How to brief an AI or automation specialist so you get a clear quote

Describe the business problem before choosing an AI solution

Start with the task that causes delays, repeated effort or mistakes. Explain who does it, who is affected and why it matters, rather than opening with a request for a chatbot or a custom AI tool.

Describe the result you need without prescribing the technology. If staff repeatedly copy enquiry details between systems, the problem is duplicate handling, not necessarily a missing AI assistant. The specialist can assess whether simple automation, AI or a process change makes sense.

Define what good looks like in terms you can check. Include what must remain under human control, particularly when work involves customer communications, payments or sensitive information.

• Which task is difficult, and where does it get stuck?
• What should staff or customers be able to do differently?
• What must the system never do without someone approving it?
• How will you decide whether the completed work is acceptable?

Map your current workflow, volume and exceptions

Write down the current steps from the initial trigger to the finished outcome. A simple numbered list is enough. Include who handles each step, which information they use and where they save the result.

Show the real process, including workarounds, rather than the version in an outdated procedure. Mention missing information, duplicate records and jobs that need a manager’s judgement. These exceptions often influence the amount of work involved.

Describe how much work passes through the process and how often it happens. Include busy periods, deadlines and whether tasks must run immediately or can wait for a scheduled batch.

Attach a blank form, a sanitised sample or a simple diagram if it helps. Remove customer and employee details before sharing, and label anything that is only an example rather than a complete description.

List your software, account access and data requirements

List every system involved, including spreadsheets, email, accounting software, booking tools and shared folders. Common tools such as Microsoft Excel or Google Workspace can be named as examples, but include the actual products your business uses.

Note your subscription plan where known, who administers each account and any existing integrations. Access to an integration can depend on the plan, permissions or approval from another provider. Flag anything the specialist needs to investigate before confirming the quote.

List the logins or access permissions the work may require, but do not send passwords by email or put them in the brief. Where possible, arrange individual user invitations with only the permissions needed, and agree how temporary access will be removed.

• Identify where data comes from, where it goes and who may access it.
• Flag personal, confidential or sensitive information before sharing samples.
• Ask how data will be stored, retained and deleted. Check privacy obligations with a qualified adviser where relevant.

Set your budget range, timing and maintenance expectations

Share a realistic budget range privately with the specialist. Separate what you can allocate to setup from what you can sustain for subscriptions, usage and support. This helps them suggest a suitable scope rather than guessing what you can afford.

Explain your preferred completion date and why it matters. Distinguish a genuine deadline from a preference, and mention when your staff can answer questions, provide access and test the work. Those dependencies can affect delivery.

Decide who will look after the automation once it is running. Someone needs to notice failures, update instructions and review changes when connected software behaves differently.

• Name a business contact who can make decisions and approve changes.
• Identify who will monitor the process and handle exceptions.
• Ask which ongoing costs are included, estimated or charged separately.
• Consider a smaller initial scope if the full project is not yet clear.

Ask the specialist about approach, risks and handover

Ask the specialist to explain their proposed approach in plain language and connect it to your business problem. They should distinguish confirmed facts from assumptions and identify what needs testing. If discovery work is needed, ask what it will deliver before committing to a build.

Discuss failure scenarios as well as normal operation. For AI, ask how unreliable outputs will be checked; for automation, ask what happens when a connection breaks or information is missing.

• Why would you use this approach, and is there a simpler option?
• What could go wrong, how would we know and what is the manual fallback?
• What testing, staff training and documentation will you provide?
• What support is available after handover, and what does it exclude?
• Who owns and controls the accounts, workflows, code and documentation?
• Can another specialist maintain the work if we change providers?

Agree the scope, quote and delivery terms in writing

Before work starts, turn the discussion into a written agreement. Make sure both sides understand the deliverables, exclusions, assumptions and responsibilities. A clear quote should show what is included and what could trigger additional work.

Record the price, payment arrangements and delivery schedule, including any milestones and acceptance checks. Agree how proposed changes will be assessed and approved before extra work proceeds.

• Confirm required access, sample data and decisions your team must provide.
• Document account ownership, handover materials and ongoing maintenance responsibilities.
• Clarify the difference between fixing defects and adding new features.
• Record support arrangements and any separate recurring costs.
• Seek advice from a qualified adviser if contractual or ownership terms are unclear.