund even at the seed stage, demanding excess proof before backing a company, defeating the whole purpose.
There are Founders Who Stayed
Ashutosh Singh, co-founder and CEO of RevRag, summarised the issue more bluntly. He said investors often want traction before imagination, a mismatch for AI startups that require compute burn and long-term conviction.
“Foundational and platform bets require patience,” he said, pointing out that even companies like OpenAI and Anthropic continue to burn capital.
Unlike Sood and Jha, Singh is doubling down on India for the next two years, betting enterprise maturity will follow. “The capital is available while conviction has to catch up,” he said.
Sood also pointed to a cultural layer. Global validation still commands respect among Indian enterprises. Sales cycles shorten not because the product changes, but because perception does.
The sales cycle tightens because the founder is no longer seen as a local vendor but a global contender.
Despite different paths, all three founders agreed on one thing. India should not fight global competition. It should match it. “Build the best product and let people use it,” Sood said. “Fair competition, fair market, confidence.”
Jha echoed similar sentiments. He hires young engineers under 25 in India and pays them at a global benchmark. “That’s what world-class building actually requires,” he said.
So why do founders still leave?
Because the confidence loop has not caught up with the talent loop. Because enterprises still chase free proof of concepts. Because investors want someone else to go first. Because early buyers look westward before trusting locally.
Yet something is shifting. Indian consumers have embraced AI faster than expected. Enterprises are beginning to deploy AI solutions at scale. Engineering depth remains unmatched. The market is warming, slowly but visibly.
The post Why Indian AI Startups Still Seek Validation from the West appeared first on Analytics India Magazine.
OpenAI has claimed that it built and shipped the Sora Android app in just 28 days, relying heavily on its AI coding agent, Codex.
The company said the initial production version of Sora for Android was developed between October 8 and November 5, 2025, by a four-engineer team working alongside Codex and consuming “roughly 5 billion tokens.”
The app launched publicly in November and reached number one on the Google Play Store on its first day, with Android users generating “more than a million videos in the first 24 hours.”
OpenAI engineers Patrick Hum and RJ Marsan wrote that the team deliberately avoided adding headcount under tight deadlines, citing the famous observation by American software engineer Fred Brooks that “adding more people to a late software project makes it later.”
Instead, each engineer worked with Codex to multiply output. “We assembled a strong team of four engineers – all equipped with Codex to drastically increase each engineer’s impact,” they said.
According to OpenAI, Codex handled an estimated 85% of the codebase, using an early version of the GPT-5.1-Codex model, which the company said is now available to developers via its CLI, IDE extension, and web app.
Despite the compressed timeline, OpenAI claims the app has a “99.9 per cent” crash-free rate.
The engineering team described treating Codex like “a newly hired senior engineer,” focusing human effort on architecture, system design, and user experience rather than implementation.
“We leaned on Codex to do a huge amount of heavy lifting inside well-understood patterns and well-bounded scopes, while our team focused on architecture, user experience, systemic changes, and final quality,” the authors wrote.
OpenAI said Codex excelled at reading large codebases, translating logic across platforms, and generating broad test coverage. “Codex is (uniquely) enthusiastic about writing unit tests,” the blog noted, adding that engineers frequently pasted CI logs into prompts to diagnose failures.
However, the company acknowledged limitations. Codex “isn’t yet great at inferring what it hasn’t been told,” and struggled with “deep architectural judgment” when left unguided.
To address this, the team invested heavily in documentation, such as AGENTS.md, to enforce patterns, coding standards, and tooling requirements.
One notable technique was to use Codex as a cross-platform translation layer rather than as a shared framework. “Forget React Native or Flutter; the future of cross-platform is just Codex,” the engineers wrote, explaining that Codex translated Swift logic from the iOS app into Kotlin while preserving behaviour.
As development accelerated, the bottleneck shifted from writing code to reviewing and coordinating parallel Codex sessions. “Our bottleneck in development shifted from writing code to making decisions, giving feedback, and integrating changes,” OpenAI said.
