Three years into the artificial intelligence boom, the companies that get the most attention from investors are still those that are easy to see. Chatbot answers questions. The agent completes the task. The visible layer of the AI economy drives much of the narrative.
But a different kind of artificial intelligence is making progress, but it goes mostly unnoticed. It sits within the systems that generate invoices, transfer funds between organizations, assess credit risk, and capture billable time.
When it works, no one calls it artificial intelligence. They just noticed fewer bills being declined and the time it took to get paid decreased.
The gap between where AI gains traction and where it actually creates value is increasingly important to investors. Legal services is one area where the gap is closing rapidly, and companies providing legal services are starting to look like a serious infrastructure area.
Also read: Apple’s biggest artificial intelligence advantage may face new privacy challenges
The most effective AI is AI that no one uses
Most enterprise software still requires people to get used to it.
The more interesting shift happening now is the opposite: Intelligence is being embedded so deeply into workflows that the people in those workflows never have to think about it.
The attorney does not need to know whether the AI reviewed the invoice against the client’s billing guidelines before it was issued. The finance director does not need to know why the exceptions were reduced. The value is displayed in the results, not in the interface.
This is where enterprise artificial intelligence becomes economically meaningful, according to Ahmed Shaaban, CEO of legal operational technology company Fulcrum GT.
“We’ve shown the world that artificial intelligence can have a conversation,” Shaaban said in an interview with TheStreet. “Now let’s put that into practice. Tie that to how businesses bill, manage risk and get paid, and you start to open up possibilities that were previously unrealistic. That’s the next chapter: bringing intelligence inside the business to help people do things they simply couldn’t do before.”
The challenge is that embedding intelligence into operational processes requires that the underlying data and systems are already in a reasonable state. If billing, time capture and transaction management exist in disconnected silos, then adding artificial intelligence will expose these gaps rather than close them.
Remix/Getty Images
Law firms first build operational foundations
Some of the world’s best-known law firms have already been laying the groundwork for several years. Their experiences are instructive, not because they represent the final product, but because they demonstrate how much more work is required before any AI deployment can actually deliver.
DLA Piper International starts with process standardization and a shared services center in Warsaw. The company established clear senior ownership of its finance function, implemented enterprise technology, and built a coherent internal team capable of delivering cross-border transformation.
Garry Swaine, the leader of the project, said the platform was launched in 30 countries on the same day. Fundamentals make speed possible.
Eversheds Sutherland has taken a similar route, integrating around a core platform that can grow with the business. The company spent 180 hours consulting directly with attorneys to understand the friction they encountered in their daily work.
The work revealed a collection process that requires manual updates of approximately 300 records per month, and the company is currently exploring whether an artificial intelligence agent can handle this without overhauling the broader collection model.
Both companies describe their AI applications as an area still under development. The platform and processes are already in place. Next, efficiency will be improved by layering artificial intelligence.
A new ownership model is emerging
In addition to traditional partnerships that invest through their own structures, managed services organizations (MSOs) are emerging as a separate avenue for modern legal operations.
An MSO provides shared non-legal services, including finance, technology and management, through a separately organized entity, while leaving the law firm itself in the hands of attorneys.
More artificial intelligence:
- Nvidia just made a move Wall Street wasn’t ready for yet
- Microsoft just picked a side in the AI policy battle
- OpenAI just revealed something truly shocking
Orion Legal MSO, established in January 2026 by Uplift Investors, provides an early example. Louisiana personal injury firm Dudley DeBosier joins as founding partner.
Uplift said the structure is designed to provide operational support without a change in ownership or control of the company. Said investment themes include technology, data and artificial intelligence integration between companies supported by MSOs.
It’s unclear whether the MSO model is better than a well-run partnership that invests in its own shared services.
The examples of DLA Piper and Everett show that mature companies can achieve significant operational transformations without the need for external capital or separate organizational structures. The MSO model adds a path; it doesn’t build a superior system.
What enterprise software investors are paying attention to
For investors evaluating enterprise software companies, the dynamics in the legal services space reflect a broader question: where artificial intelligence creates lasting business value.
An established company expanding internationally, a partner improving its existing operations, and an MSO building a shared infrastructure across multiple companies may all require the same underlying capabilities. Demand can come from many directions at once.
Scott Mozarsky, former president of Bloomberg Law, framed investment questions around proximity to mission-critical workflows. “From an investment perspective, just saying a product contains artificial intelligence has become less interesting,” Mozarski told TheStreet.
“The more important question is whether the company is in mission-critical workflows, has access to differentiated data and context, and whether its technology is tightly connected to its customers’ operations to create measurable economic value.”
This framework is useful precisely because it separates hype from signals. The software vendor’s position on key processes is a starting point, not a conclusion.
Investors still need to examine how embedded the product actually is, how much it costs to implement, whether adoption is truly sticky, and whether promised efficiency improvements translate into numbers that customers can point to.
Governance cannot be added after the fact
The more deeply artificial intelligence is integrated into operational processes, the greater the risk when things go wrong. A system that can change financial records or initiate payments requires more than just a capable model. It requires reliable data, defined access controls, and a clear path for human intervention before errors compound.
At Eversheds Sutherland, compliance, cybersecurity and in-house legal counsel are part of the delivery team from the outset.
This is a practical acknowledgment that once AI is already embedded in a workflow, governance cannot transform it. It has to be part of the design from day one.
This requirement raises the bar for any software provider looking to integrate artificial intelligence more deeply into legal and financial operations. The companies that make it work are likely to be those that view data quality, licensing, and auditability as engineering issues rather than policy issues.
Companies that later try to add controls will discover why the order of operations is so important.
RELATED: Jim Cramer reveals secrets to AI stock investors