Design
Telegram-Native Multiplayer
Contract cards that are real instruments, push alerts that arrive where players already are, and a 1.5 MB cold-start budget that disciplines everything.
Most games that ship on Telegram are ports. The interesting question is what changes when you design for it.
The deal room already exists
Trading in persistent-economy games has always happened in chat. Sim Companies has forums, EVE has out-of-game channels, every serious economy game has a Discord where the actual negotiation occurs. The game sees a transaction; the conversation that produced it happened somewhere the game cannot reach.
Running inside Telegram collapses that. A B2B contract is shared as an interactive card inside a group chat: parties, SKU, quality tier, quantity, delivery interval, unit price, duration. A buyer taps to review and accepts or counters without leaving the conversation.
The card is not a screenshot of a contract. It is the instrument. A group chat becomes a trading floor because the floor and the chat are the same surface.
Deep links carry a start parameter, so tapping a card in a group lands a buyer on exactly that contract after signature validation. Contract viewers must still be a party to it — posting a card publicly does not expose its terms to everyone who sees the message.
Alerts where the player is
Push notifications in a game app compete with every other app's notifications and lose. A message in Telegram arrives in the same place as the player's conversations.
The alert set is deliberately operational: filled orders, cargo arrivals, commodity price swings over 10%, perishable spoilage warnings, and loan due notices. Each is actionable and each has a deep link to the exact screen that acts on it.
The spoilage warning is the one that matters most. Five products expire within 24 to 72 hours and a sweeper runs hourly. A notification that arrives two hours before milk spoils is worth real money, and it only works because it reaches a player who is not currently playing.
Zero install
The funnel from a shared link to a first retail sale has no install step. That is the single largest structural advantage, and it sets the design target: a first sale within 15 seconds of founding a company in a guided run, and a first collection within 180 seconds.
There is no captive tutorial moment in a game entered from a chat message. Guidance therefore lives in the factory view, next to the decision, and can be minimised or skipped. One visible decision at a time.
The budget
The constraint that disciplines everything: a sub-1.5 MB gzipped cold start for the Mini App.
That rules out a great deal. No heavyweight component library shipped to every player on first paint. Map rendering is lightweight vector with GeoJSON coastlines, not tiled imagery. The isometric plot view runs on a 2D renderer sized for the budget rather than a general-purpose engine.
It also shapes what the server does. The client should not need to download a model of the whole economy to show you three numbers. The executive health card, the market station, the freight station and the contract card are each a small payload against a small API surface.
A budget stated in megabytes is more useful than one stated as "fast", because it can be measured on every build and it fails loudly.
Accessibility is not optional here
A Mini App runs inside somebody else's chrome, on a phone, at whatever text size they have chosen. The web shell allows zoom, respects reduced-motion preferences, keeps visible keyboard focus, and uses text rather than colour alone for prices, state and errors. Plot buttons are keyboard-reachable, which gives a second route into the factory inspector for anyone who cannot reliably hit a small isometric target.
The playtest targets include keyboard-only play and a 360-pixel viewport, with screen-reader focus checked after navigation.
What it does not solve
Telegram gives distribution and a social layer. It does not give an economy. Every structural problem — the faucet, the wage brake, the price collar, the ledger — is the same problem it would be anywhere, and solving it is most of the work.
The platform is a very good answer to "how do players find each other and talk". It is not an answer to "what are they trading, and why is it worth anything".