In the company’s State of Enterprise AI 2025 report, released a few days ago, the company stated that over the last six weeks, there was a 2x increase in weekly active Codex users. Further, the company observed a ~50% increase in Codex messages over the same period.
In October, Sam Altman, the CEO of the company, revealed that “Almost all new code written at OpenAI today is from Codex users.” He added that engineers in OpenAI complete 70% more pull requests (PRs) each week using Codex.
The post In Just 28 Days, OpenAI Built Sora’s Android App Using Codex appeared first on Analytics India Magazine.
Broadcom disclosed during its Q4 2025 earnings call that it received a $10 billion order in the previous quarter to supply Google’s latest Tensor Processing Units (TPUs) to Anthropic.
“In Q4, we received an additional $11 billion order from this same customer for delivery in late 2026,” said Hock Tan, CEO of Broadcom. This brings Anthropic’s total TPU orders to $21 billion.
Furthermore, the company revealed a $73 billion backlog of AI product orders, which are expected to be shipped over the next six quarters (18 months).
TPUs are specialised accelerators developed by Google for AI workloads. Now in their seventh generation, TPUs are available to customers through Google Cloud and power many of Google’s internal systems, including training and deployment of the Gemini family of models. Google designs the TPU architecture, while Broadcom converts those designs into manufacturable silicon and handles volume production. The relationship mirrors Google’s long-standing strategy of controlling key AI hardware design while relying on semiconductor partners for fabrication expertise.
Anthropic, a long-term user of TPUs, recently announced plans to significantly scale its infrastructure. The company intends to deploy one million TPUs, backed by more than one gigawatt of new compute capacity coming online in 2026. This represents one of the largest dedicated AI compute buildouts in the industry.
Several other companies have also confirmed their use of TPUs, including Meta, Cohere, Apple and Ilya Sutskever’s new startup, Super Safe Intelligence (SSI).
A report from The Information indicates that Meta is evaluating the deployment of TPUs in its data centres starting in 2027.
The growing adoption of TPUs stems from their power efficiency and tight optimisation for AI training and inference, creating increasing competitive pressure on NVIDIA’s GPU dominance.
Broadcom said it now has five TPU/XPU (custom AI accelerator) customers—with Google and Anthropic named on the call. Reports and industry analysis indicate that Meta and ByteDance are also among its custom AI chip customers, though Broadcom has not publicly confirmed the full roster.
The rise of TPUs over the years, thanks to their power efficiency and being fine-tuned to specifically handle AI workloads, poses a challenge to NVIDIA’s dominance with GPUs.
According to new analysis from SemiAnalysis, TPU v7 demonstrates that although it has roughly 10% lower peak floating-point operations per second (FLOPs) and memory bandwidth than NVIDIA’s GB200 platform, it still delivers a stronger performance-per-total-cost-of-ownership (TCO) profile.
SemiAnalysis estimates that Google’s internal cost to deploy Ironwood is about 44% lower than deploying an equivalent NVIDIA system.
Even when priced for external customers, TPUv7 offers an estimated 30% lower TCO than NVIDIA’s GB200, and roughly 41% lower TCO than the upcoming GB300.
SemiAnalysis notes that if Anthropic achieves around 40% machine-fraction utilisation (MFU) on TPUs—a realistic figure given the company’s compiler and systems expertise—the effective training cost per FLOP could be 50-60% lower than what GB300-class GPU clusters are expected to deliver.
The post Broadcom Reveals $21 Billion Google TPUs Order from Anthropic appeared first on Analytics India Magazine.
United States President Donald Trump, on December 11, signed an executive order aimed at curbing state-level AI regulations and accelerating the creation of a unified federal framework.
The order, titled Ensuring a National Policy Framework for Artificial Intelligence, represents the administration’s most decisive move yet to preempt what it calls a “patchwork” of state rules that threaten US dominance in AI development.
