Typical MVP delivery — 7 days/Priced and agreed before it starts/Taipei · Eindhoven · Remote/You own the code and the cloud account/Reply within 1 business day/
All work
Web3 / 2021

Seedit

A multi-chain crypto tipping bot for Telegram. Every community member gets a custodial wallet, and tipping, airdrops, and peer-to-peer trading happen with a single chat command, across TRON, Ontology, Binance DEX, and Solana, with over a million wallets created.

1M+
Wallets created across the network
4
Blockchains supported from one codebase
Single command
Tip, airdrop, or trade, straight from chat

Overview

Seedit is a multi-chain crypto tipping bot for Telegram. It gives every member of a community a custodial wallet, quietly, the first time they interact, and turns moving tokens around into a single chat gesture: reply to a message, type /tip 5 SEED, done. On top of that base it adds airdrops to active members, randomised "rain" giveaways, and peer-to-peer token trading, all without anyone leaving the group.

The problem it solves is friction. Crypto communities want to reward good posts and onboard newcomers with small amounts of tokens, but ordinary on-chain transfers mean wallets, addresses, gas, and waiting, which is far too heavy for a casual "here's a tip." Seedit collapses that into chat, and does it across four blockchains from a single codebase. More than a million wallets have been created through it.

The product

At its core Seedit is a wallet and a set of social money primitives wrapped in a conversational interface. A user never has to think about key management or network selection. The bot provisions an account per chain on demand and presents balances, deposits, and transfers as plain chat interactions, complete with QR codes for deposit addresses and in-place message edits that turn "tipping…" into "tipped … [tx]" as the transaction confirms.

Around tipping sit the community mechanics. /airdrop splits an amount evenly across recently active members; /rain distributes it randomly for a bit of fun; admins get locking controls to keep things orderly. A hashtag-driven trading system lets members post buy and sell offers (#buy … for …) that others confirm with a single command, with offers expiring automatically so the order book stays fresh.

Architecture

The codebase is a monorepo split along its most important seam: a chain-agnostic core package that owns wallets, balances, models, and services, and a thin Telegram package that owns commands and chat UX. Every blockchain implements one common interface (createAccount, getBalance, transfer, and a set of pre-flight checks) and a service registry maps each network to its implementation. Adding a new chain means writing a single service class; the Telegram commands never change. That decoupling is what let one codebase grow to support TRON, Ontology, Binance DEX, and Solana without the bot logic fragmenting per chain.

State is persisted in PostgreSQL through Sequelize, and the whole stack runs from Docker Compose with Postgres, Redis and RabbitMQ, so a contributor can bring up the full environment with one command.

Safety and correctness

Moving real value through a chat bot is unforgiving, so the design leans hard on guardrails. A per-chat work queue serialises commands for a given group, so two people tipping at once can't race on the same wallet or balance. Every transfer runs through a safeTransfer path that pre-checks sufficient balance and network fees before broadcasting and returns a structured success-or-error result rather than throwing an opaque failure into the chat. Token math runs on bignumber.js to avoid floating-point drift, and a fuzzy matcher quietly corrects mistyped token symbols so a slip of the keyboard never sends the wrong asset.

Outcome

Seedit grew into a multi-chain piece of community infrastructure, with four live blockchains behind a single chat interface, with more than a million wallets created across the network. One clean line between the platform layer and the blockchain layer is what let the project absorb entirely new chains as drop-in additions rather than rewrites.

Next case

Companion AI

Read it
/ Start

Tell us the one thing it has to do.

If it's sharp enough to prove in a week, we'll say so and give you a date. If it isn't, we'll tell you that too, along with what we'd build first instead.