AI workflow discovery: what to bring to the first meeting

A productive AI discovery meeting should end with a workflow the buyer and builder can both describe. Invite the person accountable for the result and the person who handles the awkward cases. In a small firm they may be the same person. In a larger organization, bring the administrator who understands access to the systems involved.

Bring a normal case and a difficult one

Suppose a regional distributor wants help with requests for product substitutions. Bring a redacted request the team resolved easily, one the team escalated and one it declined. Include the incoming email, the source people checked and the final decision. Remove names, commercial terms and personal information unnecessary for the discussion. Screenshots or invented examples can stand in when real records cannot be shared.

Walk backwards from the accepted outcome. An inside sales representative might say the task is “find a replacement.” The operations manager might explain that the proposed replacement must also fit a specific installation. That extra detail changes the design. The system may be able to assemble a shortlist while a specialist confirms compatibility.

Leave with a written boundary

Ask the group to finish four sentences together. “The work starts when …” establishes the trigger. “The agent receives …” defines inputs. “The person responsible sees …” identifies a usable output. “Stop and ask for help when …” captures exceptions.

Put unresolved questions next to named owners. If access to the catalog belongs to a manufacturer, identify who can confirm permitted use before the team estimates integration work. If the specialist travels, name a second reviewer. Send the one page back to the people who actually handle requests. Their corrections are inexpensive at this stage and valuable when it is time to compare proposals.

Put this into practice

Bring a sample of the work and the tools your team uses. We can help define a first project, who reviews it, and how to tell if it works.

Explore AI implementation · Discuss your project with Escencion