In the order, Trump argues that the US is “in a race with adversaries for supremacy” in AI and that companies must be able to innovate “without cumbersome regulation”.
He directly criticises emerging state laws that seek to govern algorithmic discrimination, transparency or model outputs.
Citing Colorado’s recently enacted rules, the president claims such laws “may even force AI models to produce false results in order to avoid a ‘differential treatment or impact’ on protected groups”.
The executive order establishes a new AI Litigation Task Force inside the justice department, directing it to challenge state laws the administration views as unconstitutional or obstructive.
It also instructs the commerce department to publish, within 90 days, an evaluation identifying state AI laws that are “onerous” or inconsistent with federal policy.
Trump’s directive further calls on the Federal Communications Commission (FCC) and the Federal Trade Commission (FTC) to explore federal reporting and disclosure standards that could explicitly preempt conflicting state requirements.
“It is the policy of the United States to sustain and enhance the United States’ global AI dominance through a minimally burdensome national policy framework,” the order states.
The administration will also send Congress a legislative proposal establishing a federal artificial intelligence framework that overrides state regulations, except in areas like child safety and government procurement.
“We remain in the earliest days of this technological revolution,” Trump wrote, adding, “It is imperative that we act now to ensure that America wins the AI race.”
The order comes amid escalating friction between state governments and Washington over how AI should be governed.
In recent months, scrutiny from the states has intensified as 42 attorneys general, including those from Colorado, Florida, Massachusetts, Texas, Virginia, Washington and Illinois, warned that generative AI systems may already be violating consumer-protection and child-safety laws.
The bipartisan group has also demanded independent audits from Microsoft, Google, Meta and Apple, arguing that developers have not done enough to curb harmful or misleading outputs.
State legislatures have also moved aggressively to craft their own rules, creating the patchwork the White House says it is now trying to dismantle.
California passed its Transparency in Frontier Artificial Intelligence Act, which requires developers of large-scale AI systems to publish risk assessments and safety documentation.
Texas adopted a different approach, enacting criminal penalties for the possession or promotion of AI-generated obscene material involving minors.
The post Trump Cracks Down on State AI Regulations, Launches National Policy Push appeared first on Analytics India Magazine.
Indian companies have spent the last two years fielding bold claims about what AI can deliver. Every conference, pitch deck and internal strategy memo carries the same warning: AI adoption is no longer optional.
This urgency, however, has not always been matched by clarity. Many business owners and managers hear about ambitious use cases, yet struggle to see a clear path to implementation. As a result, they often fail to achieve the desired return on investment (ROI).
They also carry the belief that meaningful AI requires huge budgets and niche technical teams. This leaves SMEs unsure about where to invest and what returns to expect realistically.
A shift is underway. Instead of large standalone AI projects that demand parallel tech teams, businesses are turning to embedded capabilities within the tools they already use. Platforms such as Bitrix24 now offer summarisation, intelligent lead scoring and contextual recommendations directly inside existing workflows.
Integrated platforms are folding these features into their core systems so organisations can capture value without tearing their architecture apart. This marks a practical distinction. AI becomes useful when it augments processes rather than forcing companies to build a new operational layer dedicated to experimentation.
The Data Quality Gap
Every meaningful AI conversation eventually comes back to one unglamorous truth: models are only as reliable as the data they work with. Many firms believe they have data simply because they maintain CRM entries, spreadsheets or contact forms.
In practice, this information is often inconsistent, duplicated or incomplete. AI systems recognise patterns. When names vary, contacts repeat or histories are missing, the signals become unreliable. Even something a person would ignore, like a stray bracket or mismatched quote from a sloppy data transfer, can derail an automated process entirely.
For resource-constrained companies, the first step towards AI readiness is not a new model. It is an operational discipline. Standardised capture, consistent tagging, proper validation and regular hygiene create the foundation on which automation can function.
