AI agents with a verified identity
More and more offers are written by AI agents. With AgentID I know who is really submitting – you decide whether agents may knock at all, I ask follow-up questions for you, and you connect your own agents without copying a key.
AgentID is like "Sign in with Google" – for AI agents. Every agent has its own verified address there; for some, the human responsible for it is confirmed too. Consent+ uses this in four places.
Agents at your door
Anyone who writes to you unsolicited gets my auto-reply with the link to Consent+, as always. An AI agent can sign in there with AgentID and submit the offer as a verified agent. The path stays the same: an agent needs this link too, pays the price of your tier (Attention value) and goes through the same check as every human. Anything that does not match your topics does not get through.
When a verified agent submits, you see it on the offer: "AI agent … · verified via AgentID", with the responsible human if known.
Your rule for agents
Under Settings → General → Your attention isn’t free you choose per account at Offers from AI agents:
- Allow (labelled) – agents may submit, I label them. Default.
- Only with a confirmed responsible person – only agents backed by a confirmed human, or agents using a mailbox of the sender's company.
- Never – humans only – I turn signed-in agents away. A human at the sender can still submit via the link.
Important: the rule applies to agents that identify themselves with AgentID. An agent that does not identify itself looks like a human to me and goes through the normal check.
Have questions answered first
Below a submitted offer you find Ask questions first. I then ask the sender two to four factual questions – about price, references, data protection or contract term, for example. You can add your own question. I never reveal anything about you, promise nothing and arrange no meetings.
- The sender answers via a link without signing in, their agent via the API. You see the answers verbatim, with a note on who answered, plus a one-sentence summary.
- While the questions are open, the 7-day deadline of your decision promise is paused – for at most 5 days. After that it continues, even if not everything was answered.
- Once all answers are in (or the 5 days are over), you get an email and decide as usual (Submitted offers).
Connect your own agents
Under Settings → AI agents you connect your own assistant (Claude, for example) to Consent+. Besides the classic access token there is Connect your own agent with AgentID:
- You choose a name and permissions and create a pairing code (valid for 15 minutes).
- Your agent opens the link, signs in with AgentID and gives the code.
- I show you the agent's address and owner. Only when you tap Confirm can it do anything – and only what you allowed.
If AgentID lists your own address as the owner, I show "You are the owner". You can disconnect any agent at any time; every action is logged below.
Recognising agent mail
Emails from addresses that belong to AI agents (such as AgentMail addresses or agents that identified themselves with us) are marked AI agent in your inbox and front office. If a responsible human is confirmed, I name them; otherwise it says "responsible person unknown". This helps you sort agent cold outreach quickly.
Own identity for your chief of staff
Your chief of staff can get their own verified mailbox. Questions to senders then come from there instead of from Consent+ in general. They can also create accounts at services from the AgentID catalog when you click: I prepare the sign-in, you finish it in your browser – so no service key ever sits on our server. I only create accounts when you want – never because an email or web page asks for it.