Sanjay Narang

As customers are becoming more conscious of data privacy and selective, retailers are increasingly looking at optimizing the touchpoints, orchestrating channels, fine tuning tailored campaigns, and moving away from "spray and pray" to "start with dart" approach to augment both physical and digital customer footprints. Glad to be featured in Adtech Today Middle East to discuss how generative AI with the right mix of agentic workflows can maximize conversion, and what it takes for retailers to get there

116

3 Comments

Reflecting on the journey as a co-founder of Embibe - India's leading AI-driven EdTech platform - and following its meteoric trajectory over the years, fills me with immense pride and gratitude. Kudos to Aditi Avasthi Surri and the Embibe team for building something that truly epitomizes "technology for human flourishing". Back when we started building Embibe, AI wasn't the industry buzzword it is today. We were driven by a simple yet powerful vision: democratizing quality education through intelligent technology that could adapt to every student's unique learning journey. While others were focused on digitizing traditional teaching methods, Embibe took a fundamentally different approach by building AI systems that could: ⭐ Analyze individual learning patterns and knowledge gaps ⭐ Provide personalized study paths for millions of students ⭐ Predict performance outcomes and recommend targeted improvements ⭐ Make quality education accessible regardless of geographic or economic barriers What started as an ambitious idea has now impacted millions of students across India, helping them achieve their academic dreams through truly personalized learning experiences. The best part? Embibe was solving real problems with AI when it was still considered "niche technology." Today, as AI transforms every industry, I'm proud of how Embibe has pioneered this space by applying it meaningfully to education - not for the sake of technology, but for genuine human impact. To every entrepreneur building with purpose: sometimes being early to the game means standing alone for a while. But when you're solving real problems for real people, the market eventually catches up to your vision. Here's to the power of education, the potential of AI, and the incredible team that made Embibe a reality.

46

4 Comments

See how IBM is working with Oracle to bring the power of watsonx to OCI—designed to fuel a new era of multi-agentic, AI-driven productivity and efficiency across the enterprise. https://lnkd.in/gck-K-YV

3

Very Interesting perspectives, I may disagree with a few and align with Many of the views shared, interesting views, perspectives, and practices nevertheless !! 1. The Counterintuitive Truth: Code Quality Does Not Equal Product Success Prasanna’s most striking lesson for engineers is that there is no direct relationship between high code quality and building a successful product. While technical excellence is a worthy goal, it is not the end goal. The ultimate measure of success is whether you are solving a real problem for people. < I am on the Edge with this one. While I am a strong proponent of Quality output/outcome, the ultimate measure of success is, however, the real problem solved :) Ramanan Natarajan - Would love to hear your thoughts on this :) 2. The Real Productivity Engine: Structure as the Foundation for Tools While Block is gaining attention for its powerful internal AI agent, Goose, Prasanna reveals that the single most impactful change for boosting productivity wasn't a tool, but a fundamental organizational restructure. The company shifted from a siloed structure to a unified, functional one. This new functional structure unlocked three primary benefits: > It created a singular focus on technology and technical excellence across the entire organization. > It allowed leaders to drive forward AI and platform depth for the whole company, rather than in isolated pockets. > It changed the hiring mindset. Instead of treating engineers as a "commodity" to be added to a business unit to increase feature output, "mythical man month", the focus shifted to building common platforms and leveraging deep experts across the entire organization. This unified structure not only drove technical depth but also created the ideal environment for small-scale experiments to be identified, supported, and scaled across the company, rather than dying within a business unit silo. < It's not always the Tool, it's the Mindset !! 3. The Principle of Momentum: Start Small to Win Big: The most successful and durable innovations rarely begin as massive, top-down corporate initiatives. Instead, they grow from small, focused experiments. < "Start small with everything." 4. The Human Element: Hire for a "Learning Mindset": The company now optimizes for a "learning mindset." They are less concerned with finding "amazing AI practitioners on the get-go." Instead, they prioritize finding people from recent college graduates to industry veterans who are eager, open to learning, and willing to embrace new tools. < Meta cognition is undervalued !! #organisationCluture #AI #Productivity https://lnkd.in/gw6saKXp

11

Automation is entering a new phase. In a recent conversation with Techzine, our VP of Automation Sathish Balakrishnan explains how Red Hat Ansible Automation Platform is evolving into the execution layer for agentic AI. As organizations move from experimentation to real AI-driven operations, having a trusted automation foundation becomes critical. A very insightful discussion on what’s next for AI and automation. Check out the full interview below. #RedHat #Ansible #AI

42

5 Comments

#AI #Governace #development #velocity Amazon implemented a 90-day safety reset for its e-commerce systems following massive outages linked to its "Q" AI tool and unapproved code changes. The new policy requires dual-human reviews for all code and strict use of internal approval tools to prevent further multi-million order losses. https://lnkd.in/gzi5khhq

