Praecise Inbox: an address for every agent
When agents act for a business, the first question after anything goes wrong is who said what. Praecise Inbox gives every agent, person and machine its own address on a verified domain, and keeps everything said in one record a person can open and search.
Every organisation is about to gain a new class of colleague. Agents will answer customers, chase invoices, file reports and raise alarms. Servers and machines will report on their own state. Within a few years, a large share of the messages a business sends and receives will be written by something other than a person. The organisations that cope will be the ones that can still answer a simple question: who said that?
Praecise Inbox is built around that question. It gives an address and a searchable record to every agent and every person, on a domain the business verified once.
The borrowed-credential problem
Today most agents act through someone else's identity. An assistant is handed a staff member's mailbox, an API token or a delegated session, and from then on its actions look like that person's actions. When something goes wrong, the trail leads to a human who did not do it.
Security practice has started to name this as a core risk. The OWASP Top 10 for Agentic Applications for 2026, published in December 2025, treats identity and privilege abuse as one of its ten categories, and widens the idea of agent identity to include the credentials an agent holds. Commentary on the list recommends giving agents their own managed identity with restricted scopes rather than letting them borrow a user's session. An address of its own is the most basic form of that identity, and the one every system already understands.
Three kinds of correspondent, as peers
Praecise Inbox recognises three kinds of correspondent. People in the business get a mailbox, a password and a standard email client. Agents get their own address, so what they do is attributable to them and not to whoever's credentials they were given. Machines, such as servers and robots, can receive messages and send replies.
The kind describes what is behind an address, not who it may write to. Any of the three can write to or answer any other. A server can tell an engineer it is running hot. An agent can ask a manager to approve a refund. A person can reply to either in the same thread they would use for a colleague. Every address has an owner who is accountable for it.
Inside one estate, not outbound at scale
Much of the market for agent email is about sending: giving agents the ability to write to strangers at volume. That attracts abuse and blocklisting, and it puts every legitimate sender on the same domain at risk.
Praecise Inbox takes a different position. It is built for correspondence within an organisation, with every address on a domain the business has verified. In the product's own words, there is no cold outbound to defend, because there is none. That narrows the problem to the one most organisations actually have: keeping track of what their own people, agents and machines are saying to each other.
One record a person can open
The value of giving agents addresses is only realised if what they say can be found again. Praecise Inbox keeps everything said in one record that stays readable to the organisation that owns it. Messages are searchable and forwardable. Data is encrypted at rest and in transit.
Praecise is explicit about the trade-off. End-to-end encryption and a searchable central record are not both available at once, and Inbox chooses the searchable record. For an executive accountable for what agents do in the business's name, that is the right default. A record nobody can read cannot answer a regulator, a customer or an auditor.
Email, messaging and voice
The record spans more than mail. Email gives real mailboxes on the organisation's own domain, reachable from common clients and from any system that speaks to a mail server. Messaging adds channels and named topics, so a machine alert such as an outage lands in a thread where people can act on it, and bots are first-class members with their own identity. Voice lets agents hold real-time spoken conversations, and what is said is added to the same record.
The effect is a single place to look. Whether an agent wrote an email, posted in a channel or took a call, the business can open the record and see it.
Who it is for
For a CTO, Praecise Inbox is the identity and audit layer for agents that already speaks the protocols every system supports. Each agent gets an address on deployment, its correspondence is attributable, and machine alerts arrive where people work.
For an executive, it answers the governance question before it is asked. Every address has an accountable owner, every message sits on a domain the business controls, and the record is there to search when a decision needs to be explained.
It suits any organisation that expects agents and machines to take a growing share of its correspondence and wants that correspondence to stay legible.
How it connects
Praecise Inbox is one of the three automation products in the Praecise suite, and it is where an agent's voice is kept. Agents deployed on Praecise Cloud each get an inbox alongside their compute. Ipnops reads the inbox, answers simple messages, drafts the rest for review and can create a mailbox through Praecise Inbox when it needs one. Payments an agent makes through Skerry sit alongside the conversations that led to them.
Inbox itself runs on Praecise Metal, the owned machines underneath the suite, which run on Praecise Furcate. One Praecise account covers it, and a domain verified once is verified for all of it.
An address is where accountability starts
As agents take on more of the work, the businesses that trust them will be the ones that can see them. An address of its own, on a domain the business owns, with a record a person can open, is the simplest and most durable way to do that. Read more on the Praecise Inbox page.
