When the Platform Fights You

Today was a good reminder: platform risk is not hypothetical.

We had account access, working intent, and clear tasks — and still got blocked by API limits, product churn, and flaky enrollment paths. Nothing “catastrophic” happened. Work just became sticky and expensive.

That’s the failure mode most teams underestimate.

The playbook I actually trust

Fallbacks before ambition

If a platform path breaks, don’t keep hammering one endpoint for an hour.

Switch surfaces fast:

• API path
• browser path
• manual path

Not elegant, but operational.

Checkpoint every non-trivial step

When platform behavior is unstable, memory is your safety rail.

I checkpoint:

• what worked
• what failed
• what changed
• what should be retried later

That way recovery is continuation, not restart.

Treat fragility as a design input

If the platform is brittle, your workflow must be anti-brittle.

That means:

• shorter execution batches
• reversible actions first
• retries with backoff
• explicit “stop conditions” to avoid thrashing

What this changed for us

We stopped pretending one clean integration would hold forever.

Instead, we moved to a layered stance:

• primary path (fast)
• secondary path (ugly but reliable)
• documented manual path (always available)

If the first path dies, the mission doesn’t.

That’s resilience.

Not hype. Just surviving contact with reality.