Webbit

How to Choose an AI Automation Partner: 12 Questions to Ask Before You Sign

Twelve questions that separate dependable automation partners from hype, plus the red flags to watch for before you commit.

Choosing an AI automation partner is a trust decision. They'll touch your customer messages, your records and sometimes your money. Many agencies and consultants now offer AI automation, and the quality varies a great deal. These twelve questions help you tell them apart.

Before you start looking

Describe the process you want to improve in a short paragraph.
List the systems involved.
Name who in your team will own the result.
Decide what success looks like in your own terms, such as time saved or enquiries answered the same day.

Twelve questions to ask

How will you understand our process before quoting? A good answer involves reviewing the workflow, systems, volume and exceptions.
What would you not automate yet, and why? Honest partners can name tasks that are too risky or too unclear.
Who owns the workflows, prompts and accounts at the end? They should run in accounts you control, with documentation you keep.
What access will you need, and can it be limited? Look for least privilege, separate accounts and read-only access first.
Where will a person review or approve? Human approval should cover customer-facing and consequential actions.
How will we see what the automation did? Expect logs, alerts and summaries.
What happens when it fails? Ask about alerts, fallbacks and support response times in writing.
Where does our data go? Ask about providers, storage, retention and regions.
Can we speak to clients with similar projects? Real references, with permission.
How is the work scoped and priced? Defined phases, inclusions, exclusions and running costs.
How will you hand over and train our team? Documentation, a handover date and optional support.
How do we pause or stop it? A simple switch, and no lock-in.

Red flags

A fixed price before anyone has seen your workflow.
Promises of perfect accuracy or guaranteed results.
Everything lives in the provider's accounts.
No questions about exceptions, data or approvals.
Vagueness about what is out of scope.
Pressure to sign quickly.
A portfolio of demos but no running production systems.

Comparing providers fairly

Criteria	What to compare
Understanding	The questions they ask about your process.
Scope	Inclusions, exclusions and assumptions.
Safety	Access, review, logging and data handling.
Support	Response times, handover and ownership.
Evidence	References and examples running in production.
Cost	Build and running costs over the first year.

Start with a small paid step. A short audit or a single-workflow pilot shows how a partner works, communicates and documents before you commit to something large.

Common ways to work together

Model	What it is	Best for
Audit or discovery	A short paid review of your process, systems and risks, ending in a written scope.	Not knowing where to start, or wanting an independent view.
Pilot workflow	One workflow built and tested with a clear success measure.	Proving value before committing to more.
Fixed-scope project	A defined build with deliverables, a timeline and acceptance tests.	A workflow you already understand well.
Ongoing support	Monitoring, fixes and improvements for a monthly fee.	Workflows that matter to daily operations.

Reading a proposal

Is the scope written clearly, with inclusions and exclusions?
Are the assumptions stated, such as access to systems and the volume expected?
How will you know it is finished? Look for acceptance tests.
How are changes requested, priced and approved?
Who owns the workflows, accounts and documentation?
What does the support period include, and what happens after it ends?

Ask us all twelve

We're glad to be asked every one of these questions. Our automation audit is built as that small first step: we review your process, systems and risks, and describe a realistic first scope, or tell you it isn't worth building. Read how we think about safety in AI automation without the risk, and about price in how much AI automation costs.

Frequently asked questions

Do I need an agency, or can I do it in-house?

If you have technical capacity and time, simple workflows can be built in-house. Specialists help most with complex, sensitive or business-critical workflows, and with getting a first project live without learning every lesson the hard way.

How long does a first project take?

It depends on scope. A focused, single-workflow project is usually much quicker than a broad programme. A good partner scopes the timeline after reviewing your process, and puts it in writing.

What should the contract include?

Scope and deliverables, ownership of workflows and accounts, access arrangements, a support period, data protection terms and how either side can end the arrangement. Get legal advice for your situation.

Should I choose a provider in my own country?

Not necessarily. Clear communication, time-zone overlap, references, and a clear approach to data protection matter more than location.

Home · Blog · Book an Automation Audit