Modern CRM systems make this easier by enforcing structure. Bitrix24, for instance, offers duplicate detection, merge workflows, required-field rules and automation that fills or standardises fields during capture. Its activity log and telephony integrations consolidate calls, emails and chats into a single customer record so the system holds the full context.
With these basics in place, even modest automation produces real improvements in response times, conversion rates and error reduction. Data governance is not a technical footnote. It is the primary enabler of any scalable AI effort.
Fragmented Tools, Fragmented Value
Another barrier lies in the way many companies assemble their tech stack. Marketing teams use a content generator. Sales installs an analytics plugin. Customer support deploys a chatbot. None of these tools talk to each other. They create pockets of automation rather than an intelligent customer journey.
This fragmentation weakens ROI. A lead generated through marketing may not carry behavioural context into sales. A support ticket may not inform future product or marketing decisions.
For Indian businesses mindful of cost, adopting one specialised tool after another raises expenses and reduces the chance of system-wide improvement.
Integrated platforms solve this by merging communication, CRM, tasks and automation into one environment. When the entire workflow shares a common data model, AI can operate with continuity. Lead scoring can incorporate marketing signals.
Customer success tools update themselves after support conversations. Sales playbooks adjust based on outcome analytics.
The lesson is straightforward. AI delivers systemic improvement only when it operates on unified data inside connected workflows.
Anxiety About Automation
Technical barriers are only part of the challenge. Many employees fear that automation will replace them. This fear is not abstract. In Indian workplaces, concerns about job security are tangible. Teams often resist adopting tools they believe could make their roles redundant.
This leads companies to pay for AI features but never entirely use them.
Yet, practical deployments show that the most valuable AI does not replace judgement or relationship building. It removes repetitive chores—automating data entry, producing first-pass summaries of customer interactions or routing requests simply frees people to focus on work that requires nuance and experience.
Framing AI as support rather than displacement is essential. Teams adopt tools more readily when they see workload reduction in mundane tasks while retaining control over decisions.
Measuring ROI: Metrics that Matter
AI often enters organisations as an aspiration. Without clear metrics, it becomes a cost instead of an engine of efficiency. A metrics-first approach gives clarity. ROI should tie directly to operational change rather than abstract claims.
The most practical indicators focus on time saved, conversion lift, accuracy improvement, faster response times and revenue impact from AI-assisted engagements. Leading platforms now display these metrics on their dashboards. They translate automation into numbers that the business can act on.
Companies should define KPIs before deployment, run controlled pilots and compare post-implementation outcomes against baselines. If a gain cannot be measured, it cannot be managed.
A Pragmatic Roadmap to Adoption
Indian SMEs that want to move beyond hype benefit from a staged approach. High-leverage processes come first. These are repeatable tasks where automation cuts time to value, such as meeting summaries, lead qualification or ticket routing.
The next step is to get data in order with consistent fields, deduplication and tagging standards within the CRM.
After this comes tool selection. Integrated solutions that unify CRM, communication and workflow automation allow AI features to operate across the entire process.
A pilot with defined KPIs and the right internal champions sets the tone for scalable deployment.
Communication remains essential throughout. Positioning AI as augmentation builds trust and signals that the goal is human efficiency, not human replacement.
Practical adoption is a sequence of small improvements that compound over time. Features like automated lead scoring, contextual suggestions and instant summarisation are low-risk and high-reward. They deliver faster wins than grand AI overhauls.
The most effective AI is almost invisible. It blends naturally into workflows, improves reliability and shows its value through measurable change rather than loud announcements. It depends on disciplined data practices, unified systems and clear metrics. It succeeds when employees feel supported rather than sidelined.
Integrated platforms that weave AI into everyday tasks prove that companies do not need large specialist teams or inflated budgets to benefit. The businesses that succeed with AI will not be the ones chasing spectacle. They will be the ones stacking small, consistent improvements until they add up to real operational strength.
The post Beyond the AI Buzz: How Indian Businesses Can Use Realistic, Affordable AI Tools That Actually Work appeared first on Analytics India Magazine.
