Devayan Kumar Sarkar

Most of the work we’ve been doing these past weeks hasn’t been loud. It’s been the kind of product work that only becomes visible when the system starts getting out of your way. Here are a few improvements we’ve been strengthening inside askSOPia — the kind of things that make daily operations feel less chaotic: 1. Voice → SOP (with citations) Teams can now talk through a process once, and askSOPia turns it into a structured SOP with steps, tools, and roles. It’s fast, it’s accurate, and it removes the bottleneck of “I’ll write that later.” 2. Document workflows that actually reflect how teams work Creation → review → approval is now a proper flow, not a scattered set of files. Version control is clean, changes are tracked, and everyone sees the same truth. 3. File management that keeps everything connected askSOPia can now handle almost any document type — including video — and reference those files directly inside AI chat responses. No more jumping between tools to find the source. None of these features are flashy on their own. But together, they reduce drift, speed up handoffs, and make knowledge easier to trust. Quiet work → visible stability. If you could automate or simplify one routine in your operation today, what would it be? #operationsleadership #processimprovement #saasfounder #aiinbusiness #workflowdesign

3

“Is this the right version?” That was the question. And no one could answer it. The policy was in the system. The folder was labeled “latest.” But there was another copy in someone’s inbox. And another printed out — with handwritten notes. So the team paused. Looked at each other. Made a call. The version wasn’t verified. But the job had to move forward. That’s what trust failure looks like. Not a software bug. A slow collapse of confidence. 76% of business leaders say AI is “very difficult” to implement. → BSI, 2024 Not because it’s too technical. Because their team doesn’t trust the answers it gives. And if they can’t verify it — they won’t use it. #TrustInTech #SOPErrors #AuditReadiness #HealthcareLeadership

5

I came across a Forbes piece today asking whether a solopreneur — or a tiny team — could build something massive with AI as leverage. Some people say yes. Some say it’s a myth. What stood out wasn’t the promise of a one-person unicorn, but the reminder that small teams can move with a kind of clarity bigger companies rarely get to enjoy. That part felt familiar. We’ve been building askSOPia with a very small team — switching between product decisions, customer conversations, marketing, strategy, and all the quiet operational work that keeps a system reliable. It’s a lot of hats, but it’s also a lot of speed. When you’re small, you don’t wait for alignment. You don’t drown in handoffs. You see the whole system end-to-end, and you feel the consequences of every decision immediately. That pressure shapes better judgment than any dashboard can. AI gives leverage, yes — but leverage only works when the foundations underneath it are solid. And that’s been our biggest lesson so far: being small isn’t a limitation — it’s a design advantage. Curious to hear from other founders and small teams: What has being small allowed you to do that a larger team couldn’t?

1

1 commentaar

I tested and compared all the features of Sunsama and Morgen side by side. They're both $15-20/month daily planners, but they solve opposite problems. Sunsama makes you plan manually. Every morning, a ritual walks you through: ➜ What matters today? ➜ How long will this take? ➜ What can wait until tomorrow? It's slow. But that's the point. When you estimate tasks yourself, you can't lie about what's possible in a day. Sunsama warns you when you're overcommitted. Morgen does the opposite. The AI looks at your tasks and suggests a complete daily schedule. You see the plan, adjust what doesn't work, then approve it. Faster. Less manual work. But you need to teach it how you work using "Frames", time blocks where only certain types of tasks belong. The pricing surprised me. Morgen includes scheduling links (replaces your $15/month Calendly subscription). Sunsama doesn't have them at all. And Sunsama's AI voice assistant, Sunny? It's legitimately impressive. But it costs $50/month instead of $20. If I were you, I'd ask myself: ➜ Do I consistently overcommit and need structure to plan realistically? ➜ Or do I just hate the manual work of planning and want AI to handle the first draft? Find the full breakdown in comments—features, integrations, and which tool fits which workflow.

12

2 commentaren

Code is becoming cheap. Trust is becoming expensive. 📉 I recently read a great piece by Kailash Nadh titled “Code is Cheap”, and one takeaway really stuck with me: “One is now slowly being compelled to also look much more closely at the provenance of software — the who, why, their track record, and plans of governance.” In an era where AI can generate thousands of lines of code in seconds, the volume of software is exploding. But as the cost of production drops to near zero, the risk of black-box dependencies skyrockets. We can no longer just ask, “Does this code work?” We have to ask: - Where did this come from? - Who is maintaining it? - What’s the long-term governance model? When software is everywhere, provenance becomes the real currency. This is why the industry is shifting toward Secure by Design thinking , not just vulnerability scanning, but visibility across the entire lifecycle of a package. Platforms that focus on software supply-chain integrity (like what we’re building with JFrog AppTrust) aim to address this gap: ensuring what teams deploy is verified, governed, and trustworthy. The future of development won’t be about who writes the most code. It’ll be about who curates the most reliable ecosystem. In a world where code is easy, trust is the hard part. Are you looking more closely at the “who” and “why” behind your dependencies lately? 🔗 Link in comments #SoftwareSupplyChain #DevSecOps #SecureByDesign #JFrog #AppTrust

