Enterprise Sales Engineer

Meta and Sierra announce open standard for personal AI agents. Key points * Meta and Sierra announced the Personal Agent Protocol on October 6, 2026, with six named industry partners. [[1]] * The proposed standard covers authentication, permissions and agent access through websites, APIs or company agents. [[2]] * Customers would decide whether an agent receives read-only or write access, while businesses set the actions they allow. [[2]] * Sierra plans to publish a v0.1 specification, design materials and a reference implementation later in October. [[2]] Meta and AI startup Sierra are developing an open standard for how personal AI agents interact with businesses, the companies announced October 6, 2026. Sierra's announcement named Genesys, Instinct, Rocket, Shopify, Stripe and Walmart as industry partners. [[1]] [[2]] The Personal Agent Protocol is intended to help businesses identify an agent acting for a customer, determine what it is authorized to do and control how it accesses business systems. Customers would decide whether an agent receives read-only or write access. [[2]] The announcement comes as Meta expands Muse, its personal AI agent, into tasks such as booking travel, filling out forms and making purchases with user approval. [[3]] How the proposed standard works. Sierra said the protocol would let an agent begin as a guest to check product availability or review a returns policy, then authenticate through the company's website when an account-specific task requires access. [[2]] The session would use OAuth, an established authorization standard. Businesses could choose whether agents interact through ordinary web pages, APIs built on standards such as MCP and OpenAPI, or the company's own conversational agent. [[2]] The protocol is still under development. Sierra said it plans to publish version 0.1 later in October, hold design workshops and release a reference implementation. [[2]] The companies involved. Sierra's announcement named Genesys, Instinct, Rocket, Shopify, Stripe and Walmart as partners. The group includes customer-service software, retail, payments and enterprise technology companies. [[2]] Sierra co-founder Bret Taylor, who also chairs OpenAI, is leading the initiative, according to CNBC. Taylor said companies still need to determine when a personal agent should be allowed to access information and perform tasks. [[1]] Meta's agent push. Meta introduced Muse in September as a personal AI agent that can act across a user's apps and services. Meta says Muse can open a browser, fill out forms, book travel and continue working until it needs user approval for a sensitive action such as sending an email or making a purchase. [[3]] Meta separately said in June that more than 1 million businesses were using Meta Business Agent on WhatsApp and Messenger to respond to customers. Its business-agent tools can answer questions, recommend products, book appointments and close sales. [[4]] What has not been decided. The announcement establishes a working group and a proposed framework, not a completed or widely adopted standard. Sierra has not yet published the v0.1 specification. [[2]] Sierra described possible future extensions for more detailed action limits, shipment and flight notifications, and payments that could let an agent complete a purchase without sharing a customer's card number. Those features remain proposals rather than confirmed capabilities of the initial protocol. [[2]] OpenAI and Anthropic were not listed among Sierra's named partners in the announcement. CNBC reported that Taylor expects other major AI companies to participate, but broader industry adoption remains unresolved. [[1]] Companies mentioned. At the intersection of AI, tech, and markets. The stories that matter, in one email. Free - unsubscribe anytime.

Meta & NiCE back AI agent protocol for customer service. Wed, 7th Oct 2026 (Today) Meta has previewed the Personal Agent Protocol, an open standard for interactions between consumers' AI agents and businesses. NiCE is co-developing the protocol with Meta, Sierra and other industry partners. The proposal outlines how a customer's personal AI agent can find a business, authenticate itself and act on the customer's behalf. It offers an early indication of how companies may need to handle a growing volume of requests initiated by software agents rather than people. According to NiCE, the standard is intended to address a series of operational questions for customer service teams, including how businesses verify which customer an agent represents, what the agent is allowed to do and how those interactions are managed across automated systems and human staff. It also points to a shift in demand patterns. Rather than reducing service volumes, personal agents could increase them by pursuing tasks customers might otherwise abandon, such as repeated follow-up questions, refund requests or booking changes. "Personal AI agents have endless time that consumers don't, and that brings a new era of growth for customer service. Customers' agents can sign in, ask follow-up questions on their behalf and get to resolution faster, which means more demand, more persistence and higher expectations. An open standard is how that demand reaches businesses with trust, and NiCE is proud to shape what comes next alongside Meta and others. Companies that compete every day agreeing on a standard so customers and enterprises both benefit is the industry at its best. Personal agents bring customers to the door, and resolution brings them back," said Scott Russell, Chief Executive Officer, NiCE. How It Works Under the proposed approach, a company would publish information at a known digital address telling personal agents where to sign in and which interfaces they can use. The protocol is designed so an agent can discover that information in a single request before continuing the interaction across application programming interfaces, web pages or conversational channels. NiCE said the model also relies on customer-specific authentication tokens, with access limited according to the business's own rules. That structure is intended to let companies decide which personal agents they will accept, what those agents may do and when a person should take over a case. Philipp Heltewig, Chief AI Officer at NiCE, said the protocol creates a common language between a customer's software agent and a company's systems. "Personal agents already work for consumers, and today a customer's agent and a company's agent share a common language. When a customer's agent arrives with a goal, like a refund or a rebooked flight, the business's AI agent checks who it represents and what it is allowed to do, then works the request to resolution using the systems the business already runs on. When the case needs judgment, a person joins the same conversation with the full context, and every reply says who is responding. That is AI agents and people working as one crew, with clear jobs, shared context and a clean handoff. Enterprises decide which agents they serve, what access those agents get and when a person steps in, and now the conversation turns to what we build with it," said Heltewig. Operational Pressure For contact centre operators and IT teams, the prospect of machine-driven customer requests raises questions not just about access, but also scale. Traditional service models have been built around human behaviour, where delays, drop-offs and limited persistence naturally constrain demand. Personal agents could change that pattern by sending requests at machine speed and continuing interactions without the time limits of a human customer. That would require businesses to think not only about external connections to personal agents, but also about internal routing between AI systems, workflows and staff. Neeraj Verma, Vice President, Agentic AI, NiCE, said the proposed design aims to keep those interactions consistent while preserving business control. "What I like about the Personal Agent Protocol is how much it solves with a simple design. A company publishes one document at a well-known address that lists its sign-in endpoints and interfaces, so an agent finds everything in a single request. Each personal agent gets its own token for each customer, with access scoped to what the company allows, and that same token works across APIs, web pages and conversations, so the experience stays consistent from first request to resolution. Every reply says whether an AI agent or a person is responding, and when a case calls for judgment, a person joins the same conversation with full context. Enterprises keep control throughout: they set which agents they serve, what access those agents get and when a person steps in," said Verma. The initiative comes as businesses across customer service, digital operations and enterprise software examine how personal AI assistants could change online interactions, particularly in sectors where identity, permissions and audit trails are central to the transaction. For companies preparing for that shift, the protocol suggests the challenge will extend beyond opening a technical connection. They will also need systems that can distinguish between customers, agents and staff at each stage of an interaction, while handling a potentially much higher volume of service requests than today.

