AI Will Save the Public Company. But First,
the Regulators.
Cheaper to be public. More retail ownership. More global investors. More IPOs. That is the future. But the SEC and the CSA are coming first. The companies that prepare own the next decade. At the speed AI is moving, the ones that wait may not be able to catch up.
By George Fleming, Founder & CEO, Versance Technologies
The public company is about to be rebuilt around AI.
Not gradually. Not optionally. Not on a regulatory timeline.
The issuers that move first will set the AI standard for the public market. Everyone else gets measured against it. The issuers on the wrong side of the standard will discover what that means in court. In a D&O renewal. In the next letter from the SEC or a Canadian securities commission. In the audit committee meeting where nobody had the answer ready.
Public markets shrank for twenty-five years. Fewer listed companies. Higher costs to be public. Retail investors pushed away from the issuers they own. Capital concentrated in fewer hands. The market's original promise of open access on both sides broke. AI reverses every line of that story for the companies that get to it first.
That ends now. Not gradually. Not in a decade. In the next three to five years.
AI will reshape how public companies operate. How investors evaluate them. How capital moves between the two. Adopt it and the math flips. The company runs leaner. The team communicates faster. Investors who never saw the company find it. Investors who use AI to read the record will be sharper. Better-informed. More globally distributed than at any time since online brokerage went mainstream.
The friction that made the public market a worse option for two decades collapses. Disclosure cost. Communications overhead. Regulatory complexity. All of it falls for the companies that get this right.
This is the manifesto for those companies.
What the next decade of public markets looks like
Being public gets cheaper. The regulatory burden today is brutal. Not because the rules are wrong. Because complying with them eats the calendar. Public companies burn most of their IR, communications, and disclosure budget on work AI built for the task can do in a fraction of the time. Filings drafted with citations built in. Investor email handled with audit-grade defensibility. Disclosure reviews that used to take weeks done in hours. The time-and-attention cost of being public collapses.
Retail ownership expands. Retail investors got priced out of serious research for decades. No AlphaSense. No sell-side coverage. Reading filings unaided. AI ends that. The retail investor who runs your 10-Q through Claude today already gets analysis that used to require an institutional subscription. By 2028, every retail investor with a phone will have a citation-backed research agent. The information moat institutional money built around itself collapses. Retail ownership expands behind it.
The investor base goes global. Language and disclosure barriers fall together. An investor in Frankfurt reads a Vancouver mining company's MD&A in German with the same depth a Toronto analyst gets in English. An investor in Singapore runs cross-company comparisons across NYSE, TSX, and TSX-V issuers without an institutional research subscription. Information becomes the lingua franca of public markets. Capital follows information. The pool of investors paying attention to your company goes global.
More companies go public. When the cost of being public collapses, the math for staying private changes. Companies refuse the IPO today because the regulatory and communications burden eats the value. That burden falls when AI handles most of the IR, disclosure, and communications work. The IPO pipeline reopens at smaller scales. The de-listing trend of the last twenty-five years reverses.
Capital moves faster, with better information. When investors have AI reading every filing across thousands of issuers in real time, capital allocation tightens. Mispricing closes faster. Underfollowed companies get attention they could never have earned on a sell-side budget. The public market gets sharper as a discovery mechanism. Capital moves faster. Capital moves harder. Capital moves in more directions than at any time in modern markets.
This is not a vision. It is a trajectory. The mechanisms are already running. The only variable is which public companies move first. And which ones get moved on.
Why generic AI cannot deliver this
There is a catch. None of the above happens with the AI most companies use today.
A public company is not a marketing department. It is not a law firm. It is not a hedge fund. It is a regulated entity. Every record you publish can become evidence in a complaint, a subpoena, or a regulator's letter. When a generic AI tool helps your team write a release and something goes wrong, the company answers. The AI vendor does not. There is no model card that exempts you from Reg FD. There is no LLM provider you can name as a co-defendant when a forward-looking statement blows up.
That is the asymmetry. The model writes the sentence. The issuer carries the consequence. The vendor captures the productivity. You inherit the exposure. The deal is already in place, whether you have looked at it or not.
Generic AI cannot drive this revolution. The cost of being wrong sits entirely on the issuer. Every public company that scales generic AI into disclosure and communications accumulates unauditable risk faster than it captures productivity. The economics break the moment a regulator asks for the record. Or a plaintiff. Or an auditor. Or a board member. Speed. Leverage. Scale. None of it pays until the AI is built to the standard the work requires.
Most CEOs think about AI the way they think about email. A productivity tool. Helpful. Harmless. That mental model is wrong. Inside a public company, generic AI is an unsupervised contractor. With full access to your communications. With no securities training. With no record of what it said or why. No CFO would knowingly hire that contractor. A chatbot is playing that role today across IR, communications, governance, and operations. That is what stands between every public company and the revolution above.
The bar: Compliance-grade AI
There is an AI that can actually do this work. I call it compliance-grade AI. The label matters because the bar matters.
Compliance-grade AI has four defining attributes. Any tool that cannot claim all four is not in the category. No matter what its marketing says.
It knows your record. The model retrieves from your filings, releases, presentations, approved Q&A, and governance documents. Not the open internet. The output is grounded in the documents it should be citing.
