The next phase of business AI is moving beyond chatbots.

Meta is now putting its AI agent, Muse, directly into the workflows of small businesses, allowing owners to connect the agent to many of the applications they already use to run their companies. The announcement comes as Meta simultaneously builds a dedicated enterprise AI business, signaling a broader shift from AI that answers questions to AI that can work across a company’s operations.

For small-business owners, that distinction could matter considerably.

Muse Can Now Connect to the Business Behind the Business

Meta’s new Muse for Small Business can connect to Facebook and Instagram business accounts as well as services including Shopify, QuickBooks, Slack, Canva, Dropbox, Asana, Notion, Stripe, Klaviyo, Zoom and HighLevel.

The idea is relatively simple: instead of asking an AI about information that has been manually copied into a conversation, the business can give the agent access to the systems where its information already lives.

Meta says an owner can give Muse goals such as analyzing sales performance, finding new customers, improving advertising or figuring out which business expenses deserve attention. Muse can then work across the connected information to produce recommendations and carry out portions of the work.

That represents an important change in the AI market.

For years, the typical small-business AI pitch was essentially:

“Ask our chatbot a question.”

The emerging pitch is:

“Give our agent a business objective.”

The Approval Button May Be More Important Than the AI

There is an important limitation built into Meta’s system.

Meta says Muse will not publish content, send messages or spend money without the user’s approval.

That distinction becomes increasingly important as AI agents gain access to business systems.

An AI that drafts a Facebook advertisement is one thing.

An AI that can access a company’s advertising account, customer information, financial records and payment systems is something entirely different.

The more useful agents become, the more consequential their mistakes can become. That is one reason the industry is increasingly focused not only on what AI agents can do, but on what they are permitted to do without human authorization.

Small Businesses Are Becoming the Next AI Battleground

Meta’s announcement did not happen in isolation.

On September 28, Meta launched Meta Enterprise Platform, a new business focused on bringing its AI models, agents and developer tools to companies. Meta specifically identified Muse, Meta Business Agent, Muse API and Muse Code as components of the platform.

A day later, OpenAI announced Dots, always-on AI agents designed to continue working on goals between conversations. OpenAI says a Dot can operate from its own cloud computer, connect to the applications it needs and continue making progress for the user.

The timing is revealing.

The major AI companies are increasingly competing for the same thing: the work performed inside businesses.

That potentially puts traditional business software in an interesting position.

If an AI agent can sit between a business owner and dozens of applications, the owner may eventually interact less with individual software products and more with the AI layer connecting them.

What This Means for a Local Business

Consider a small plumbing company.

Today, the owner might have one system for leads, another for scheduling, another for accounting, another for marketing, another for customer communication and another for social media.

The owner or employees have to move information between those systems.

An agent changes the potential workflow.

A business owner could theoretically ask:

«“Find out where our leads are coming from, identify the campaigns producing the best customers, follow up with prospects who haven’t booked, and prepare next week’s marketing plan.”»

The important development is not that AI can write the resulting report.

It is that the agent can increasingly access the underlying business information required to produce it.

Meta is explicitly building toward this model by connecting Muse to the applications small businesses already use.

The Opportunity for Businesses Is Also a Warning for AI Service Providers

This shift creates a more complicated market for companies selling AI automation.

An AI receptionist that simply answers questions may no longer be the entire opportunity.

The larger opportunity is connecting AI to the business’s actual workflow: lead generation, qualification, follow-up, scheduling, customer records, payments, marketing and reporting.

At the same time, some basic automation services could become commodities as platforms such as Meta and OpenAI incorporate more capabilities directly into their products.

That does not necessarily eliminate the need for implementation.

For many small businesses, the difficult part isn’t having access to an AI agent. It is determining what the agent should actually do, which systems it should access, what permissions it should have, and where humans must remain in control.

That creates a different kind of service opportunity: designing and managing the business process around the AI rather than simply selling access to the AI itself.

The Bigger Story

The significance of Meta’s announcement is therefore bigger than Muse.

Meta is betting that small businesses will eventually want AI that understands their business context and can operate across the software stack rather than another isolated chatbot.

OpenAI is making a similar bet with its always-on Dots.

The competitive question is increasingly shifting from:

“Which AI gives the best answer?”

to:

“Which AI can actually get the work done?”

For small businesses, that could ultimately be a much more consequential question.

