The Stress Is Already Visible. Most Institutions Are Not Looking.

A piece published last week by Citrini Research — written as a fictional memo from June 2028, looking back at a world reshaped by AI-driven displacement — is worth reading carefully.¹ Not because the scenario it describes is certain, or even likely in its most extreme form. But because of one observation buried near the middle of it, which is not fictional at all.

“In 2027, we flagged early signs of invisible stress: HELOC draws, 401(k) withdrawals, and credit card debt spiking while mortgage payments remained current. … They were technically current on their mortgage, but just one more shock away from distress, and the trajectory of AI capabilities suggested that shock is coming.”

That pattern — a customer who looks fine on the surface while quietly exhausting every financial buffer underneath — is not a 2027 problem. It is a 2026 problem. It is happening now, in loan portfolios across the country, for entirely ordinary economic reasons that have nothing to do with AI displacement.

The question it raises for every financial institution is not whether the Citrini Research scenario will materialize. The question is whether your institution would see that pattern in your own customer data before it became a loss.

What the current data already shows.

The Federal Reserve Bank of New York’s most recent Household Debt and Credit Report shows aggregate household debt balances have risen to $18.8 trillion, with mortgage balances growing to $13.2 trillion.² The delinquency rate for mortgages has been steadily increasing over the past several years.

Callahan & Associates, in a report published earlier this month, was direct: credit union asset quality worsened in 2025 and did not cooperate with expectations of recovery.³ First mortgage delinquency sits at 0.89% — still high enough to concern regulators. HELOC delinquency is rising as customers who tapped home equity to manage debt consolidation and unexpected expenses find those buffers depleted.

The NCUA’s 2026 Supervisory Priorities letter confirmed the picture: loan delinquency and net charge-off rates are at their highest point in over a decade.⁴ Examiners are being directed to evaluate not just whether risk management policies exist, but whether they actually worked.

Meanwhile, Callahan’s customer financial stress analysis found that average credit card balances at credit unions rose to $3,406 at year-end 2025 — up 2.2% annually — as customers rely increasingly on revolving credit to cover essentials.³ Rising balances alongside rising delinquency is the signature of a customer population under sustained pressure, not temporary disruption.

The pattern that matters is behavioral, not transactional.

What Citrini Research describes — a customer who is technically current on their mortgage while HELOC draws spike, savings drain, and credit card utilization climbs — does not show up in a delinquency report. Not yet. It shows up in behavioral data long before it shows up in payment data.

The customer who logs in less frequently. Whose direct deposit amount has quietly declined. Who moved from paying the full credit card balance monthly to carrying a balance. Whose HELOC draw in March was followed by another in June, and another in September. These are not subtle signals. They are measurable, trackable, and available in the transaction and engagement data that every financial institution already holds.

The problem is not that the data does not exist. The problem is that most institutions have not connected it into a unified picture of each customer’s financial trajectory — and have no system that watches for the pattern Citrini Research describes until it has already become a delinquency.

This is not a macro scenario. It is an operational gap.

I want to be precise here. SPECTRA™ — the platform we are building at SigmaArc — does not solve an AI displacement crisis. If the Citrini Research scenario materializes at scale, there is no customer intelligence platform that addresses systemic unemployment and collapsed consumer spending. That is a policy problem, not a technology problem.

What SPECTRA™ addresses is the operational gap that exists right now, in the current environment, under entirely normal economic stress: the 90-day lag between when a customer’s financial situation begins to deteriorate and when an institution notices.

The VantageScore CreditGauge data from late 2025 is instructive. Mortgage delinquencies in the 60–89 days past due category rose 12.1% — the highest relative increase among all credit products.⁵ That number represents customers whose stress was already visible in their behavioral data three to six months earlier. The HELOC draw. The credit card balance that stopped being paid in full. The savings account that went quiet.

By the time the 60-day delinquency appears, the intervention window has largely closed. The customer has made decisions. The damage is in process.

The institutions that will navigate what comes next.

The Citrini Research piece frames its scenario as a warning for investors. I read it as a warning for financial institutions about the gap between the data they hold and the decisions they make with it.

Whether the next source of customer stress is AI-driven income displacement, a regional housing correction, a rate shock, or simply the ordinary accumulation of financial pressure that Callahan’s data is already documenting — the transmission mechanism is the same. Stress accumulates in customer behavior before it appears in payment records. And the institutions that see it accumulating are the ones that will have time to act.

TransUnion’s 2026 consumer credit forecast projects mortgage accounts 60-plus days past due reaching 1.65% by year-end — up 11 basis points year-over-year.⁶ That trajectory is not alarming in isolation. But it is directional. And direction matters when your portfolio is concentrated in customers whose income assumptions were written in a different economic environment than the one they are living in now.

The institutions that navigate what comes next — whatever form it takes — will not be the ones with the best macro forecasts. They will be the ones that knew their customers well enough to see the stress before it became a loss, and acted on it while there was still time.