Two arrangements, for two different fears. Multisig splits the power to spend across several keys, so losing one device loses nothing. A guarded wallet keeps one key in charge but caps what it can move alone, so a stolen phone cannot empty the wallet. This is what each one actually is, how to set it up, and which fear each one answers.
An m-of-n wallet is n extended public keys and a threshold. Nothing else. There is no account somewhere holding it, and no server that knows about it — the wallet exists because those n keys and that number exist.
Each address is built by taking every co-signer's key, deriving its child at the same index, and writing them into a script that says "any m of these may spend":
OP_2 <pubkey A> <pubkey B> <pubkey C> OP_3 OP_CHECKMULTISIG
That script is hashed into a P2SH address. Two consequences worth knowing, because both surprise people:
Any of them, from anywhere. A co-signer key is an ordinary account xpub, so it can be one Aegis mints for the purpose, or one exported from another wallet on another device — Electron Cash, Bitcoin.com, a second Aegis. Aegis will show you its own under Show account xpub.
Aegis can hold all of the co-signers, some, or none:
Do not reuse your everyday wallet's xpub. An xpub lets anyone holding it derive and watch every address of that account. Hand your spending account's xpub to two co-signers and you have given them a permanent view of everything you do with it — and one compromised key now costs you both wallets. Use a key made for the purpose.
Spending then needs m signatures, and getting the half-signed transaction between devices is its own problem — solved by the co-signing protocol, which needs no pairing step and no server.
Multisig is all-or-nothing: either every payment needs two devices, or none does. That is the wrong shape for a wallet you actually spend from, where small payments should be easy and large ones should not be.
A guarded wallet has an owner key, a guardian key and a limit. Under the limit, the owner spends alone. Over it, the guardian has to sign too. The rule is a covenant, so it is the chain that enforces it — not Aegis deciding whether to allow the send, which malware on the same machine could simply skip.
Three details that change what you can do with it:
They answer different fears, and the right question is which failure you are protecting against.
A 2-of-3 survives any one device dying, being stolen or being forgotten. The cost is that every payment needs two devices present, which is why it suits savings far better than daily spending.
Someone with your phone can take the limit and no more. The cost is that losing the owner key still needs the guardian to recover from, so it bounds theft rather than surviving loss.
Neither of these can be rebuilt from a seed phrase alone, and this is the part people get wrong. A seed gives you back your key. It does not tell anyone what the other keys were, and without all of them the money cannot be found, let alone spent.
For a multisig wallet, keep all n extended public keys and the threshold.
For a guarded wallet, keep both public keys and the limit.
Keep them wherever you keep the seed. They are public — they reveal the wallet's addresses to anyone who reads them, but they can never spend from it.