12

1 commentaar

Requirement gathering fails when we only listen with a “developer mindset.” When clients explain their business workflow, they rarely speak like developers. They speak in pain points: • “It’s difficult to control approvals.” • “We don’t know the real stock.” • “Reporting takes too long.” • “Everything depends on one person.” If a developer collects requirements and focuses too early on features, the real problem can be missed. Because the brain starts translating everything into: “Okay, we will build X, add Y, customize Z…” And some important details never get explored. On the other side, if a non developer collects requirements, there can also be a gap: They may capture the conversation, but they may not understand how the ERP works end to end. So the developer receives a simplified list like: “They want this, this, this…” Then something common happens: The developer suggests a better way to build the workflow (based on ERP logic), but the collector rejects it and says: “No, the client wants it exactly this way.” (እነሱ እንዲህ ነው የፈለጉት) Later, during testing, the client sees the system and says: “Can the system work like that instead?” (እንዲህ መስራት አይቻልም?) And that was actually the right approach from the beginning, but it was missed in requirement collection. That’s why I believe requirement gathering is the most critical phase in ERP. The best requirement collector is someone who can do both: • listen like a business partner (not like a coder) • understand how ERP works (flows, data, controls, reports) • translate pain points into a clean process and system logic Because ERP implementation is not just installing software or building custom apps. ERP is about solving real operational pain and enabling better decisions. Your code is only as good as your understanding of the business. How do you handle requirement gathering in ERP projects, who usually collects it in your team? Heleus AI #ERP #HeleusAI #AI #BusinessAnalysis #Requirements #DigitalTransformation #Implementation

10

What does India’s next generation of travelers want — and how can smart policy keep up? 🌏 With India becoming one of the world’s most dynamic travel markets, Yang Li from Booking.com explores how digital adoption, social media inspiration, and a passion for authentic experiences are reshaping tourism, and what policymakers can do to support this momentum. A fascinating read for anyone interested in the future of travel, tech, and policy in South Asia. #TravelTrends #SmartTourism

8

From a beginner who only learned A/B testing from textbooks to now running over 20 A/B tests — the journey hasn’t been linear. In my first few experiments, the success rate was roughly 1 in 3. In 2025, every experiment I ran was successful — not because the ideas were magically better, but because the approach improved. Over time, it became clear that better outcomes were driven by thoughtful design and strategy, not simply by choosing different variants. I learned to be more intentional about: * what’s actually worth testing * how to size opportunities (even when estimates are wrong) * why focus and sequencing matter more than running “more” experiments I wrote a short article reflecting on my A/B testing journey — what changed, what increased impact, and what I’d do differently if I were starting again. 👉 https://lnkd.in/d9Tgid2n Hope this article can help you make your experiments more effective and impactful.

30

4 commentaren

Want to get more leads? (spend more time on this) Many salespeople I know just build lists based on the basic criteria from data providers. Looks fast. Feels easy. But it’s a gamble. They think they need better copy or a better offer. But what they really need is better lists. You can have: - Great copy - Irresistible offer - Top cold calling skills - Premium deliverability - A big brand on LinkedIn When your list is hyper-relevant, everything else in sales and marketing gets easier. Just build hyper-relevant lists. Or you won’t get the results you expect. Agreed?

31

19 commentaren

"The driver asked me to cancel and take the ride offline." > We're all building "user-first" products. But building for India, a nuance that rarely gets attention: > What users do != what users say, or in the fancy PM language, the noise associated with implicit v/s explicit user signals. I've seen this 9/10 times around me: 1. Cancelling a cab? Pick: "The driver asked me to cancel and go offline" - whether or not it happened. 2. Didn't like the food - want a refund from Swiggy? Select: "The food was raw/uncooked" - fastest route to full refund. 3. Didn't like the crochet top you just ordered? Choose: "Product doesn't fit" - less explanation, faster approval. > In India, the intent behind the input is often to minimise perceived friction, not to share actual feedback. They fill them strategically, or the easiest way out, or to avoid penalties. > And unfortunately, these are loss-making use-cases for startups, and everyone is behind optimising these, whereas the core problem might be different. We gospel this truth in dashboards, end up misreading our users, and shipping the wrong fixes. TLDR; if you're someone or know someone close to a better solve on these noisy signals, feel free to comment or DM! 😃 Listen to the noise - sometimes it's just around you. Amazon Flipkart Myntra Meesho Swiggy Zomato Zepto Ola Uber Rapido Shashwat Singh

174

10 commentaren