Inside one company, a first name is enough. You say, "Sven, take a look at this" or "Kairi, prepare the proposal," and everyone knows who you mean. Sven helps the owner keep track of the business. Kairi handles clients and sales.
Now imagine hundreds of companies using Studio, each with its own Sven and Kairi. They read emails, prepare proposals, ask partners for information and help move agreed work forward. Eventually, one company's Kairi sends a message to another company's Sven.
A signature that says "Kairi" suddenly feels incomplete. Which company does she represent? May she ask for a price, or may she also confirm an order? Who gave her that authority?
That is why Estonia's idea of digital identities for agents feels so natural to us. This becomes a practical question as soon as a digital colleague starts dealing with others on a company's behalf.
What is Estonia actually planning?
On 17 June 2026, the government announced that the Eesti.ai advisory board supported developing digital identities for AI agents. The aim is to let agents act for a person, company or institution with defined, verifiable limits. Government announcement, 17 June 2026.
Euronews brought the story to an international audience with its headline about AI ID codes. Computerworld and The Register also covered the initiative. Euronews, 19 June 2026.
In September, RIA's Aruait project lead clarified to ERR that agents would not receive human personal identification codes. The essential requirement is linking an agent's actions to the person it represents. ERR's clarification.
The ministry's 24 September update describes a digital identifier, draft legislation and planned trials. Agents would gain no independent rights. Service providers could set their own conditions for admitting them. This is a published plan, rather than evidence of a service already open to everyone. Ministry update, 24 September 2026.
For us, the value is practical: the other party should be able to verify who is acting and the authority behind that action.
The company as the agent's surname
A Studio agent already works in a company context. Kairi uses that company's client information, pricing and agreements. Another company's Kairi has her own workspace, information and owner. The shared first name describes the role; the company tells you where she belongs.
For a person reading a message, we could express this simply: "Kairi · Kase Construction" or "Sven · Saare Design." These are fictional businesses, but the principle is familiar. The company name becomes a kind of surname.
Systems communicating with each other need more. A name can be copied, a business can change its name, and one company might eventually have several agents in the same role. We therefore see a need for a persistent, verifiable agent identifier linked to the company it represents.
The first name tells you who you are talking to. The company tells you whom they represent. The mandate tells you what they may do.
An ID needs a mandate beside it
Kairi's ability to prepare an order does not automatically give her permission to confirm it. Representing a business involves steps with very different consequences: asking for information, preparing a quote, committing to a price, placing an order and making a payment.
We think the other party should be able to verify which of those steps an agent is authorised to take. Permission should cover a defined purpose, relevant data, financial limits and an expiry. The person granting it should be able to withdraw it.
Agent: Kairi · Kase Construction. A verifiable agent identifier links her to this company.
Task: obtain materials quotes for a particular project.
Allowed: share the quantities and delivery address for this inquiry, and receive offers.
Confirming an order and paying: require separate human approval.
Valid until Friday; the person granting permission can revoke it sooner.
The useful part is the boundary. The supplier knows which company is requesting the quote. Kairi does not expose the builder's whole client database, and permission to ask does not quietly turn into permission to buy.
When one company's agent talks to another
Imagine the next step. A builder's Sven is asked to find an installation partner for a project. The service provider's Kairi receives the inquiry, clarifies the scope and prepares a response. The people involved review the price, timing and terms before confirming the agreement.
If this becomes ordinary business, verifiable information about the parties needs to travel with the conversation. The buyer's agent must know which company stands behind an offer. The seller's agent must distinguish a question, an expression of interest and an authorised acceptance.
We are thinking about this opportunity as we build Studio Network: helping businesses find each other and move from an inquiry to a concrete proposal. A shared approach to agent identity could make those exchanges more trustworthy across different software providers.
RIA's Aruait project includes limited, revocable authority, an agent trust registry and secure cooperation between agents among its development goals. RIA: Aruait / Reason Reserve.
An identifier helps us trace responsibility
A correctly identified agent can still make a mistake. A quote might contain the wrong quantity or a delivery date the business cannot meet. Alongside identity, we want a readable record of what happened: the information used, the authority granted and the person who approved a consequential step.
As a business owner, I want to see what happened in my company's name and stop further action when necessary. As a trading partner, I want to know whom to contact when something is unclear or wrong. An agent's name and identifier should help me get there.
Studio already has company-specific workspaces, agent roles, activity history and approval workflows. Connecting to a national agent identity system is a future possibility. We are not claiming that Sven or Kairi currently has a state-issued ID or can use one to act in a bank or government portal.
So why not?
As a company's digital team takes on more real business administration, its authority to represent that company needs to be understandable and verifiable by others. That feels like a natural next step.
Sven can remain Sven to us. But when there are hundreds of Svens, we also want his surname: the company he belongs to. And when he agrees to something, we want to see the authority to do so.
We think tomorrow's Sven needs all three: a familiar name, a verifiable identity and a clearly bounded right to act.
Further reading
Sources reviewed on 12 October 2026. The account of Estonia's initiative follows the announcements linked above; the agent surnames and business scenarios express our own view of Studio's future.
Computerworld · Estonia's plans for government-backed agent identities, 17 June 2026
The Register · Agent identities, permissions and responsibility, 18 June 2026