The companies that figure out how to safely delegate repetitive operational work to AI agents could gain significant leverage without necessarily adding more employees.

But the businesses adopting these systems will also have to treat permissions, oversight and security as part of the implementation—not as an afterthought.

The AI assistant is becoming an AI operator.

And for small businesses, that may be the development worth watching most closely.

It was a night of shock, awe, surprise, and enormous talent as Randy Roberts brought a star-studded Laurie Beechman Theatre to its feet several times with a brilliant display of singing, comedy, acting, and female impersonation.

The biggest surprise of all? Randy began the evening as a spot-on Bette Midler, then transformed into a remarkable Liza Minnelli and Cher. But by the end of the night, there was no mistaking who the real star was: the incredibly talented Randy Roberts himself.

Eda and I were at the Copa back in the day when Bette Midler first entered the room to begin her performance. The excitement was electric, and I can honestly say that the wall-to-wall audience reaction when Randy entered the Laurie Beechman Theatre as Bette last night was every bit as thrilling to experience. The audience erupted. For a moment, it was as if Bette herself had walked through the door.

What a performer. What a night. And what a thrill to watch Randy Roberts remind a room filled with sophisticated cabaret and theatre connoisseurs that, while his impersonations are extraordinary, the real revelation is the immensely talented artist that is Randy Roberts himself. As usual, no spoilers here! Not with the buzz racing along 42nd Street that the D’Angora’s may have plans to keep Randy on that Laurie Beechman stage for many more nights for the kind of dazzling show business that left this audience begging for more.

No spoilers—except this one: Randy’s rendition of Stephen Sondheim’s “The Ladies Who Lunch” from Company left the entire company we were with last night completely spellbound. And speaking of that company, check out my photos from last night’s show.

MICHAEL D’ANGORA, RANDY ROBERTS, TOM D’ANGORA

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

RANDY ROBERTS

MICHAEL D’ANGORA, RANDY ROBERTS, TOM D’ANGORA

TOM D’ANGORA, FRAN DRESCHER, MICHAEL D’ANGORA

PETER MARC JACOBSON, RANDY ROBERTS & FRAN DRESCHER

TOM D’ANGORA & MICHELE LEE

EDA SOROKOFF, MELISSA ERRICO, LORAN DALLAS, MICHELE LEE

LEE ROY REAMS & LORNA DALLAS

MICHELE LEE & FRAN DRESCHER

RICHARD SKIPPER & MICHELE LEE

EDA SOROKOFF, MICHELE LEE, MARDIE MILLIT

SUSIE MOSHER, IIENE GRAFF, RICHARD SKIPPER

RANDY ROBERTS & LORNA DALLAS

MELISSA ERRICO

RANDY ROBERTS & MELISSA ERRICO

GABRIELLA STRAVELLI, RANDY ROBERTS, SUSIE MOSHER

MARK SENDROFF, FRAN DRESCHER, STEPHEN SOROKOFF

LAURIE BEECHMAN THEATRE

WEST BANK CAFE

A Google Gemini security test has exposed a problem that businesses deploying increasingly autonomous AI agents will have to take seriously: an AI system can be given a legitimate task, operate inside what is supposed to be a controlled environment, and still end up interacting with real-world systems if the surrounding safeguards fail.

In May, Gemini was being evaluated by AI security testing company Irregular in a simulated “capture the flag” exercise. The model was supposed to attack a fictional company inside a controlled environment. But the test environment unintentionally provided internet access, and the fictional company shared a name with a real company. Gemini subsequently accessed systems belonging to three real companies.

According to Google, Gemini used publicly available information and credentials it either found or guessed. In all three cases, the model stopped once it recognized that it had reached real companies rather than the simulated targets. Google said the affected organizations were notified and that testing procedures were changed.

The business lesson is bigger than the incident

The important issue for companies isn’t that Gemini “turned rogue.” This was a testing failure involving unintended internet access and insufficient separation between simulated and real systems.

The bigger lesson is that AI agents need to be treated as systems with access, not simply as software that produces text.

That distinction becomes increasingly important as companies give AI the ability to browse websites, access databases, send emails, write and execute code, interact with customers and make changes inside business applications.

A human employee might have access to ten systems. An AI agent could potentially have access to dozens, while operating at machine speed and continuously pursuing a task.

