Vedant Awasthi

Why manufacturing AI now? Not because factory data suddenly became perfect. It did not. Because systems can finally work with more than clean database fields: transactions, notes, schedules, documents, and human questions can participate in the same decision. For years, the order lived in the ERP, the sequence in Excel, the quality concern in a note, and the latest reality inside one person's head. Connecting that context was slow and expensive. The technology has become more flexible. That creates a real opportunity for manufacturing, but it also creates responsibility. Our job is to make that flexibility dependable enough for a real decision on a real shift.

The planning spreadsheet may be fragile, but it often contains a working theory of the factory. Customer priorities. Tooling constraints. Jobs that can safely move. Exceptions the ERP never modelled. Do not replace the spreadsheet before understanding the intelligence inside it.

Steel is a weight-losing raw material, meaning logistics will drain up to 35% of your operating budget if you set up in the wrong state. The highest-margin steel plants in India aren't in the commercial hubs—they are in the mineral belt: Odisha, Jharkhand, and Chhattisgarh. By setting up directly on the Eastern Freight Corridor and eliminating cross-state raw material transport, factories can achieve 15% to 22% gross operating margins (compared to just 8% inland). The biggest untapped opportunity right now? High-grade structural steel for solar mounting and pre-engineered buildings (PEBs) to feed India's infrastructure boom. Build where the minerals are mined, not where the finished goods are bought. #SteelIndustry #ManufacturingIndia #BusinessStrategy #MakeInIndia #SupplyChain #HeavyIndustry #Odisha #BusinessTips

Every factory already has an operating system. Part is in the ERP. Part is in Excel. Part travels through calls and messages. The most important part may live in the habits of people who know when the official process no longer fits reality. Calling this "no system" misses the point. The system grew around customer demands, old software, missing information, and the need to keep production moving. Some workarounds create risk. Others are protecting the plant because the formal process cannot handle the exception. New software should understand that operating system before trying to replace it. Manufacturing transformation should begin with respect for the intelligence already keeping the factory alive.

A long product roadmap can hide weak understanding. Three factories may ask for three different features while describing the same operating problem. Adding all three is easier than understanding the common decision underneath them. The roadmap should record problems understood deeply enough to solve.

“Any update?” 😭 “Stock hai na?” 💀 “Just one quick follow-up.” 🫠 Factory life really said: choose your fighter. [AICAN Optiwise, Factory Life, Inventory Management, AI Agents, Manufacturing Technology] #AICANOptiwise #FactoryLife #ManufacturingMemes #IndustrialAI #SmartManufacturing

The least useful product feedback is: "Looks good." The useful version is uncomfortable: this field is wrong, our operators will bypass that step, your recommendation ignored a tooling constraint, nobody trusts this data after the second shift. That feedback removes the clean product we imagined in our own room. It also brings us closer to something a manufacturer can trust. I want Optiwise tested on real operating questions and messy data, not only a polished demo. If you run a factory and are willing to challenge us, bring one problem your current system cannot explain. I would rather find where Optiwise fails now than protect a comfortable demo.

Before manufacturing AI earns autonomy, it has to earn the right to be taken seriously. Show the evidence. Expose the assumption. Admit uncertainty. Let the person remain in control.

Your old tyre doesn't necessarily end its life as waste. It can become carbon black, pyrolysis oil and recovered steel — all materials with industrial applications. Finster Carbon says it processes around 15 million end-of-life tyres annually, with reported pyrolysis-oil production of around 36,000 tonnes a year. The interesting business isn't the tyre. It's what happens after the tyre stops being useful to you. A discarded product becomes a source of industrial raw materials. And that's where recycling gets much more interesting: when waste starts competing with virgin materials. #Manufacturing #CircularEconomy #TyreRecycling #WasteToValue #IndianManufacturing #Recycling #IndustrialManufacturing #SustainableManufacturing

Predicting that an order will be late is useful. But the planner still has to decide: change the sequence, split the quantity, add a shift, expedite material, call the customer, or accept the delay? Industrial AI becomes valuable when it helps with the decision after the prediction.

Most software looks good in a calm meeting room. The real test starts when the line is down, the customer is calling, the supervisor is handling three issues, and the person who knows the workaround is elsewhere. On that shift, every extra login matters. Every unexplained recommendation matters. I keep asking one question while building Optiwise: would this feature still help during the worst hour of the week? If it requires perfect data, patient users, and someone from our team standing beside it, it is not ready for a factory. Manufacturing software should reduce pressure when the factory is already under pressure. That bad shift is where trust is either built or lost.

The ERP is not dead. It holds the orders, items, inventory, suppliers, and years of transactions that keep the business standing. The opportunity is the decision layer above it: connecting records, live conditions, exceptions, and human context quickly enough to act.

This is Optiwise letting you know about all that happened this week! [news, manufacturing, weekly, aican, optiwise]

The best operators notice things software misses. If an experienced operator rejects an AI recommendation, that should not disappear as an "override." The system should ask what condition it missed and learn from the answer. Good operators should make manufacturing AI better. Manufacturing AI should make their judgment travel further.

“How’s your business going?” 😀🤝 Me juggling 27 urgent requests, 15 client calls and 47 “final_final_v3” files. 💀😂 Big ambitions. Busy days. Smarter operations with Optiwise.🏭🧡 #AICANOptiwise #BusinessMemes #FactoryLife #ManufacturingAI #Optiwise [manufacturing, business memes, client requests, factory life, Optiwise]

Real time is not a feature by itself. If nobody knows what the signal means or what to do next, the factory only receives confusion faster.

"Bad data" is not always a people problem. A late entry may mean the terminal is too far away. A blank reason code may mean none of the options matches reality. A private spreadsheet may exist because the official system cannot answer the planner's question. Sometimes the data is reviewing the product.

88,000 solar cells. Every hour. That is the designed output of Premier Energies' new 7 GW solar-cell facility in Naidupeta, Andhra Pradesh. The ₹3,293-crore plant spans 101 acres and takes the company's total solar-cell manufacturing capacity from 3.6 GW to 10.6 GW. But the bigger story is how this factory is being built. N-type TOPCon technology. Automated material movement. Automated packing and packaging. AI-assisted process control and predictive performance analysis. And Premier isn't stopping at cells. The company is also moving towards ingot and wafer manufacturing, pushing further upstream in the solar supply chain. Once stabilised, the facility is targeting around 25.8% average cell efficiency. This is what modern manufacturing increasingly looks like: scale + automation + process control + deeper integration. #Manufacturing #SolarManufacturing #FactoryAutomation #Industry40 #ManufacturingTechnology #SolarEnergy #TOPCon #IndianManufacturing

Sample data makes every manufacturing product look intelligent. Real factory data decides whether it actually is.

A machine stops. Someone says, "We had this problem before." Then the calls begin. Production checks with quality. Quality remembers maintenance changed something. Maintenance searches an old message. The person who solved it is on leave. An hour later, ten capable people have reconstructed knowledge the factory had already earned. This is one reason manufacturing intelligence matters to me. The transaction tells us what finally happened. It rarely preserves why the team made that decision or which small intervention solved the problem. The factory already paid for that lesson once. Its system should remember it for the next person and the next shift.