10

People are what make a service exemplary and people are also what can ruin it. Service industries are inherently more difficult to scale due to high dependency on people and the quality of their output. Typically there are people on both ends of a system - - Internal users who operate behind the scenes, intuitively and efficiently. + - External users who benefit from the experience, often without needing to know how the system works. How these two sides work well together depends on how well the system is designed for everyone. Take the example of a fine dining restaurant service: - Chefs, sous-chefs, and kitchen staff managing time, prep, plating. - Waitstaff, sommeliers, and hosts reading cues, adjusting pace and meeting the demands of customers. - Managers managing reservations and guest experience. - Diners, who come in, order, and enjoy an experience that feels seamless and thoughtful. (If they entered the kitchen, they would see a whole different world!) The end customers don't even notice the silent or not so silent orchestra of internal cues, softwares or subtle hand signals playing across the kitchen, service team, and front-of-house—but they see the results: a great meal, on time and an elegant experience. What does it take to design a system with the best customer experience- 1: Each person involved in strategising, execution and on the ground must understand the 'why'. This helps to arrive at appropriate solutions for each problem with the larger goal in mind. 2: Each step of the system should be intuitive to use. If someone has to ask for support for their daily operational tasks, then it is not a well designed system. It should be designed in a manner that it reduces errors and multiple reworks. 3: Training and retraining matters. It helps to listen, iterate and improve the work flows and processes set up. My biggest take away and one that I apply to all problem solving exercises, is that 'Simplification is a business's super power'. Also mine (ask my team!) 😊 #systemdesign #processdesign #serviceindustries #scalingservices

19

Varun Bansal, CIO, Bata India, speaks about the company’s steady digital transformation—leveraging Oracle Cloud to modernise core systems, enable scale, and build resilience across stores and supply chains. https://lnkd.in/gRTCaY6A

1

Chapter III — Signals to Systems 🔱 Day 1 — Leadership (Detection) Signals in Systems Reflections from Ravi Leadership systems rarely fail suddenly. They weaken quietly. A shift in tone. A delayed response. A hesitation before speaking. Small signals precede visible breakdowns. Most teams respond to outcomes. Few develop the discipline of detection. What is visible is often late. What is subtle is usually early. Strong leadership is not only about execution. It is about noticing. Systems create order. Signal awareness protects it.

37

42 Comments

🚀 Modern Infrastructure is Changing the Rules of the Game 🔧💡 Gone are the days when vertical scaling meant upgrading bare-metal servers with fingers crossed and long change windows. In the cloud, scaling is no longer a blocker — it's a dropdown option. 🌩️💻 Today, we can spin up instances with: 🔹 160+ vCPUs 🔹 TBs of memory 🔹 Low-latency SSDs All within minutes. That’s not just convenience — that’s transformation. This shift is redefining how we approach traditional RDBMS design. The old concerns of CPU bottlenecks, memory limitations, and rigid capacity planning are fading. With compute power at our fingertips, we can: ✅ Focus less on infrastructure ✅ Focus more on faster feature delivery ✅ Build with scale, resilience, and performance by default 💬 For DBAs and engineering teams, this is the call to evolve: from gatekeepers of resources to enablers of velocity. Even the CAP theorem conversations are changing — availability isn’t the headache it once was. Cloud-native design + powerful infra = new rules. The cloud isn't just where we host systems now. It's where we reimagine them. 🌐✨ #CloudComputing #InfrastructureEngineering #DatabaseManagement #DevOps #Scalability #SystemDesign #ModernEngineering #CloudNative #Innovation #DBA #CAPTheorem #FeatureDelivery #TechLeadership

10

Future into AI. Mark my words, there will be two kinds of engineers soon into future. One who grow fast in their career at immense pace because they perform 100x. They build pocs in hours and not days and weeks using AI. They react to customers problems almost immediately and resolve their issues in no time. Others who, rely “only” on AI. They struggle to find working solutions with AI. Like others they too are using AI in their day to day task but have not yet figured out their way to identify the right way to “prompt”. Prompting better and being able to identify the hallucinations of AI and augmenting it with your own intelligence will be the key skill going forward. As an example, there is lot of knowledge about financial markets but only a few out of billion people on earth can still make proportionally high margin of money out of the market compared to rest of the people. These folks usually have the skill to understand and pick the right investment strategy at right time. Its going to be very similar now. Engineers who work with AI “and” have a lot of subject matter insights, will ask right questions and identify hallucinations and course correct sooner will now grow faster! Dont fight AI rather develop deeper insights and subject matter expertise using AI to supplement your world view and adapt.