It cites every fact. Every output names the source document and the date. The citation is part of the output. No unsourced claims. No paraphrase that drifts.
It respects every rule. Reg FD. NI 51-102. NI 43-101. Section 21E. Rule 10b-5. The securities framework that governs your work is encoded in the engine and applied automatically before the output ships.
It is audit-ready by design. Every interaction is logged. Every output is traceable. The audit trail is a default behavior of the system. Not something assembled retroactively when someone asks.
That is the bar. It is a structural property of the system, not a feature list. The same four attributes apply across every output. Investor reply. Release. Board memo. Peer comparison. ESG footnote. The use cases differ. The standard does not.
The asymmetry has teeth
The asymmetry above is not theoretical. It has teeth. They are already cutting.
When a generic AI tool drafts a forward-looking statement that turns out to be wrong, the class action names the company. Not the AI vendor. When a generic AI tool sends a single investor information beyond the approved talking points, that is selective disclosure. Reg FD. When the CEO and CFO sign their Section 302 certification without knowing what AI is touching disclosure controls, the certification is a misrepresentation.
The plaintiffs' bar is already requesting AI-generated communications in discovery. The training data. The logs. The prompts. Companies without an audit trail produce a defective record. The defective record becomes the evidence.
D&O carriers are surveying AI governance in renewal applications. Companies without a policy face exclusions. Officers eat the gap personally.
The SEC's Division of Enforcement has an AI task force. Their job is not to wait.
The cycle is already running. Boards are asking. Counsel is drafting policy. Audit committees are reviewing AI exposure. Institutional investors are scoring it. Plaintiffs' lawyers are probing it in depositions.
Every public company arrives at the same compliance-grade standard. Some arrive by choice. The rest arrive in court.
Move first or get moved on.
The market is not waiting
The legal exposure is one layer. The market is the other.
Three things are already true. Your competitors run AI through their investor communications. Investors run AI through your filings. Retail platforms push AI-generated research to millions of users every week.
None of those mechanisms wait for your approval. You did not pick them. You did not vote on them. You cannot turn them off.
They are forming opinions about your company. Capital is moving on those opinions. Every day. With or without your involvement.
This is happening this quarter. Not next year. Not after the SEC writes formal guidance.
The companies that move to compliance-grade AI first will set the standard. Everyone else gets evaluated against it. On terms they did not write. At valuations that have already priced in the lag.
Where to begin
A few starting points. No budget. No board resolution. No procurement cycle.
Find out what AI your team is using. Not an audit. A conversation. The two or three people in the company who write things that go out under your name. What tools are they reaching for? Where? On what content? Most leaders are surprised by the answer. Visibility is the first move.
Get familiar with the four criteria above. They are the vocabulary you need to evaluate the stack you have. The pieces you are missing. The questions to ask the next vendor who calls.
Have one conversation with counsel. Not to draft a policy. To understand what they are watching. What they would advise if asked. Thirty minutes. The readiness curve moves more than most leaders expect.
Map where the risk sits. Generic AI exposure shows up in different places depending on the issuer.
Investor email. Press release drafting. Social media posts. Earnings Q&A prep. Board materials. Disclosure review. Peer research. ESG and sustainability writeups. Internal memos that touch material information. Identify which your team is touching with AI today. Start there.
None of this is a procurement plan. It is orientation. The point is not to push you into a decision this week. It is to make sure you are already moving when the revolution arrives. Whether it arrives through your competitors. Through your investors. Through your counsel. Or through your regulator.
Why I built this
I have spent a career in public markets. I have built and taken tech companies public. I saw what AI was going to do to public companies. The opportunity was obvious. So was the problem nobody was solving.
AI for public companies has to clear two bars at the same time. Accuracy and compliance. Off-the-shelf AI does neither.
Generic AI optimizes for plausibility. Not accuracy. It hallucinates. It paraphrases when it should cite. It cannot tell the difference between a footnote in your 10-K and a paragraph from a competitor's blog post. Use it to draft an investor email and you have introduced unverified content into your continuous disclosure record.
Enterprise AI built for law firms and hedge funds was built to read about public companies. Not to write as them. Tuned for billable hours. Not continuous disclosure. The closest those tools came to your company was as a research subject.
Build AI that clears one bar and not the other and the issuer pays. Build AI that clears both and the issuer wins. That is the engineering problem I started Versance to solve in the spring of 2023.
It is the hardest thing we have built. It still is.
Accuracy and compliance pull against each other at every layer of the stack. Tighten the rules and the system stops being useful. Loosen them and it stops being defensible. The work is to hold both at the same time. Without compromising either.
The choice
If you run a public company, you are already in this story.
AI is reading your filings. Summarizing your news. Comparing you to peers. Answering investor questions. Shaping how the market understands your company.
You do not get to decide whether AI enters the capital markets.
It already has.
The choice is whether the AI telling your company's story is one you picked and can defend. Or one operating in your name that you did not pick and cannot defend.
That is the line.
The companies that move first will make disclosure easier to file and access, investor communication easier to scale, and their story easier for retail and global investors to understand.
The companies on the right side of this operate with leverage their competitors cannot match. They reach investors their peers cannot reach. They build trust at scale.
There is compliance-grade AI. There is everything else.
The revolution is not coming.
It is here.
George Fleming is the Founder and CEO of Versance Technologies. versance.ai