Enterprise tools present a compelling business opportunity as Facebook and Instagram parent Meta looks to diversify its revenue

Meta, co-founded as "The Facebook" in 2004 by Mark Zuckerberg, gets nearly all of its revenue today from advertising. That could change if the company has success with new enterprise AI moves.

Meta Platforms has sent a clear signal that it wants more from businesses than just their advertising dollars.

On Monday, the company announced its Meta Enterprise Platform, with plans to bring the Muse agent and related tools to enterprise customers. Chirantan "CJ" Desai, MongoDB's chief executive, is leaving that job to lead the Meta effort, reporting directly to Mark Zuckerberg.

The hire shows Meta (META) now wants a place in the software budgets that help businesses operate. For shareholders wondering how the company will earn back as much as $145 billion in capital spending this year, the new initiative opens up a potentially important source of revenue. As it stands, Meta gets essentially all of its revenue from advertising.

Don't miss: Meta turned Muse into a viral hit. Now comes the hard part.

Meta's capital spending - expenses put toward chips and data-center assets - supports its existing apps and broader AI ambitions. Because of that, the new enterprise platform would not have to earn back that entire spending budget on its own. Meta's announcement does not disclose enterprise pricing, launch customers or a detailed rollout schedule. Those omissions leave investors unable to estimate the size or profitability of the opportunity.

I see genuine potential here, provided Meta can pair powerful AI with dependable business software. Its models could give customers a reason to choose it. Reliability, integration and support will also matter in turning that advantage into a lasting business.

Meta has a head start. Advertisers and other businesses already use its services. A company paying to attract customers might also consider tools that answer customers' questions, arrange appointments or assist employees with follow-up work. Existing relationships could lower the effort required to introduce those products.

Consider a retailer that wants an agent to handle order inquiries. The retailer needs accurate access to inventory and delivery information. If the agent can issue refunds, it needs spending limits and approval rules. When something goes wrong, employees need a record of its actions and a way to correct them.

For nearly two decades, Meta has sold businesses attention. The enterprise platform asks them to buy something else: software they depend on every day.

The value comes from going through that process reliably. Meta Business Agent, already used by more than 1 million businesses on Meta's WhatsApp and Messenger, is built for that job. Connecting the agent to existing systems, controlling its access and supporting the retailer are what Meta still has to prove it can do.

See also: Could Meta's viral Muse app be the company's ChatGPT moment?

Microsoft (MSFT) and Salesforce (CRM) have an advantage here because businesses already use their software for daily work. Meta will need to give those customers a compelling reason to look beyond the AI tools their existing suppliers offer.

Microsoft lists its enterprise Microsoft 365 Copilot offering at $30 per user per month, paid yearly, with a separate qualifying Microsoft 365 license required. It connects the assistant to business data and applications such as Word, Excel and Outlook. Microsoft includes Copilot Chat at no additional charge with eligible subscriptions.

Meta would compete with both Microsoft's paid Copilot and the chat tool customers already get with eligible subscriptions. A lower subscription price alone might not persuade customers to replace Copilot with Meta's offering, especially if they must pay to set it up while already having access to Microsoft's tools. But if Meta's tools handle everyday office work well, companies could rely on Meta's agent for most employees and buy Copilot only for heavy users. Free Copilot Chat already encourages that.

Salesforce's Agentforce offers both usage-based pricing and per-user licensing: $500 per 100,000 usage credits or flat-fee access from $125 per user per month. If Meta can perform comparable work reliably at a lower total cost, customers could use that alternative to negotiate better Agentforce terms without replacing Salesforce's core software. That could put pressure on how much Salesforce earns from adding AI to its existing customer relationships. A business might retain its existing customer database while buying an agent elsewhere. If Meta can connect effectively to those systems, it could compete for new AI spending without replacing every application underneath it.

The question is how much additional revenue AI can generate when customers have more suppliers to choose from. A vendor could retain its core subscription business while finding it harder to charge a premium for new capabilities.

Meta will also need to fight off fresh competition from OpenAI - and convince businesses it is committed to serving them for the long term. Its earlier corporate-communication product, Workplace, closed in May. Meta had announced a phased shutdown two years earlier. The new enterprise initiative arrives just months after that exit.

Switching costs make a supplier's staying power part of the purchasing decision. The shuttering of Workplace does not establish that the new platform will fail, but it gives prospective customers reason to ask how long Meta will support software they build into their daily operations.

Investors can see how customers answer that question by watching whether they move beyond trials into recurring paid deployments. Renewals and expansion would show whether the tools remain useful after their initial appeal. Clear evidence of time saved or work completed would be more informative than user registrations alone.

They should also ask what remains after the expense of serving those customers. Running models, supporting integrations and maintaining reliable service could absorb a meaningful share of revenue. Reusing existing technology helps, but additional enterprise sales will still carry additional costs.

Cheap introductory offers would complicate that assessment. They could attract customers and give Meta useful experience, but investors would need to know whether those customers stay when discounts end. Winning contracts by absorbing their costs would establish demand without establishing an attractive business for shareholders.

For nearly two decades, Meta has sold businesses attention. The enterprise platform asks them to buy something else: software they depend on every day. An advertiser can move its budget elsewhere next quarter. A retailer that runs orders and refunds through Meta's agent cannot switch so easily. If Meta earns that dependence, it will have business customers whose daily operations run on its software. Those are much harder to lose than an ad budget.

More from MarketWatch: OpenAI launches a rival to Meta's Muse, as the battle for AI agents kicks into high gear

-Jurica Dujmovic