197

6 Comments

Sometimes the most powerful insights come from revisiting the basics. Watching Google’s gRPC team break down service mesh fundamentals reminded me: clarity is a form of leadership. As infra architects, we often juggle complexity—protocol risk matrices, compliance automation, business model mapping. But simplicity isn’t the enemy. It’s the foundation. Back To the Basics - Service Mesh Fundamentals - Kannan Jayaprakasam & Pawan Bhardwaj, Google https://lnkd.in/gBN9nd3N

6

I had the pleasure of joining nybl in their podcast series Deep Tech Boardroom where we discussed the challenges that organisations face when it comes to AI, and a lot of the discussion was around the tech itself but after un-packing this more culture and education is far more important than the tech itself to enable the organisation. Now the funny thing about this is we shot this podcast in 2024 where organisations were at their very start of their AI journeys - but yet here we are in Jan 2026 and as organisations have begun to adopt AI I still see the same challenges from back in 2024. - We can't blindly apply AI and think the the value will be created, stress test the hypothesis. Have we truly asked why and mapped that to a impactful business KPI? - Do we understand the process? Are we simply applying AI to remove humans and manual steps and the value is created by removing human-capital/time? - Perhaps in certain use-cases we re-imagine the process completely? Remember a lot of the processes were designed before a lot of the technological advances that we have today so is the process right and is there a more efficient and impactful process we can design to get higher value? So lets take a step back - lets learn from the past 2 years (and the past 50 years of tech!) - lets ask why and enable our teams to understand the processes and how we can, in some cases - reimagine the process. Remember - technology is here to make our lives better, not harder - so lets continue to bring our organisations and teams along the journey with us. p.s I'm saying AI to cover off the entire field of AI, not just GenAI and Agents but also ML, Computer vision, Deep learning and all the other areas of AI that GenAI and Agents are a part off :) p.p.s - no AI was harmed in the writing of this post :) #nybl_is_everywhere #Boardroom_by_nybl #AI #ML #DeepTech #MEA #agenticAI #proccess #AIvalue nybl Hala Alnsour Nathan Hanley Mohammed Shono Nicholas Borsotto Machado Monteiro

54

5 Comments

✨ New on Medium: I’ve shared key insights from Microsoft’s “MCP for Beginners” series and why the Model Context Protocol (MCP) is shaping the future of generative AI. 🔑 Highlights include: How MCP acts as the “USB-C of AI” to connect models with tools, APIs, and data. Why it reduces hallucinations and boosts scalability. The client–server architecture powering MCP. Security best practices and real-world use cases. If you’re curious about building smarter, more secure AI applications, check out the full write-up here:

13

1 Comment

🛒 How the SPAR Framework Powers AI in POS & Retail |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| In retail and POS ecosystems, AI isn’t about replacing point of cashiers—it’s about making smarter, real-time decisions across millions of transactions. The SPAR loop (Sense → Plan → Act → Reflect) provides a structured way to build AI agents that are reliable, adaptive, and aligned to business goals. Here’s how SPAR can guide scalable architecture in POS/retail:   1. Sense Gather data and signals from diverse sources: ◦ Point-of-sale terminals (transaction logs, item scans) ◦ Inventory / supply chain systems (stock levels, delivery timings) ◦ Customer profiles & loyalty data ◦ Real-time context (weather, store footfall, local promos) This sensory layer ensures retailer AI “knows the state” before making decisions.   2. Plan/Process Generate decision paths based on sensed data and business goals: ◦ Dynamic pricing or discount decisions ◦ Inventory replenishment thresholds ◦ Personalised cross-sell / upsell recommendations ◦ Fraud risk scoring and flagging The planning engine evaluates trade-offs (profit vs. margin vs. stock risk) and lays down execution paths.   3. Act Execute the chosen plan via system integrations: ◦ Push price updates to POS terminals or digital systems ◦ Trigger stock replenishment orders ◦ Deliver personalized offers at checkout ◦ Block suspicious payments or require extra verification “Act” is where AI becomes tangible in customer experience and operations.   4. Reflect/Refine Evaluate outcomes & improve continuously: ◦ Did the discount uplift sales as expected? ◦ Were items overstocked or understocked? ◦ Did any fraud go undetected? ◦ Use this feedback to re-train, refine thresholds, or adjust decision weights Reflection closes the loop so agents improve over time rather than staying static.   🎯 Why SPAR Matters in Retail / POS AI ‣ Reliability & Governance: Avoid brittle, one-off AI models by building continuous feedback loops and enhancements. ‣ Scalability: SPAR enables modular agent components per store, region, range and/or product line. ‣ Business Alignment: Each cycle ties back to metrics (margin, conversion, stock turns) — not just technical accuracy. ‣ Future-Proofing: As new signals or systems emerge (e.g. IoT sensors, external demand data), they plug into “Sense” or “Reflect” without rearchitecting the whole system.   If you’re architecting AI for POS, retail, or omni-channel systems, consider SPAR as your foundational mental model. It elevates AI from proof-of-concept to resilient, meaningful business instrument. What retail AI scenario would you apply SPAR to next? #EnterpriseArchitecture #RetailAI #POS #ComposableArchitecture #SPARFramework #DigitalTransformation #MachineLearning #EdgeAI

4