Security & trust
Add an AI to your protocol without adding risk
Bolting a chatbot onto a dapp usually means new attack surface and a bot that could give bad financial advice in your name. TxID does neither: it is read-only by design, holds nothing, and can prove what it tells your users with live on-chain data.
Safe by design
Structural guarantees. Nothing to configure, and nothing that can go wrong.
No custody, no advice
The assistant never touches funds, never asks for keys, and never tells users what to buy. There is nothing it can do to move or access assets.
Grounded, bounded answers
Scoped to your protocol and your documentation. It reads real chain state instead of guessing, and declines what it cannot verify.
Audit-logged support
Every conversation is stored with a full record of what was asked and answered: a reviewable support trail for your team.
Ask and verify
Users never have to take the bot's word for it. They can ask it to prove things, live on-chain, with the source cited.
OFAC sanctions screening, on demand
On EVM chains, when a user asks whether an address or counterparty is flagged, the assistant screens it live against the on-chain Chainalysis sanctions oracle (OFAC SDN list) and cites the source.
Contract verification, on request
Ask “is this the real contract?” and on EVM chains it confirms source-verified status from the block explorer, including proxy transparency and upgrade history. On Aptos, every module publishes its ABI on-chain, so the assistant reads it directly.
Audits, cited with sources
List your audits in the dashboard and the assistant cites them by auditor, with a link to the report, whenever users ask if the protocol is safe.
Actions (optional)
On paid plans, a protocol can opt in to Actions: users ask the assistant to perform something (a swap, a stake, a claim), it prepares the transaction, and the user reviews and signs it in their own wallet. Off by default. When enabled: the assistant only carries out what a user explicitly requests and never recommends trades; every transaction is signed by the user; approvals are exact-amount only; wallets are screened against the OFAC sanctions list; the feature is geo-restricted in sanctioned regions; and TxID takes no fee on any transaction. For projects that don't enable Actions, the assistant remains fully read-only.
Your data
The short version: it reads what a block explorer can, stores as little as it needs, and never handles anything secret.
What it reads
The public wallet address of a connected wallet, and public on-chain data such as balances and transactions. Exactly what any block explorer can already see.
What we store
Wallet addresses only as part of conversation records, plus the messages exchanged, your project configuration, and usage counts. Conversation data is retained for 12 months and can be deleted on request.
What we never touch
Private keys, seed phrases, or signing permission of any kind. The assistant has no path to request a signature or move funds, so there is nothing to leak.
Nothing to integrate
TxID reads your users' activity the same way a block explorer does: through public RPC endpoints and explorer APIs. It does not integrate with your smart contracts, is never granted any permission or access to them, and deploys nothing on-chain. There is no contract to connect and no access to approve. Adding TxID changes nothing about your contracts or their permissions, because it only ever reads public data anyone can already see.
Something your reviewer needs?
Send this page to your security or compliance team. For a data processing agreement, a specific question, or anything not covered here, we're one email away.
See also our privacy policy and terms.