Shadow AI, or when artificial intelligence tools are used without the green light from a firm’s higher-ups or information technology department, has become widespread in the workplace, with 78% of employees confessing to using unauthorized tools.

While employees may do this to boost productivity, by entering sensitive or proprietary information into unapproved or public chatbots and third-party apps, they could inadvertently expose their workplace to data breaches, compliance issues and other risks.

Many commercial real estate companies may be aware of shadow AI but don’t know the extent to which it occurs, said Sienam Ahuja, founder and CEO of Bryckel AI, an AI enablement firm helping companies manage their CRE workflows within their own AI ecosystems.

“Employers don't know how to govern what they can't even see,” she said. “To manage, control or contain something, you first need to have complete visibility of your landscape.”

In her book The Shadow Partner, Ahuja helps CRE leaders take the reins of their company’s AI use and guides them in developing an adoption strategy that works best for their firm. The book dives into the risks and implications of shadow AI, the difference between building versus buying institutional-grade tools, and how to get the most value out of their AI rollout.

In a conversation with Bisnow, Ahuja dove deeper into what companies need to know about protecting their businesses and the strategies they can embrace.

Bisnow: Your book posits that CRE has a “strategic sequencing problem” due to unstructured use of AI. How did we get to this point?

Ahuja: The AI tools arrived before there could be any sort of governance over them. Individual employees took the initiative to start using these tools because it seemed convenient for them to automate their rote work, and firms never got ahead of the situation or did a deliberate rollout.

The companies’ response to employees using AI tools not provided by their employer was to start piloting them in an ad hoc way. This uncovered a few issues: These tools may not solve the employees' problem, employees don't have sufficient training to use them, or their legacy data is not ready for those tools. Therefore, companies are not going to see the return on investment.

For companies, this is a strategic sequencing problem, because they have not gone through the entire process of evaluating their organization and contemplating what a thoughtful rollout for their specific organization looks like. Instead, fear of missing out is doing the driving: roll some tools out, cross your fingers, hope for the best.

Bisnow: What is the real risk for CRE specifically?

Ahuja: In CRE, every piece of data is sensitive. The risk that comes with the disclosure of proprietary information without data governance is extremely high, because every piece of data touches financials, transaction histories, tenant-landlord relationships, your underwriting assumptions or your borrowing data.

AI tools are designed to be easy to use: You upload something, you get a structured output which automates a week's worth of tedious work, and employees use it. AI is so easy and confident in its responses that someone who isn’t tech-savvy may not think about how to use it responsibly and whether the output is reliable or not. The cost implications can be millions of dollars.

Bisnow: How can a firm begin to approach this issue?

Ahuja: Trying to come down on the behavior with a hammer doesn't change the behavior. The only thing that does is structured enablement: teaching your employees what reliable, safe AI looks like for their specific role.

Giving them a chat assistant with an empty chat box is not solving the problem. Educating them on how to build skills and work with AI agents to automate repetitive tasks, how to clean the underlying data before you even run that agent, and how to avoid tokenmaxxing — these are the solutions that are needed first.

Bisnow: How can they ensure they’re applying the correct remedies?

Ahuja: Companies need to look at outcome metrics, or how many employees today are successfully using AI skills and AI agents to automate rote work. How much more productive and engaged are they, and of the AI outputs that they are generating, how much of the information is reliable?

A cottage industry of influencers and ad hoc trainers has sprung up, charging for AI education and making the technology look like magic. Employees follow them on social media, hoping to master the AI assistants their companies have handed them. But the demos gloss over the risks. Some trainers even encourage connecting these assistants to third-party application programming interfaces the organization never sanctioned.

It is imperative for organizations to not only have strategies in place, provide safe tools and enable employees with those tools, but really double down on structured education on how to safely use those company-provided tools without leaning into random training from the internet.

Bisnow: What role does Bryckel play in this?

Ahuja: I coach organizations to build AI fluency where it pays off first: repetitive, templatizable tasks, with the firm's own language captured as skills they own. Because I pair domain expertise with hands-on AI proficiency, my guidance covers where AI should take a pass and where human judgment prevails. Every firm should start here, whether they work with me or on their own.

Where Bryckel comes in is the next step. Complex, high-stakes workflows can't be one-shotted. They need multiturn agents, enterprise-grade tooling and token optimization to be reliable. We deploy those agents inside the firm's current environment, working through the AI assistants employees already use, so consistency is engineered in rather than left to individual prompting skill.

The AI coaching solves the companies’ employee behavior issue, and the AI agents that Bryckel provides solve the infrastructure problem.

This article was produced in collaboration between Studio B and Bryckel AI. Bisnow’s editorial staff was not involved in its creation.

Studio B is Bisnow’s in-house content and design studio. To learn more about how Studio B can help your team, reach out to studio@bisnow.com.