OpenAI is arriving on Google Cloud, giving businesses a new way to buy its AI.

Bojangles’ AI Drive-Thru Is Moving Beyond the Experiment

An automated ordering system can sound impressive in a demonstration. The harder test comes when cars are lining up, customers are changing their orders, and employees are trying to keep food moving through the pickup window. Bojangles’ extended partnership with Hi Auto brings that practical challenge into focus. With the technology deployed at nearly 500 locations and the companies reporting more than one million orders handled weekly, the conversation is moving beyond whether AI can take an order to whether it can reliably support a working restaurant.

That distinction matters because taking an order is only one part of serving a customer. The information needs to reach the kitchen correctly, employees need to prepare the food, and the right bag needs to reach the right vehicle. Improving one step helps most when it improves the whole process. A faster conversation at the speaker has limited value if it creates confusion or additional work at the window.

Restaurant ordering also involves more variation than a simple demonstration can capture. Customers speak differently, ask questions, change their minds, and request substitutions. Background noise and several people talking from the same vehicle can add difficulty. A useful system needs to handle ordinary requests efficiently and recognize when an employee should step in. Reliability includes knowing when to hand over a task rather than making the customer repeat an unsuccessful interaction.

The lunch rush makes those requirements especially important. A small mistake can interrupt several employees while other orders continue arriving. Someone may need to confirm a request, correct a payment, or replace an incorrectly prepared item. Automation creates value when it reduces that friction. If staff regularly have to repair its work, some of the apparent time savings disappear into tasks that are less visible than the initial ordering conversation.

That is why deployment numbers should begin the evaluation rather than end it. Restaurant operators need to understand order accuracy, completion times, customer complaints, employee interventions, and the cost of correcting errors. They should also examine whether the experience holds up during busy periods. A system that performs comfortably during a quiet afternoon may face a different test when the queue stretches around the building.

The employee experience deserves equal attention. An ordering tool could allow staff to concentrate more fully on preparation, packaging, and customer service. But that benefit depends on clear responsibilities and a practical way to take control when necessary. Employees need to understand what the system handles, how to spot a problem, and how to resolve it quickly. Introducing technology without adjusting the surrounding workflow can leave people managing another layer of complexity.

Bojangles’ expansion illustrates the stage many businesses are approaching with automation. The question is becoming less about an impressive capability and more about dependable everyday performance. For a restaurant, success means accurate orders, manageable workloads, and customers receiving their meals without unnecessary frustration. The same standard applies elsewhere: automation earns its place when the business runs better while using it, including during the busiest hour of the day.

---

Salesforce Bets $5.66 Billion That Real-Time AI Decisioning Is the Future of Customer Engagement

Salesforce announced on October 20 that it has signed a definitive agreement to acquire Braze, a customer engagement platform, for about $5.66 billion in enterprise value. Braze shareholders will receive $47.50 in cash per share. The deal is pending and is expected to close in Salesforce’s fiscal year 2028, subject to Braze shareholder approval and regulatory clearance.

**What Salesforce is buying**

Braze coordinates messages across mobile apps, websites, email, SMS, and WhatsApp, and updates customer profiles as people browse and buy. Its BrazeAI layer optimizes content, offers, channels, and timing. The company serves roughly 2,770 customers, including Canva, Dairy Queen, Instacart, and Venmo. Salesforce says marketers now need agentic tools that can understand customers, make decisions, and act in real time. Braze is meant to supply that capability inside Salesforce’s Marketing Cloud, Customer 360, and Agentforce.

**What it means for businesses**

The strategic signal is more important than the price tag. The valuable part of marketing software is shifting from sending messages to deciding, in the moment, what to send, to whom, and through which channel. Salesforce is paying a premium to own that decision layer rather than build it. Expect competitors to respond with acquisitions or aggressive AI roadmaps of their own.

For companies already using Braze, nothing changes immediately. Braze says it will keep building under its own name, and its CEO will stay on. Still, the deal is more than a year from closing, which provides a good window to review contract terms, renewal dates, and integration dependencies, and to ask how pricing and product direction could shift once the platforms merge. Both companies describe their approach as open and composable, but a vendor’s stated intentions and what happens after integration can differ.

For small and mid-sized businesses, this is mostly an indirect story, since Braze’s customers are large consumer brands. The lesson that carries down is about data. Real-time personalization only works when customer events from your website, app, point-of-sale system, and support channels flow into one place in a clean, usable form. Businesses that sort that out now will be better placed to use whatever AI decisioning tools become affordable later, from whichever vendor.

**The open questions**

Regulators have yet to weigh in, and integrations of this size often take longer to deliver value than the announcement suggests. Salesforce has not updated its financial guidance for the deal, so its expected return remains unstated. Watch for the Braze shareholder vote and any signals on pricing and product roadmap as the process moves forward.

---

Meta, Sierra, Walmart, Stripe and Shopify Back a Shared Standard for AI Shopping Agents. Here Is What Businesses Should Do Now

Sierra and Meta announced the Personal Agent Protocol (PAP) on October 6, 2026. It is an open standard that defines how personal AI agents authenticate with businesses and act on consumers’ behalf. Founding partners are Walmart, Shopify, Stripe, Rocket, Genesys, and Instinct.

**What the protocol does**

Sierra says the design splits control: customers decide what access their agents get, while businesses decide which actions those agents may perform. That could give businesses a clearer way to tell an authorized customer’s agent apart from an unidentified bot.

The problem is already real. Meta launched its Muse shopping agent on September 8, and Amazon blocked it on September 20, saying unauthorized agent access violates its terms and that the agent does not identify itself as an agent. Other companies reportedly blocked Muse for the same reason.

**What it does not do yet**

The protocol is at an early stage and is not an industry-wide standard. Sierra and Meta plan to publish the v0.1 specification later this month, along with design workshops and a reference implementation. Payments are not in the first version. Sierra lists payment extensions under future work, so identity and permissions come first and money comes later.

The field is crowded. Walmart and Shopify also take part in the Universal Commerce Protocol with Google, Target, Etsy, and Wayfair, and Stripe co-developed the Agentic Commerce Protocol with OpenAI. Visa’s Trusted Agent Protocol covers overlapping ground from the payments side.

**What this means for businesses**

1. Customer agents are becoming a channel. Customers will increasingly send an agent to book, buy, reschedule, or ask questions. Businesses that can identify those agents and define what they may do will serve them. Businesses that cannot will end up blocking them, as Amazon did, or letting them through unchecked.

2. Permissions are a business decision. Under the model Sierra describes, the business sets which actions an agent can take. Decide now what an outside agent should be allowed to do without a human: check availability, book an appointment, request a quote, or make a payment. Write those rules down before a standard forces the question.

3. Service businesses are exposed as much as retailers. Phone lines, booking systems, and intake forms were built for