The duality of Indian cricket is hard to ignore. While money and opportunities flow under the high-mast floodlights of stadiums that host international ODIs and cash-rich cricket leagues, the real sport is played in the dusty district grounds and makeshift pitches, where players hone their talent with sweat and blood.
However, getting discovered in one of those cricket nurseries still factors more on chance than skill. A good performance might matter, but only if the right selector is watching.
Satyendra Kumar found himself on the wrong side of that uneven system. “I played eight years of cricket,” the 19-year-old said. “I was in the under-16 state team probables… but I didn’t get enough chances even being a left-arm fast bowler.”
Left-arm fast bowlers are rare in Indian cricket, but that didn’t guarantee opportunity. “There is no benchmark, no structure or no transparency in selections,” he recalled.
He kept pushing. He earned spots at special IPL camps with Royal Challengers Bengaluru (RCB), Delhi Capitals, and Lucknow Super Giants. But as his cricket career edged forward, life threw curveballs.
This experience inspired Kumar to create opportunities for others facing similar challenges, believing that if the system couldn’t be fair to him, it was time to build something new. In September this year, the teenager launched ScoutEdge as an AI-driven talent-scouting and video-analysis platform.
Building a Tech Backbone for Talent Discovery
Before starting up, Satyendra plunged into AI, realising Big Data and AI can assist scouts across India in overcoming biases related to geography, religion, background, or even physical appearance. Additionally, these technologies offer a cost-effective scouting solution, as travelling across the country can be expensive.
In just two months, he earned 84 Microsoft Learn AI badges through an upskilling programme offered by the tech giant.
He then became a familiar face at startup events and competitions, including Startup Singam (top 25 finalist), SRM Bootstrappers Research Council, Global Student Entrepreneurs Awards (top 3 finalist), CRFTHQ Fellowship, StudUp TBI, NextUnicorn Global Awards, ISB IVI, and media features like Sharks of India.
He is currently pursuing a Bachelor of Science in Data Science and Application at IIT Madras.
Over the last two years, Satyendra studied Python, MySQL, machine learning, and GenAI to architect the core of the AI platform. He also designed the database schemas for structuring player data, providing automated score computation pipelines, a dashboard for tracking growth, and an intelligent filtering system for athletes.
On the video analytics side, Kumar experimented with open-source frameworks OpenCV and MediaPipe for pose detection, AlphaPose for skeletal tracking, and YOLOv8 for real-time cricket action detection. Advanced deep learning frameworks like I3D and SlowFast enabled ScoutEdge to read motion across frames to classify bowling and batting techniques.
Budding players like Karan Kannan and Vaibhav Mishra swear by the platform’s possibilities.
Despite representing Andhra Pradesh at U-14, U-16, U-19 and U-23, and later joining Puducherry’s Ranji squad, Kannan hit the same invisible wall Satyendra once did. “Even if I was performing… I was not given that proper opportunity,” he told AIM.
ScoutEdge’s AI-driven insights gave him structured feedback, helping him identify strengths, address weaknesses, and train with greater purpose, he said.
AI Steps In Where Human Eyes Can’t Always Reach
ScoutEdge’s foundational idea is that technology can correct structural imbalance by producing consistent, objective assessments where subjective judgement once dominated. “We are mostly leveraging AI to get better insight for scouting,” Kumar explained. “We use deep learning, NLP, computer vision… to get better insight from the stats we have.”
The founder and CEO embedded this philosophy in the Player Growth Index (PGI), a multidimensional score that evaluates players through performance analytics, AI-driven video analysis and feedback processed using natural language processing, and fitness and mental resilience assessments.
For temporal analysis, understanding movement over time, video recognition frameworks like I3D and SlowFast allow the system to dissect a bowler’s run-up or a batsman’s backlift across frames.
