Definition
Agent Authority is the deliberately granted, bounded, and accounted-for mandate a brand gives its agents.
Granted means: decided, not drifted into. Bounded means: it is defined what the agent may promise and where a human takes over. Accounted for means: there is an address that stands behind it.
Distinctions. Not the same as access rights, which govern what a system may technically reach. Not the same as AI governance, which orders how an entire company deals with such systems. Agent Authority answers a narrower question: what may an agent promise bindingly in the brand's name, and who stands behind it? It is the counterpart to the Machine Readable Brand. One makes the brand legible outward, the other binding inward.
Why the question exists. A tribunal answered it before the industry had asked it: Air Canada is liable for what its chatbot promised. The evidence, with its source, is in the evidence collection.
The expression comes from the law of agency. Carried into the brand context and used in this sense by Robert Haase, since July 2026; older uses of the word pair in other contexts exist. Related terms: Accountable Decision · Accountability · Agentic Brand Readiness.
A Chatbot Before the Tribunal
When Air Canada's chatbot promised a customer a refund policy that did not exist, the airline argued before the tribunal that it could not be held liable for information provided by its agents, servants or representatives, the chatbot included. The tribunal member noted that Air Canada was in effect suggesting the chatbot was a separate legal entity, responsible for its own actions.
The tribunal called that defense, verbatim, "a remarkable submission" and rejected it: the chatbot is part of the website, and the brand stands behind everything promised there. The amount in dispute was a few hundred dollars. The question behind it is considerably larger, and most brands have never asked it: Who is actually acting in our name, and what is he allowed to do?
Brands Have Started to Act
Agents no longer just answer. They refund, book, grant, and negotiate. In doing so, brands are effectively granting a mandate, most of them without ever deciding to.
The cases are piling up, and they share a pattern. Cursor's support agent invented a company policy that never existed, operating under a human name; users cancelled their subscriptions, and co-founder Michael Truell apologised publicly. A Chevrolet dealer's chatbot confirmed the sale of a new Tahoe for one dollar and called it "a legally binding offer", after a user had fed it that exact phrase; the dealership never honoured the sale. And Klarna, which in February 2024 had publicly celebrated its AI assistant for doing the work of 700 full-time service agents, announced a good year later that it would recruit human agents again. Chief executive Sebastian Siemiatkowski told Bloomberg in May 2025 that cost had unfortunately been too dominant a factor in how this was organised, and that the result was lower quality. Nobody was made redundant for the 700: Klarna's headcount fell from 5,527 to 3,422 full-time positions between the end of 2022 and the end of 2024, by the company's own account through AI, a hiring freeze and natural attrition.
The pattern: an agent is not a channel. Channels transmit what the brand has decided. Agents decide themselves, a thousand times in parallel, in the brand's name. German business language has a precise word for this relationship: Prokura, the registered authority to sign for the company. Legally it is tightly regulated, entered in the commercial register, bounded. What brands are currently granting their agents is Prokura without a contract: a mandate whose scope nobody has defined.
This is exactly what brand leadership needs a new discipline for, and I call it Agent Authority: the deliberately granted, bounded, and accounted-for mandate a brand gives its agents. Granted means: decided, not drifted into. Bounded means: it is defined what the agent may promise, grant, and disclose, and where a human takes over. Accounted for means: there is an address that stands behind its actions. Air Canada has demonstrated what the alternative looks like.
Go-to-Market: There Is an Agent on the Other Side, Too
While brands field agents, the other side already buys through its own. When mandate meets mandate, go-to-market gets redistributed.
The buying side is delegating faster than the brand side. Bain surveyed US consumers in November 2025: 17 percent of online shoppers said they would begin their holiday shopping on an AI platform such as ChatGPT or Perplexity, and 30 to 45 percent were already using generative AI for product research and comparison. The two figures measure different things and cannot be added up; the same survey records that around half of respondents remain cautious about letting an AI handle a purchase from start to finish. Gartner, surveying US consumers separately, found that the purchase decision itself is one only a small minority would give away. Awareness, consideration, and purchase still collapse into a single conversation. In Bain's flight-booking tests, agents reached the airline's own website directly in only about five percent of cases: the direct channel and the loyalty program, the most expensive assets many brands own, are structurally bypassed in the agent layer. And what the agent weighs up and discards along the way, the merchant never sees. It sees the customer only once the decision has been made.
What the transaction itself will look like was settled in 2026 in a remarkably fast experiment. OpenAI launched purchasing directly in the chat and rolled it back a good five months later, with the source given in the evidence collection. Walmart put two opposing figures on it, both self-reported: Daniel Danker, who oversees product and design at Walmart, told Wired in March 2026 that items purchasable directly inside the chat converted at one third the rate of items requiring a click through to walmart.com, while ChatGPT brought Walmart new customers at about twice the rate of search engines. Discovery happens at the agent, purchase happens at the brand, for now. In parallel, Google's Universal Commerce Protocol is building the infrastructure meant to make agent-led purchasing the standard, carried by more than twenty partners from Shopify to Mastercard.
