Why Telegram Chatbots are the Secret Weapon for Crypto & Web3

What a Telegram chatbot does for a Web3 community

A Telegram chatbot for a crypto or Web3 project answers the questions new members ask every hour (contract address, vesting schedule, how to join the whitelist, where the docs are) from your approved documents, collects whitelist and support details in a private chat, and alerts a moderator when something needs a person. It is for teams running token launches, NFT collections, DeFi protocols, exchanges and wallets whose community lives in a Telegram group that never sleeps.

The decision is which jobs to give a moderation bot, which to give an AI agent, which questions the agent must refuse, and how to run intake without creating a privacy problem. This guide covers each, with a cost example built on Botcadence pricing. For the channel itself, see the Telegram chatbot page.

Moderation bots vs AI support bots: what is the difference?

Two different tools get called "the Telegram bot". Keep them separate in your plan.

Botcadence provides the second column: a Telegram customer support agent and a Telegram lead generation agent. Pair it with whatever moderation bot your admins already trust for the first column. We build Botcadence, so we are not neutral on the second column; on the first, use the tool your moderators know.

Which questions should the bot answer, and which must it refuse?

Grounding is the whole game in crypto. A chatbot that improvises a contract address is a security incident, not a bug.

Botcadence document training restricts answers to what you upload, and refusal wording is set in the agent's instructions. The hallucination glossary entry explains why ungrounded answers happen and how grounding limits them.

How to run whitelist and airdrop intake in DMs

Asking members to paste wallet addresses into the main group is a privacy risk and a scammer's harvest. Move intake to a private chat with the bot.

1. Announce the campaign in the group with the bot's handle and the exact steps.
2. In DM, the agent confirms the eligibility questions you define (for example, the campaign code or an account email), then asks for the wallet address once.
3. The agent repeats the address back for confirmation; one wrong character is a lost allocation.
4. Records go to Google Sheets, your CRM through Zapier or a webhook, never to the group.
5. The bot never asks for a seed phrase, private key or signing transaction, and says so in every intake conversation.

What the intake record should hold

Keep the record small and consistent, because the allocation script downstream will read it.

Export the sheet once at the cut-off, deduplicate on wallet address, and have a person review any row the agent flagged as uncertain. On-chain checks (holding a specific NFT, a minimum balance) are not something the chat agent does by itself; if you need them, they run in your backend and the agent asks for the result through a webhook. Keep that distinction clear in the campaign plan.

What does a Telegram agent cost for a large community?

Botcadence chat plans run Free (web chat and WhatsApp), Starter $39, Growth $99 and Business $199 a month, with Enterprise custom; which plan includes the Telegram channel and how many replies it carries is on the pricing page.

A worked example, with every input labelled as an assumption. Say a 50,000-member group produces 250 answerable questions a day, 7,500 a month. Say a moderator takes 2 minutes to find and type each answer: 250 hours of moderator time a month, or roughly 1.5 full-time people across time zones. Say the agent answers 70 percent of those from the litepaper and FAQ: 5,250 automated, 2,250 escalated or ignored as chatter, and moderator time falls to about 75 hours. Compare that against the plan price for the reply volume, and against what the team currently pays for round-the-clock moderation. Outcomes depend on how complete your documents are and how much of the chatter is actually a question.

Staffing a launch window

During a token launch or mint, questions can arrive many times faster than on a normal day, and most of them are the same three. Pin the answers, let the agent repeat them, and put two moderators on the escalation queue in shifts for the first 48 hours. The agent's job in that window is to absorb the repeat questions so that the moderators only see the wallet problems and the reports of impersonation that need a person.

Multilingual communities

Train the agent in one primary language, usually English, and let it reply in the member's language. Botcadence chat agents can reply in the member's language, across 50+ languages, from the same approved documents; see multilingual support. The contract address and refusal wording should be identical in every language, so test them in the three or four languages your community actually uses.

Choose a Telegram AI agent when, do not when

Choose it when the same questions arrive hourly, when the community spans time zones your moderators cannot cover, when intake currently happens in the open group, or when wrong answers from well-meaning members have caused problems before.

Do not deploy it to make moderation decisions, to give investment guidance, or before your documents are current. An agent trained on an outdated litepaper will state the old vesting schedule with confidence. Compare Telegram with other messaging channels in Telegram chatbot vs WhatsApp Business and read the Telegram business chatbot guide for setup details.

What to measure

• Questions answered by the agent per day, and the share escalated to a moderator.
• Refusals: how many price or advice prompts were declined, reviewed weekly for wording.
• Contract-address answers: audit every one against the pinned document.
• Whitelist records with a confirmed address versus abandoned intakes.
• Moderator response time to escalations, by time zone.
• Scam reports from members, before and after launch, as a trend rather than a promise.

Common mistakes

• Letting the agent read contract addresses from the group's message history instead of the pinned document.
• Running intake in the main chat.
• No refusal wording for financial advice, so the agent hedges instead of declining.
• Treating the agent as a moderator; it answers, it does not police.
• Forgetting to update the litepaper after a tokenomics change.
• No human on call during a launch window, when escalations spike.

Botcadence human handoff sends the conversation to a moderator with full context, which is the part that keeps a launch-day spike manageable. The Botcadence chat agents page covers the Telegram channel alongside website, WhatsApp and the other messaging channels a Web3 team runs from one knowledge base.