Machine learning models like Linear Regression predict scouting index scores, while classification systems such as Random Forests and SVM (Support Vector Machine) bucket players into skill tiers. Together, these tools give ScoutEdge the analytical depth of a human coach paired with the precision and reproducibility of machine intelligence, the startup claims.
For players, this translates into meaningful, actionable insights. “Before ScoutEdge, it was very difficult for me to analyse what my strengths are, what my weaknesses are,” Kannan highlighted. “ScoutEdge changed this entirely.”
He adjusted his training based on what the platform flagged, “If I’m a little weak in a pull shot… if I could do that drill for 100 balls, now I would do it for 200.”
Mishra, an 18-year-old representing Bihar U-19, had grown up without access to even turf wickets. “Working with ScoutEdge has been a turning point for me. My training became smarter instead of just longer.”
The Chennai-based startup’s AI insights revealed nuances in his bowling patterns, such as seam position, release angles, run-up rhythm, and workload balance. “I’m hoping the platform keeps adding more tools, especially for bowlers, like advanced ball-tracking, fatigue prediction, and session planning,” he added.
For both players, ScoutEdge has become an accelerator—a structured, fair system they never had.
Funding, Scale, and Business Model
ScoutEdge’s traction has translated into significant financial backing. Through Tamil Nadu-based business reality TV show Startup Singam, three investors pledged ₹50 lakh each, totalling ₹1.5 crore. Two additional ₹25 lakh grants came from the Institute of Internal Auditor Delhi’s Ignite programme, followed by ₹10 lakh from Entrepreneurship Development and Innovation Institute Tamil Nadu, and a pending ₹40 lakh Genesis grant.
The startup is further supported by a looming six-year government scheme and incubated at IIT Madras, where Kumar can access research labs, technical support, and business mentorship.
The company’s revenue strategy is diversified.
Players access the platform through individual subscriptions for personalised analytics and improvement plans. The pricing starts from ₹199/month; however, coaches and academies can choose plans ranging from ₹999 to ₹2,499/month based on team size and analytics depth.
Meanwhile, scouts and teams can opt for custom enterprise plans featuring advanced dashboards and analytics that use a SaaS-based model tailored to talent-identification workflows.
The startup also has enterprise-scale contracts to enable leagues and franchises to integrate AI-enabled scouting into their recruitment pipelines. In addition, ScoutEdge is developing paid mentorship channels, trial facilitation services, and consulting offerings for emerging talent.
Currently, ScoutEdge has onboarded over 600 players from different regions and age groups, including district, state, and academy-level athletes. “We are onboarding more players every week as we prepare for a wider launch,” Kumar highlighted.
The startup’s 12-member team collaborates with over 20 academies and training centres, including the Sitamarhi District Cricket Association, the Dumra Cricket Academy, the Asha Indoor Academy, the Elite Sports Academy (Chennai), and the Nexus Cricket High-Performance Centre. It is also in early discussions with organisations like TNCA, BCCI units, and IPL teams like RCB and SRH to develop scouting and data-driven evaluation modules.
ScoutEdge’s key competitors in India include Vtrakit, which analyses cricket data for aspiring players; the Dartle App, a SaaS platform that enhances football academy operations and talent scouting; and Prorecruit India, which allows athletes to create digital profiles to improve recruitment opportunities.
Injury Prevention, Mental Resilience & Chatbots
ScoutEdge also positions itself as an athlete welfare platform. Its injury-prevention system allows players to log pain via a visual body map and receive targeted exercises, mobility routines, workload adjustments, and diet recommendations.
The platform includes a body-mapping system that allows players to indicate areas of pain or discomfort, rate their severity, and receive personalised recommendations. These suggestions include mobility routines, corrective exercises, workload adjustments, and sometimes dietary guidance to manage fatigue or overuse.
The Board of Control for Cricket in India (BCCI) uses injury and workload-tracking systems for national and under-19 players to prevent burnout among fast bowlers. Indian Super League clubs also use GPS vests and video analytics to monitor performance and develop strategies.
