Sales rose after your launch. Was it the feature, or December? Meet Winter editor @BhogalePrasanna at The Fifth Elephant Winter edition, with his book on causal inference. 30 copies on sale, signed on the spot. 4–5 Dec · Bengaluru Programme: hasgeek.com/fifthelephant/wi…

“Who wants free LLM keys?” This hackathon question led @boskykode to start @aigrantsindia At Fifthel Pune, he shared how AI Grants India has supported 200+ startups, and invited ideas for what comes next. Watch his pitch: youtu.be/V0yXW0yzqTM

New video out from AI & Research track at The Fifth Elephant 2023, curated by @nischalhp & @sumodm Priyanka Banik of Swiggy explains demand forecasting for Swiggy Instamart. @Swiggy

The problem: if an item goes out of stock, historical sales are not true demand. The model may look fine offline while the business gets lower availability or higher wastage. Classic ML metric mischief. Watch: youtu.be/VczDlwRn7QE

One fabric mill is late. Which products, stores and promotions are affected, and what can the retailer do next? At Fifthel Winter, Abhijith Neerkaje, co-founder at @BeyondVectors will present: What breaks next? Tracing a retail disruption and checking AI recommendations

Last year, @BhogalePrasanna spoke at The Fifth Elephant Winter Edition. This year, he is helping shape it as an editor. In this reflection recorded in 2025, he describes the care taken to review talks & give feedback & the enthusiastic engagement he experienced after presenting.

5/5 The deadline is 11 October. Your idea can develop through editorial discussion. Start with the problem, your findings, and what you can show. Share your work at The Fifth Elephant Winter, 4–6 December 2026. Submit: has.gy/KaLO

4/5 This is the inquiry we want at Winter: practical use cases, careful analysis, and evidence that changes how we work. Have a metric that concealed a failure? An analysis you reconsidered? An experiment that changed your mind? Bring that story.

3/5 Another proposal examines the process of disproving your own data story. What will make you question a convincing finding? What evidence will make you abandon it? Both ideas are under editorial review and being shaped.

2/5 In one proposal, @aloshydenny examines how a metric can reward a system for doing nothing. He proposes adding “liveness” measures: is the system actually doing useful work alongside scoring well? What might your metrics be hiding?

1/5 A system scores 100%. It has stopped answering. A convincing data story gets challenged, by its own author. Proposals are coming in for The Fifth Elephant Winter 2026. Here are two ideas on the editorial table as the 11 October submission deadline approaches. 🧵

2/6 Your organization has official dashboards. Teams still keep their own spreadsheets, and bring different answers to the same meeting. What creates those differences? What have you tried to resolve them? That might a story worth investigating ...