The most common mistake we see is not choosing the wrong tool. It is choosing the wrong first workflow: the most interesting process rather than the most valuable one. The method below takes twenty minutes and produces a shortlist your whole team can agree on.
List candidate workflows
Write down every repetitive, conversation-shaped task in your business. Examples: answering pricing questions, qualifying new enquiries, booking first appointments, sending reminders, chasing quiet leads, logging call notes, routing support tickets. Aim for eight to twelve.
Score each one from 1 to 3 on five questions
- Frequency: does it happen daily (3), weekly (2) or occasionally (1)?
- Pattern: does it follow the same steps almost every time (3), often (2) or rarely (1)?
- Cost of delay: does a slow response lose a customer or appointment (3), cause friction (2) or barely matter (1)?
- Available content: do you already have approved answers and rules (3), partly (2) or nothing written (1)?
- Low judgement: can a checklist handle it (3), mostly (2) or does it need experience and discretion (1)?
Add the scores. Anything at 12 or above is a strong first candidate. Anything below 8 belongs with your team for now.
Two worked examples
After-hours property enquiries: frequency 3, pattern 3, cost of delay 3, available content 2, low judgement 3. Total 14. Start here.
Negotiating a discount with an existing client: frequency 1, pattern 1, cost of delay 2, available content 1, low judgement 1. Total 6. Keep this with your team, and let the agent prepare the account summary before the call instead.
Why one workflow, not a platform
Starting with one workflow proves value quickly, teaches your team how the agent behaves and gives us real conversations to learn from. It also protects you: if the first version is not right, you find out with one process, not ten. Expansion is easy once the first agent has earned its place.
What happens next
Bring your top-scoring workflow to a 30-minute discovery call. We map the process, confirm whether an agent is a good fit, and send a written plan with realistic milestones rather than a promise of a fixed launch date. If the honest answer is “not yet”, we will say so.