Blog

AI Turned One-Way Decisions Into Two-Way Doors. Discipline Matters More Than Ever.

Back to blog

AI Turned Some One-Way Doors Into Two-Way Doors. Identity Is One of Them.

There was a time when choosing an identity provider felt permanent.

Auth0. Azure AD (now Entra ID). Okta. Cognito.

Once you picked one, you designed your entire system around it. Token shapes, scopes, claims, SDKs, login flows, admin tooling. Changing later meant months of rework, risk, and user pain.

That made identity a classic one-way door decision.

You thought carefully because being wrong was expensive.

AI changed that calculus.

Identity Used to Be Sticky for a Reason

Historically, identity providers embedded themselves everywhere. Frontend auth flows. Backend middleware. Role models. User metadata. Admin processes. Documentation. Even support workflows.

Switching wasn’t just a technical task. It was organizational. You had to retrain teams, rewrite docs, re-educate customers, and migrate data without breaking trust.

So teams optimized for safety. They picked something “good enough” and lived with the tradeoffs.

That friction enforced discipline.

What AI Changed

With tools like Claude Code, Copilot, and similar systems, the mechanics of switching identity providers are dramatically cheaper.

You can now:

  • Translate middleware from Auth0 to Entra quickly
  • Rewrite token parsing and claims mapping in minutes
  • Scaffold parallel auth flows to compare behavior
  • Generate migration scripts and adapters on demand

From a code perspective, what used to take weeks now takes days or hours.

That turns identity from a one-way door into something that feels reversible.

And that’s where teams get into trouble.

Reversible Doesn’t Mean Free

Yes, you can pivot faster now. But identity decisions still leak into places AI doesn’t clean up for you.

User expectations.

Support workflows.

Security reviews.

Compliance assumptions.

Operational muscle memory.

AI lowers the cost of trying, not the cost of living with the consequences.

When teams realize they can “always switch later,” they start deferring judgment. They prototype Auth0, experiment with Entra, maybe layer both, and convince themselves they’re staying flexible.

What they’re often doing is accumulating ambiguity.

Idea Velocity Cuts Both Ways

This is the real shift AI introduced.

The risk used to be choosing wrong.

Now the risk is never choosing at all.

High idea velocity makes it easy to spin up alternatives, but it also makes it easy to leave systems half-committed. Identity becomes abstracted “for now.” Temporary adapters stick around. Edge cases multiply.

Nothing breaks immediately.

But coherence erodes.

Discipline Matters More Than Ever

The teams that benefit most from AI tooling aren’t the ones that try every option. They’re the ones that use speed to arrive at conviction faster, not to postpone it.

They still treat certain decisions as one-way doors by choice, even if tooling makes reversal possible. Not because reversal is impossible, but because constant reversibility prevents systems from settling.

With identity, that often means:

  • Choosing one provider as the source of truth
  • Designing clean boundaries instead of leaky abstractions
  • Defining explicit criteria for when a switch is justified
  • Accepting short-term discomfort to avoid long-term drift

AI makes the switch possible.

Discipline determines whether it’s worth it.

A Better Mental Model

Instead of asking, “Can we undo this later?”

Ask, “What happens if we don’t commit long enough to learn?”

AI collapses the cost of experimentation.

It does not collapse the cost of indecision.

Some doors are now technically two-way.

That doesn’t mean you should keep walking back and forth.

Closing Thought

Choosing between Auth0 and Azure Entra ID used to be a permanent bet. Today, it’s far more reversible from a code standpoint. That’s real progress.

But progress doesn’t remove the need for judgment. It raises it.

Speed is no longer the scarce resource.

Clarity is.

Have a project in mind?

Lopie is a senior-led software studio in San Diego that builds custom software and AI systems. If you have a project in mind, tell us about it.