That creates a new security question for businesses:

What exactly is this AI allowed to touch, and what prevents it from touching everything else?

The Google incident is particularly relevant because similar problems involving AI models from OpenAI, Anthropic and Meta have also emerged during security testing conducted by Irregular. The company said an issue that unintentionally allowed models to access the internet had been fixed.

AI security is becoming an operational requirement

For businesses adopting AI agents, traditional cybersecurity controls are no longer enough by themselves.

Companies will increasingly need to establish separate identities and permissions for AI agents, restrict which websites and applications they can access, isolate testing environments from production systems, monitor their actions and maintain logs showing what an agent did and why.

They also need safeguards that assume the AI can make mistakes.

An agent doesn’t necessarily have to be malicious to create a security incident. It can simply misunderstand its environment, follow instructions too literally, discover credentials it should never have seen, or encounter a system that wasn’t supposed to be reachable.

That changes the economics of AI deployment.

The question for businesses is no longer simply “Can we automate this job with AI?”

It is also “What happens if the AI makes the wrong decision while it has permission to act?”

As AI moves from answering questions to performing actual business operations, controlling those permissions may become just as important as choosing the underlying AI model.

What this means for businesses: Companies should begin treating every autonomous AI agent as a new digital employee with its own identity, permissions, limits and audit trail. The more authority an AI receives, the more important it becomes to design the environment around the AI—not just the AI itself.

For years, the AI race was largely about building better models. Now another race is emerging: getting those models into actual businesses.

Google Cloud and Accenture have created a new business group dedicated to helping companies deploy Google’s Gemini AI across their operations. The partnership brings together AI specialists and engineers who can work directly with businesses to turn AI technology into functioning systems.

The development points to a growing problem for companies: buying access to AI is relatively easy. Figuring out where it should be used, connecting it to existing software and data, redesigning workflows, training employees and getting the system into production is much harder.

That creates an important shift in the AI economy.

AI Implementation Becomes the Opportunity

Businesses may not need another chatbot. They need someone to take an existing AI technology and make it useful inside their company.

That could mean automating customer service, analyzing documents, assisting employees, improving software development, processing data or connecting AI to existing business systems.

Large technology companies are increasingly building dedicated services around this implementation work because simply providing the model is only part of the customer’s problem.

For smaller businesses, the lesson is similar: AI adoption may be less about finding the newest model and more about finding the right business process to change.

The companies that can bridge that gap — between what AI can do and what a business actually needs — are likely to become an increasingly important part of the AI economy.

Like a magician who somehow keeps producing one dazzling surprise after another from up her sleeve, entertainer musician extraordinaire Gunhild Carling had the City Winery audience at Pier 57 on the edge of its seat last night, as her jazz/pop music and showmanship delivered one thrilling song after another.

If Louis Prima and Louis Armstrong had grown up on a steady diet of Swedish meatballs, the result might very well have been Sweden’s Gunhild Carling. She has Prima’s wild, infectious energy and Armstrong’s unmistakable trumpet phrasing and vocal flair—and she brings it all to the stage with a show-business sensibility and proficiency on many instruments that is impossible to resist.

Remember those circus clown cars where you were convinced there couldn’t possibly be room for another performer, only to have one more, and then another, and another, somehow climb out? That’s Gunhild Carling’s band. Just when you think you’ve seen the last great musician, out comes another one—each more talented and entertaining than the next, including her brother Max Carling.

A special shout-out to NYC and Birdland favorite drummer Daniel Glass, who delivered some amazing solos that had the audience cheering.

It felt like one great Bon Voyage party last night on Pier 57. And the next time Carling’s ship comes sailing into town, make sure you don’t miss the boat.

Gunhild Carling and her band are definitely a unique musical experience—part jazz concert, part circus, and completely entertaining.

GUNHILD CARLING

GUNHILD CARLING

MAX CARLING

GUNHILD CARLING

GUNHILD CARLING, MAX CARLING,Clarinet, Neil Fontano, Piano

DANIEL GLASS

DANIEL GLASS

GUNHILD CARLING, NEIL FONTANO, Bass

GUNHILD CARLING

GUNHILD CARLING

CITY WINERY

GUNHILD CARLING

GUNHILD CARLING

GUNHILD CARLING

GUNHILD CARLING

PIER 57

PIER 57

COMING ATTRACTIONS

COMING ATTRACTIONS

CITY WINERY