Ethinos

Big ideas. Sharp thinking. Real impact. We’re excited to spotlight Brijesh Munyal, Co-Founder & CEO at Ethinos Digital Marketing, as he takes the stage at Intrigue Madverse Singapore 2026 for a powerful keynote on Winning Brand Visibility in the Age of AI Answers. In a world driven by AI and shrinking attention spans, it’s not just about being seen it’s about being remembered. 📅 23rd April 2026 📍 Novotel Singapore on Kitchener 🔗 RSVP here: 🔗 https://lnkd.in/giND9FPz #Ethinos #MadverseSingapore #DigitalMarketing #FutureofMarketing Brijesh Munyal Tom Hales Yuvraj Rao Ephraim Ritikos Siddharth Ponnamparambil

A few weeks ago, I sat down with zero plan and just started building.  Last week, that experiment won 2nd place at the Ethinos AI Buildoff 🎉 I went in with just a genuine curiosity to build something useful, not just for SEO folks like me, but for anyone on the team drowning in messages and emails. That's how FollowFlow AI came to life. What it does: FollowFlow AI monitors your Teams messages and Outlook emails, automatically pulls out tasks and deadlines, logs them into a Google Sheet, and sends reminder emails before things are due. It also flags client emails that haven't been replied to in over a day. No manual to-do lists. No relying on memory. Nothing slipping through the cracks. How I built it: The stack, honestly simpler than it sounds: •⁠ ⁠n8n to wire everything together •⁠ ⁠Microsoft Teams and Outlook as inputs via Power Automate •⁠ ⁠ChatGPT to read messages and extract tasks •⁠ ⁠Google Sheets to store and track everything •⁠ ⁠Outlook to send the reminders The flow: a message comes in, the AI reads it, extracts the task, owner and deadline, saves it to the sheet, and sends a reminder at the right time. Simple in theory. Messier in practice. The part that humbled me: Microsoft Teams was the hardest part. There's no clean native trigger in n8n, so I had to bridge it through Power Automate to get messages flowing in. And when they did come in, they were full of HTML tags that confused the AI completely, so I had to strip those out first. On the AI side, the model kept returning markdown instead of clean JSON and breaking everything. Fix was simple: just tell it very explicitly in the prompt. "Return ONLY raw JSON. No markdown. No extra text." One line. Problem solved. Prompt engineering really is half the job. What I’d do differently: Tasks currently stay "Pending" until someone updates the sheet manually. Auto-detecting completion is something I want to solve next. Also, it works best in English. Hinglish confuses the model, which is a real gap for an Indian agency setup. If anyone has cracked this, genuinely would love to know how. If you want to build this yourself, here's exactly what I used: ✅ n8n, free to self-host ✅ For Teams: Power Automate for personal accounts or Microsoft Azure app registration. Azure works better for work accounts ✅ ChatGPT API key. Free tier works to start but you'll want a paid one once you're doing a lot of testing and retries ✅ Google Sheets ✅ A prompt that asks for structured JSON and is very specific about the format Attaching the demo video so you can see it in action. If you want the exact prompts or the workflow breakdown, comment AI Buildoff Thank you to the Ethinos team for pushing us to build and not just talk about AI. And if you've been thinking about building something with AI but haven't started, just start. Seriously. You don't need to have it all figured out. You figure it out by doing. What are you all building these days? Would love to know 👇