Sierra launches fleming-1 to detect AI agents calling by phone - unite.ai. Sierra launched fleming-1 on October 7, 2026, a model designed to detect when the caller on a phone call is another AI agent. The company described the release as an update to the constellation of models behind agents built on its platform. Sierra said the question companies face when the phone rings has shifted from who is calling to what is calling. The company first encountered the issue in 2025, when agents built on Sierra started calling agents built on Sierra in healthcare. With personal agents such as Muse and Instinct taking off, Sierra said, a large share of the calls companies receive could come from AI acting on behalf of consumers. How fleming-1 detects agent callers. According to Sierra, fleming-1 analyzes a caller's speech in real time, scoring the audio for signs that it was generated by AI. The model looks beyond how a voice sounds to a human listener, assessing whether the voice on the other end is synthetic while the call is still happening. Sierra said telling a synthetic voice from a human one over the phone is getting harder. Background noise, muffled audio, and a bad connection can obscure the clues, while AI voices keep improving. Sierra flags the calls fleming-1 identifies as likely AI. The company said the model is tuned to be conservative by default so that real people do not inadvertently get flagged, and the business decides what happens next. The Caller ID comparison. Sierra framed the model through the history of Caller ID, which BellSouth launched in Memphis, Tennessee, in 1988. Caller ID never told you whether to pick up, and Sierra said fleming-1 works the same way: it is for companies to decide how they want to handle such calls. The company said its starting principle is that companies should know what is calling. As examples of what businesses might do with that information, Sierra said a bank might want to add a verification step when the caller is an agent, and a company with high call volume might start by measuring how often it happens. Sierra said some automated callers will be fraudsters testing account security at scale, while others will be customers who handed off a chore to an agent, or people who rely on text-to-speech. Sierra said Caller ID ended up on every phone because people found it useful, and that it believes the same will be true of knowing when an agent is calling. Relationship to Personal Agent Protocol. The launch came one day after Sierra announced Personal Agent Protocol on October 6, 2026, an open standard Sierra and Meta are developing with industry partners Genesys, Instinct, Rocket, Shopify, Stripe, and Walmart that defines how personal agents interact with businesses. When an agent uses the protocol, the company knows it is an agent and who it is acting for. Sierra noted that some agents may not respect the protocol. Sierra described the best case as an agent that says who it is, which is what the protocol is for, with fleming-1 covering the calls where that does not happen. Sierra plans to publish the protocol's v0.1 specification later in October 2026, host design workshops with interested parties, and publish a reference implementation to help developers get started. Voice and the constellation architecture. fleming-1 applies to a voice channel Sierra announced on October 9, 2024, when the company said its AI agents can pick up the phone. Sierra said those agents integrate with existing call-center platforms, can sit in front of or behind traditional IVR systems, and escalate to human teams with AI-generated summaries when a situation requires a person to intervene. Sierra's Agent OS allows a company to build an agent once and deploy it across channels, including chat and phone. The model also joins an architecture Sierra detailed in a December 3, 2025 post: agents built on Sierra are assembled from 15+ frontier, open-weight, and proprietary models, each selected for the job to be done. Sierra said it runs task-specific evaluations and invests in fine-tuned models where off-the-shelf models fail to meet its constraints. Sierra said fleming-1 works with any voice agent built on Sierra and requires only that the business turn it on.