Every time I listed something on Facebook Marketplace, I felt a quiet dread.
Not because selling is hard. But because the experience was designed to frustrate you into giving up.
You upload your item. You wait. Someone messages you — "is this still available?" You say yes. They ghost you. Another person asks the same question. You answer. Ghost. Repeat until you either sell the item or lose the will to live.
I've been through this cycle with OLX, Facebook Marketplace, and Craigslist. Three platforms. Three different UX disasters. One shared soul-crushing experience.
The Problem With Existing Classifieds
Let me be specific, because vague complaints help no one.
Facebook Marketplace makes you fight the algorithm. Your listing competes with thousands of others, and unless you boost it with money, it sinks. The messaging system is buried inside Facebook's notification hell. Half the time, legitimate buyers get filtered into a "Message Requests" folder they never check. And if you're in a country where Facebook's infrastructure is spotty? Forget it.
OLX (huge in Eastern Europe, Central Asia, and South Asia) is better in some ways — but it's stuck in 2015. The app is bloated. The search is mediocre. Spam listings are everywhere. And the "premium" placement fees feel predatory when you're just trying to sell a used blender.
Craigslist is a different problem entirely. It works fine if you're in the US and don't mind interacting with a website that looks like it was designed during the Clinton administration. But outside North America, it's basically a ghost town.
The core issue with all of them: they're destinations. You have to go to them. Download their app. Create an account. Learn their interface. And your buyers have to do the same.
Then I Looked at Where People Actually Were
Here's the thing. In the markets I care about — Central Asia, Eastern Europe, Southeast Asia — people weren't on Craigslist. They were on Telegram.
Not just chatting. They had Telegram groups for everything. Local buy/sell groups. Neighborhood communities. People posting photos of stuff they wanted to sell directly into group chats. No middleman platform. Just people talking.
It worked, but it was chaotic. Listings got buried under conversation. There was no search. No structure. No way to filter "show me all furniture in my city."
I kept thinking: what if you could keep that organic, chat-native feel but add just enough structure to make it actually useful?
Why Telegram Is the Right Foundation
Telegram has crossed 900 million monthly active users. In many emerging markets, it's not just a messaging app — it's the internet infrastructure. People use it for news, for community, for commerce.
More importantly: everyone already has Telegram. No app download required for buyers. No new account to create. If someone wants to buy something from your listing, they message you on the same app they use to talk to their family.
That frictionlessness is massive. Every extra step in a buyer's journey kills conversions. "Download our app" is a step that kills a lot of conversions.
Telegram bots make this even more powerful. A bot lives in the app your users already have. It can process photos, handle commands, maintain state, send notifications — all through the chat interface people already know how to use. Mobile-first by default, because Telegram is mobile-first by default.
Building the Thing
So I built it. A marketplace bot for Telegram called k4pi (@k4pi_bot).
The stack is Python (aiogram 3), PostgreSQL, Redis, Elasticsearch for text search, and — this is the part I'm most proud of — an AI image understanding layer using SigLIP embeddings. When you upload a photo of an item, the bot automatically categorizes it, checks it for appropriateness, and indexes it for visual search.
That last part matters more than it sounds. On a mobile-first platform, most listings start with a photo, not a description. Someone snaps a picture of their old guitar and posts it. If your search only works on text, you miss all the listings where people didn't bother writing much. Visual search closes that gap.
The flow for sellers is dead simple:
- Start a conversation with @k4pi_bot
- Send your photos
- Add a price and brief description
- Your listing is live
For buyers, it's equally straightforward. Search by text, browse by category, or just scroll recent listings. Everything happens inside Telegram. No redirects. No external websites.
What Actually Works Better
After running this for a while, a few things stand out:
Notifications are natural. When someone is interested in your item, you get a Telegram message. Not an email you might ignore. Not a push notification from an app you might have muted. A message in the same thread where your friends are texting you.
Photos work the way you expect. Telegram handles image compression and display beautifully. You can send multiple photos, buyers can zoom in, everything just works. Compare this to OLX where uploading multiple photos feels like filing a tax return.
Trust is higher. Buyers and sellers can see each other's Telegram profiles. You're not interacting with anonymous usernames — you're interacting with real accounts that have profile pictures, join dates, and mutual contacts if you happen to share groups. That social graph reduces scam anxiety significantly.
Speed is real. No page loads. No switching apps. The whole experience lives in one place, and Telegram's infrastructure is genuinely fast even on mediocre mobile connections.
The Honest Limitations
I'm not going to pretend this is perfect.
Telegram bots have interaction limits. You can't, for example, show a beautiful grid layout the way a native app can. Everything is constrained by what the Telegram bot API allows — which is powerful but not infinite.
Discovery is also a challenge. If buyers don't know @k4pi_bot exists, they can't find it by browsing. You still need word of mouth or community sharing to grow. This is different from a marketplace website that gets SEO traffic.
And Telegram's API has quirks. Building on someone else's platform always means accepting their constraints and occasionally scrambling when something changes.
But here's my honest assessment: for peer-to-peer selling in markets where Telegram is already dominant, these tradeoffs are worth it. The activation energy for buyers is so low that it compensates for a lot of UI limitations.
The Broader Pattern
What I built is one instance of a broader pattern that I think is underexplored.
A lot of "we need an app for this" problems are actually "we need a bot for this" problems. Especially in markets where smartphone storage is precious and users are selective about which apps they install. Especially when the use case is occasional rather than daily. Especially when your target users are already organized into communities on a messaging platform.
The Telegram marketplace isn't trying to compete with Amazon. It's trying to serve the guy who wants to sell his old motorcycle to someone in the same city, without either of them downloading a new app, creating a new account, or navigating a platform built for a different market.
That's a real problem. And a chat-native bot solves it more elegantly than a classifieds website does.
Try It Yourself
If you're curious what this looks like in practice — or if you're in a city where the bot is active and want to buy or sell something — the bot is live and free to use.
Start a conversation with @k4pi_bot on Telegram
If you're a developer thinking about building something similar: the core insight is that Telegram's bot platform is dramatically underutilized for commerce use cases. The infrastructure is solid, the user base is massive in the right markets, and the friction reduction for buyers is a real competitive advantage.
The classifieds space is fragmented and frustrating almost everywhere outside North America. That's not a complaint — it's an opportunity.
Built with Python, aiogram 3, PostgreSQL, Redis, Elasticsearch, and a lot of curiosity about whether chat-native commerce could actually work. Spoiler: it can.
Top comments (0)