Additionally, the Khelo India initiative is creating a digital database of young athletes, recording their performance metrics to build a talent pool for future national teams.
Kumar is also building a conversational AI mentor designed to make expert guidance more accessible. The chatbot will eventually support multiple languages, allowing players across India’s diverse linguistic landscape to engage with frequently asked questions, analyse performance queries, and recommend drills or workload strategies.
Although still under development, it intends to democratise knowledge, giving disadvantaged players structured guidance once available only at elite academies.
ScoutEdge has also begun building a diverse mentorship ecosystem of over 50 people to give young players access to guidance that previously existed only within elite cricketing circles.
Among the mentors currently associated with the platform are Stephen Jones, a UK-based expert who has worked with IPL franchises like RCB and RR; Nishog Naik, widely recognised as India’s youngest scout; Mitans, who is part of the US national cricket setup; and Gor, a talent specialist connected with RCB.
Additionally, the startup is developing a mobile app, currently in beta and to be launched in early 2026.
A New Era of Discovery, Fueled by Stories and Precision
If ScoutEdge achieves its highest ambitions, cricket scouting in India and across the Indian diaspora could evolve into a system defined not by geography but by data.
Globally, sports analytics heavily rely on heat maps, biomechanics, action recognition, and fatigue modelling, and ScoutEdge signals India’s turn toward that future.
As Kannan said, “To be honest, this is just a start.”
The post How AI-driven Sports Tech Startup ScoutEdge is Democratising Athlete Scouting in India appeared first on Analytics India Magazine.
Global Capability Centres (GCCs) in India are entering a defining moment in their evolution. For years, they operated on human-driven models where people performed the critical steps and technology supported them. This aligns with the traditional Human-in-the-Loop (HITL) approach, where human oversight is essential for AI systems to function safely.
HITL now feels outdated. Advances in autonomous agents, enterprise-grade AI orchestration, and context-aware models have shifted the landscape. Among leading GCCs, Agent-in-the-Loop (AITL) is replacing HITL.
At the MachineCon GCC Summit 2025 held in Goa from Nov 29 to Dec 1, Jaywant Deshpande, Chief Solutions & Innovation Officer, Accion Labs, said, “This shift is not simply about replacing humans. It is about re-architecting how enterprises operate.”
As GCCs push for higher productivity and exponential scalability, they are discovering the limitations of human-dependent processes and the power of autonomous AI agents that don’t just assist but deliver end-to-end outcomes.
Why HITL Is No Longer Sustainable
The traditional HITL framework assumes that humans must check, validate, and correct the work performed by AI. This made sense when AI models were error-prone or lacked context. But in high-velocity GCC operations, where millions of transactions, tickets, lines of code, or data points flow every hour, humans appear to be the bottleneck.
Validation queues increase, QA cycles slow down automation, and the cost of human oversight rises linearly while AI output scales exponentially. Leaders across the ecosystem are observing the same phenomenon.
“We found that task-specific agents are better, faster, and 1000x cheaper than humans,” Jaywant mentioned.
This is not about replacing people. It is about repositioning them where their strengths matter most. Agents excel at repetitive, rule-driven work. Humans excel at strategy, judgment, and innovation.
As GCCs move towards AI-native operating models, keeping humans in every loop becomes inefficient, expensive, and unsustainable.
Task-Specific Agents
Unlike broad copilots, task-specific agents are tightly trained systems built for one job: code refactoring, test generation, compliance checks, document extraction, or infrastructure provisioning.
With specialized design, supported by validation agents and structured exception handling, they achieve extraordinary precision and speed.
“We migrated 2.3 million lines of code in under four months, 99% untouched by humans,” Jaywant added.
This level of automation was unthinkable a few years ago, yet it is already standard in advanced GCCs.
In the AITL model:
Agents execute work
Agents validate output
Only exceptions escalate to humans
By automating both task execution and quality control, GCCs are achieving throughputs that previously required hundreds or thousands of employees in HITL environments.