In B2B, the mandate already has numbers attached, though only ones supplied by the vendor and the operator. Walmart has negotiations with its smaller suppliers run by a system from the vendor Pactum; according to the procurement researchers Remko Van Hoek and Mary Lacity, such a bot can run 2,000 negotiations at the same time. Walmart put average savings at around three percent in 2023 and said three out of four suppliers would rather negotiate with the machine than with a person; the 68 percent closing rate comes from an earlier Harvard Business Review case study. None of these figures has been independently audited, and Walmart says it has more than 100,000 suppliers; the sources are in the evidence collection. Forrester expects one in five B2B sellers to be forced into agent-led quote negotiations in 2026. The consequence for sales is uncomfortably concrete: product data, prices, and terms are the new sales collateral, because they are the only thing the other side's agent reads. An agent can read a PDF price list. What it cannot do with one is compare, check and negotiate.
Honesty requires the order of magnitude, and that depends on what is being counted. Visits where someone clicks a link inside an AI assistant grew on US retail sites by 4,700 percent year on year in July 2025 on Adobe's measurement, and by 138 percent in May 2026. The share stays small, and the measurements are far apart: across more than 99 billion sessions in the fourth quarter of 2025, Contentsquare puts it at 0.2 percent of all visits; Semrush, across more than 50,000 websites for the year 2025, at less than 0.15 percent; Conductor, across 3.3 billion sessions from May to September 2025, at 1.08 percent. All three are lower bounds, because part of this traffic arrives without a referrer and is counted as direct. For 2030, Morgan Stanley puts agentic commerce at 10 to 20 percent of US online retail and Bain at 15 to 25 percent, both in December 2025; for retail outside the US there is no comparable figure. Early or not, the window is not open indefinitely, because the loyalty mechanics are shifting structurally. Gartner put it in one sentence back in early 2022: "You can't wine and dine a machine to win its loyalty, and you don't need to." A customer's agent does not bond with experiences. It bonds with suppliers whose promises are machine-readable and hold.
Authenticity: The New Trademark Problem
Once agents speak for brands, the next question is unavoidable: How does anyone know this agent is real?
Today the answer is: often not at all. In 2024 the British consumer organisation Which? found fake support accounts on X for every major airline operating in the UK and recorded that they are often quicker to reply than the real ones; two fakes answered a test enquiry to Wizz Air almost immediately. They ask by direct message for a WhatsApp number, link to phishing pages that harvest card details, or dangle compensation and get themselves paid through a money-transfer app. Amazon is running a case against Perplexity whose core is a question of identity: Amazon accuses the shopping agent of having disguised itself as an ordinary user. By one lawyer's assessment it may be the first federal lawsuit over a commercially deployed agentic system. The first court sided with Amazon in March 2026; the Ninth Circuit vacated the injunction in August 2026, holding that it was Amazon's own customers who accessed Amazon's systems rather than Perplexity, while noting that more autonomous systems could be judged differently. The question is adjourned. And the Agent2Agent protocol, through which agents are supposed to talk to each other, still does not require an agent's card to be checked for authenticity. Signed cards are provided for; mandatory they are not.
The counter-infrastructure is emerging right now, and at its core it is brand infrastructure: cryptographically signed agents at Cloudflare, Visa's Trusted Agent Protocol at checkout, Mastercard's "agentic tokens" with programmed spending limits, which is to say: a mandate as code. Trademark protection is moving from the logo space into the protocol space: the verified agent becomes the seal of authenticity, the unverified one a counterfeiting risk. Whoever registers trademarks today will register agent identities tomorrow.
What Brand Leadership Is Now
The brand does not become the agent. It becomes what governs the agent: its behavioral specification and its address of responsibility.
One might think this turns the brand itself into an agent. The opposite is true. The agent is substrate, and interchangeable: today one provider's model, tomorrow another's. What has to stay constant through every switch is the logic by which it acts and the address that stands behind it. That is the brand. Whoever fails to define this layer will find that their brand is whatever the model provider makes of it.
I have written about why brands must become machine-readable: so that other people's agents can find, understand, and recommend them. Agent Authority is the second half of the same movement. The Machine Readable Brand makes the brand legible to foreign agents. Agent Authority makes it binding for its own: concession limits, disclosure rules, tone under pressure, the defined point at which a human takes over. Legible outward, binding inward. Only both together make a brand that remains governable in the agent economy.
And this closes the loop to accountability. Authority and accountability are a pair: only those who name who stands behind its exercise can grant a mandate at all. An agent can exercise a mandate, a thousand times in parallel and mostly well. What it cannot do, the Air Canada decision put on record before the industry had even asked the question.
Authority can be delegated. Accountability cannot.
Three Questions, Today
Every brand that deploys an agent, or plans to, should be able to answer three questions in writing. What may our agent promise, grant, and disclose, and where is that written down? How does the other side know it truly speaks for us? And when it gets it wrong: Who finds out, who corrects, who is liable?
An employee who talks nonsense is a personnel conversation. An agent that makes the same wrong concession in a thousand parallel conversations is a scaling problem of brand leadership. The mandate has long been granted, at most brands tacitly. Agent Authority means turning it into a decision, before a tribunal turns it into one for you.
Read next: The Analysis Is No Longer the Product. →
Terms used in this text: See the glossary →