ERPAIIntegration
Menu
Case studiesGovernanceLessonsWatch it workThe journeyHow it's builtLive healthContact
Questions we answer
Can we trust AI to act on its own?Governance and guardrailsWho allowed the AI to do that?Identity travels with every requestCan someone talk the AI into something risky?Risk-graded autonomyCan we see who did what, and why?One thread from request to resultWhere are the keys kept?One vault, no shared secretsWhy did the AI bill jump, and who spent it?One counted doorwayIs every task getting the right model?Right model, right taskWhat if our AI vendor goes down or raises its price?No single point of supplyAre we paying for tools that do the same job?One tool per jobDoes the AI forget everything between conversations?Governed shared memoryIs this producing real work, or just answers?The digital factoryWhich work should run itself?Managed, not scriptedCan we get more from the Copilot licences we already pay for?Copilot proposes, the platform verifiesWill people actually use it?Meet people where they workHow much time do people lose sorting inbound information?Intake to insight, untouchedCan the same platform run a real business?One governed foundation for every productWhat happens when it breaks at night?Detect, fix, escalateWhere is the list of what is open, and who owns it?One list, closed with evidenceCould we recover if we lost everything?Proven by restoringCan you prove it to an auditor?Challenge before acceptanceDoes it get better over time, or just older?Every lesson becomes a ruleWill this still work after the next upgrade?Clean Core for AIDo our documents match what is actually running?Documentation as a by-product

Portfolio / Cost-Smart AI

The business question · Model routing

“Is every task getting the right model?”

The concept

Right model, right task

Each task goes to the smallest model that still gives a top-quality answer.

How it works
Each kind of work has its own lane: hard reasoning goes to the strongest model, light work to low-cost models, with a set fallback order.
Why it matters
Quality goes where it counts and cost comes down everywhere else, without anyone choosing models by hand.
The result
Each kind of work runs on the right-sized model, and switching a lane is a configuration change, not a rebuild.

See it workingRight model, right taskAnimation

How the gateway routes work: each request takes the lane that fits, and switches lanes on its own when a provider stops answering. Built in code from the real lanes, models and Teams message.

The business problem

One model for every job is either wasteful or weak.

Most AI programs send every request to the most expensive model, or to whichever model a developer picked first. The platform handles very different work: long reasoning, short summaries, approvals, searches. One model for all of it would be either wasteful or weak.

▶ Watch the answer: Why does the AI bill keep climbing?

Why this concept

Task-based lanes behind one doorway, with a defined fallback order

Matching the model to the task is the single biggest lever on both quality and cost.

Also considered

  • One premium model for everything
  • Let each agent pick its own model

How it works

Right model, right task

Task

Reasoning, summary, approval or search

Lane

Matched to the kind of work

Strong model

Hard reasoning only

Low-cost model

Light work

Fallback order

Takes over when a lane is down

What we put in place

  • Defined lanes by task: hard reasoning to the main model, light work such as titles, summaries and approvals to low-cost models.
  • Made sure a top model is never used for something like sending an approval.
  • Set a fallback order so a lane switches automatically when its model is unavailable.
  • Added a quota guard that moves work to a backup lane before a limit is reached.

The result

Before

One model for every job is either wasteful or weak.

After

Each kind of work runs on the right-sized model, and switching a lane is a configuration change, not a rebuild.

I treat AI models as a portfolio to be managed, not a single product to be bought.

In your organization

Better answers where they matter and lower cost where they do not.

Related case studies

Cost control

“Why did the AI bill jump, and who spent it?”

The conceptOne counted doorway

Read the case study

Not tied to one provider

“What if our AI vendor goes down or raises its price?”

The conceptNo single point of supply

Read the case study

Contact

Looking for someone to lead AI from pilot to production?

I would be glad to walk you through any part of this live and talk about the role you have in mind.

Email meLinkedInOne-page brief

j.walters@erpaiintegration.com

Go deeper: Lessons · The journey · How it’s built · Live health