A wrong answer from your phone might send you to a closed restaurant. A wrong answer inside a bank can expose confidential records or trigger a payment that should never have happened. Businesses don’t have room for “close enough.”
That’s why I like the investment opportunity behind the safeguards. As artificial intelligence (AI) takes on more responsibility, the information it uses and the permissions it receives become more valuable parts of the process.
More responsibility
An AI agent is software that can perform steps toward a goal, rather than simply answer a question. It might retrieve a record, prepare a letter and send the result to another system. Each step creates another opportunity to use the wrong information or exceed its authority.
Companies need to know which records the agent can access, what actions it can take and when a person must approve the result. A larger model alone doesn’t resolve those business rules. Investors looking at Canadian AI stocks should recognise the opportunity in supplying those controls.
Recently I attended OpenText World, with interviews from two executives who returned to that same foundation. Their argument wasn’t simply that AI would get smarter. It was that businesses needed their information connected and governed before increasingly capable tools could do useful work.
A useful position
OpenText (TSX: OTEX) already sells information management and security capabilities to organisations. That puts it in a useful position when customers want to extend existing systems into AI applications.
Muhi Majzoub, executive vice-president of client advocacy and technology alliances, described governance as part of the blueprints customers need when building agents. He discussed where authentication and security belong, along with help from OpenText’s architects.
I like that OpenText can sell into that need without having to invent the world’s most capable general-purpose model. It can provide the software around the information and work with partners supplying other pieces. That role is different from owning the entire AI system, and potentially broad enough to serve customers using several providers.
A reasonable entry
The earnings give this idea some substance. Fiscal 2026 reported diluted earnings were US$2.58 per share. Compared with the October 23 U.S. closing price of US$27.95, that works out to roughly 10.8 times trailing earnings.
For every US$1 of last year’s reported profit per share, investors were paying about US$10.80. That isn’t a promise of future returns, but it gives me a more comfortable starting point than paying a rich price for profits that haven’t arrived.
Reported earnings can fluctuate with taxes, restructuring and other items. I wouldn’t treat one year’s multiple as the whole investment case. The attraction is having an established earnings base alongside an opportunity to sell additional capabilities.
Keep the limits
Security software cannot eliminate every error, and established technology companies can face new competition. OpenText must demonstrate that customers value its approach enough to keep paying. A low earnings multiple can remain low if growth disappoints.
That keeps my enthusiasm proportionate. I’d buy gradually as part of a diversified portfolio, rather than make one company’s AI strategy the centre of my investing approach. Useful technology still needs a sensible purchase price and steady execution.
Majzoub described how OpenText can help customers introduce AI at a level they are comfortable using. “We wanted clients to define their comfort zone,” he told me. That gradual approach could support several rounds of spending as customers become more ambitious. It also recognises a sensible commercial reality: a customer that feels rushed can simply decide not to buy.
Bottom line
OpenText’s existing work could become more important as customers allow AI to do more. I like owning a business that can earn revenue helping them make that transition safely, with a valuation that leaves room for progress to be recognised.