From Humans Checking AI to AI Checking Humans
The most profound shift in enterprise work is a reversal of responsibility. Instead of humans validating machine output, future GCC operating models will use AI to validate human work.
This change marks the arrival of the AITL era, where agents now handle 90%-99% of tasks, provide contextual recommendations, and flag anomalies only when human intervention is required.
“Humans transition from being the principal executors to becoming orchestrators and supervisors. This unlocks a workforce that is smaller but significantly more strategic. Only the GCCs that drive AI and innovation are going to succeed,” said Mandar Garge, SVP & Global Head of Strategy Consulting & Enterprise Transformation, Accion Labs.
Humans remain essential, but not for repetitive work, which often leads to fatigue, inconsistency, and cognitive overload.
Agents deliver:
perfect repeatability
real-time response
zero deviation
continuous execution
instant scale
This shift is not just a marginal efficiency gain but an exponential leap in capability.
For GCCs, the question is no longer whether agents will replace task-driven workflows, but how quickly they can be deployed.
In AITL models, humans focus on system design, policy setting, exception handling, strategic oversight, and innovation—areas where human talent creates real value, particularly in AI-first GCCs.
What GCCs Must Do to Prepare for Agent-Led Operations
A transformation of this magnitude requires more than technology. It demands new thinking, new architecture, and new operating models.
The most advanced GCCs are moving quickly on three fronts:
Automating with agents, not copilots Copilots assist individuals. Agents transform entire workflows.
Building context layers and Knowledge Graphs, not just data lakes Data alone is opaque to agents. Context enables accuracy, autonomy, and safety.
Designing AI systems with an engineering discipline A shared enterprise context to ensure the architecture stays reliable and scalable.
When Not to Use AI
“As far as possible, don’t use AI. Use code. Only use AI where nothing else works,” Jaywant added.
This mindset ensures that AI is applied where it delivers disproportionate value, rather than adding unnecessary complexity.
The GCC of the future will not be defined by real estate, scale, or headcount. It will be defined by AI maturity, particularly the ability to deploy autonomous agents that deliver outcomes at superhuman speed. HITL served the last decade of automation. AITL will define the next.
The post Human-in-the-Loop Is Out, Agent-in-the-Loop Is In appeared first on Analytics India Magazine.
Cognizant has begun work on an 8,000 seat campus in Visakhapatnam and opened an interim techfin centre that can host 1000 associates, marking its largest expansion move in Andhra Pradesh. The ceremony took place in the presence of Chief Minister N Chandrababu Naidu.
The company will invest ₹1,583 crore to build the 22 acre campus in 3 phases by 2033. Construction of the first phase starts in 2026 and will support 3000 associates when it opens in early 2029.
The next phases will take the total capacity to 8,000. The interim techfin centre will operate till the first phase is ready and will work on AI, machine learning, digital engineering and cloud solutions.
“We are proud to welcome Cognizant to Visakhapatnam, a landmark step in advancing our vision of transforming Andhra Pradesh into a world-class destination where global enterprises can build, innovate and grow,” Naidu said.
IT minister Nara Lokesh said the investment strengthens the city’s rise as a technology hub and opens new opportunities for local talent.
Cognizant CEO Ravi Kumar S said the expansion reflects the company’s confidence in the city’s potential and talent pool. The centre includes a client experience space and collaborative areas for teams.
Cognizant has added new delivery centres in Bhubaneswar, Indore and GIFT City since 2024. India remains a key part of its global delivery network, with existing hubs across major cities including Bengaluru, Chennai, Hyderabad, Mumbai and Pune.
Just recently, Cognizant also opened a new AI Lab and a Cognizant Moment Studio in Bengaluru, deepening its India presence as the company accelerates its $1 billion generative AI investment plan announced in 2023.
The post Cognizant Begins Work on 8,000 Seat Visakhapatnam Campus appeared first on Analytics India Magazine.
