← All examples
AI reliability

Eight unreliable AI providers, one product that stays up

A product that generates video, images, and music depended on AI providers that fail, slow down, and change constantly. One dependable system now sits between the product and all of them.

Client
Broadcast media company
Timeline
Twelve months
Providers
Eight
Product changes
None
01

AI providers move. The product cannot.

Outages are routine, the best model changes every few months, and every provider works differently. None of that should reach the product team.

WhatBeforeAfter
A provider failsThe product failsWork moves to the next one
Switching AI modelsRebuild the productChanged behind the scenes
Adding a capabilityA new projectOne new field
Provider quirksLeak into the productHandled in one place
Stuck requestsHang indefinitelyTime limit, then reroute
Tracking a requestGuessworkTraceable end to end
Before and after
02

How we did it

  1. 01

    One simple agreement

    The product sends one message describing what it has and wants, and gets one answer back.

  2. 02

    Every provider behind one door

    Each provider’s quirks are handled in its own adapter, never in the product.

  3. 03

    Health decides the work

    Providers that fail are sidelined, and win work back when they recover.

  4. 04

    Five generations, no disruption

    The video model changed five times in a year. The product never changed.

03

One moment from the work

Built for bad days

“What happens when a provider goes down?”

Nothing the customer sees. A failed attempt costs the time it took, then the work moves to the next healthy provider.

Delivery routes
25
Model generations
5
Product changes
0
04

What any team can take from this

  • Put one door in front of vendors

    Your product should not know which vendor did the work.

  • Let health decide

    Never switch a vendor off by hand. Let the scores do it.

  • Describe what you have

    The request says what it has; the system decides the job.