FR : Afficher cette page en français Describe the task

Run and improve

A monthly retainer for what you have in production: monitoring, adjustments, and re-testing when the model underneath changes.

AI tools can drift without anyone touching them. The provider replaces a model, an instruction that worked starts to get in the way, a cost creeps up. The failure is often quiet, so it tends to be found late.

Describe the task in writing
Example: a month on the retainer
  1. Runs on its own 1. Your systems are monitored Runs, failures and cost per step.
  2. Alerts someone 2. Drift or a model change is found Instructions age; models get replaced.
  3. Done by us 3. Re-tested against your checks Adjusted until the checks pass again.
  4. Waits for a person 4. You read the monthly report Cost per step, checks passed, what changed.
  • Runs on its own
  • Waits for a person
  • Alerts someone
  • Done by us

What is included

  • Monitoring of the workflows, assistants and agents we built for you; a workspace on your own computer is supported in writing, not monitored
  • Adjustments, requested and answered in writing, as your work changes
  • Re-testing against your test set when the model your setup runs on changes
  • For builds we monitor, a short written report each month: cost per step where it is measured, checks passed, what changed
  • Improvements from our own tooling as we make them

How it goes

  1. The retainer starts when a build goes live, or after a workspace is installed. What it covers for your setup is listed in the quote.
  2. You write when something needs changing. We answer in writing.
  3. For monitored builds, you receive the report each month.
  4. New tools and new workflows are quoted separately.

Price

A flat monthly fee, quoted in writing. Support is in writing. New builds and live training hours are separate quotes.

Questions

What is not included?

New tools, new workflows and live training hours. Those are quoted on their own.

How fast do you answer?

We aim to answer within one business day, Monday to Friday, in writing. That is a target, not a guaranteed response time, unless your quote sets one.

What if nothing changes in a month?

Then the report is short. Monitoring continues through the month, and re-tests run when the model your setup runs on changes.

Describe the task

What happens next

  1. A person reads it.
  2. You get a written first read, usually within one business day.
  3. If it is worth doing, a written quote follows.

Prefer email? Write to info@acceleratech.ca.

How we handle what you send: Privacy Policy

Other services

Product names are trademarks of their owners. Naming a tool means we use it, not that its vendor partners with or endorses us.