Even with no one on staff whose job is process improvement, you can pick one piece of repeat work, turn it into an repeatable routine, and find something to cut that same week. Here is the order to work through.
3 min read ·
The first candidate is not a new account. It is the one job a member of staff handled the same way several times last week: copying inquiries into a spreadsheet, pasting the same reply, chasing the same missing field. The rule is already in their head. Explaining it takes five minutes, and those five minutes are the first draft of the operating standard.
It comes back often, the rule for it fits in one sentence, and the material needed to answer it sits in a document rather than in someone's memory. Miss any of the three and it is too early. McKinsey put the impact of generative AI squarely on activities that require understanding natural language, and estimated those activities at about a quarter of all working hours.
McKinsey, The economic potential of generative AI, 2023
Money leaving, a contract being fixed, personal data being changed, a message going outside the company. Those four reach a person before they happen. In the EU, GDPR Article 22 gives a person the right not to be subject to a solely automated decision with a significant effect on them, and where such a decision is permitted the controller must offer human intervention. The United States has no general federal equivalent; Colorado and California each grant a right to human review of automated decisions from January 2027.
GDPR Article 22 (2018); Colorado SB 26-189 and California's CPPA ADMT rules, both effective 1 January 2027
Replacing the tool spends the first month on training. Put intake and an approval queue on top of the mail, the messaging channel, the store and the payment record already in use, and nobody has to learn anything. What changes is not the screen but the handling time and the number of inquiries nobody answered.
Today's first-response time, today's unanswered count, how often staff had to correct a draft. Without those, nobody can say afterwards whether anything improved, and an impression takes the place of a measurement. Three numbers are enough. The point is to measure the same way twice, not to measure well.
Requests that arrive differently every time, judgments that shift with the day, evidence that lives in a different place for each person. That work needs a standard before it needs automation. Reverse the order and you have not automated anything, you have built a machine for sending the wrong answer faster.
A few short pieces you can read next, from the same operating standard.
RPA follows a fixed procedure. A Velros AI ops team reads the situation, classifies it, and prepares a draft. Here is how to tell which work belongs to which.
Open pageThe work to hand off is not some grand project, it is the small task that repeats every week. Here are the signals for spotting that task in your own company.
Open pageInquiries land scattered across WhatsApp Business, Instagram DMs, Google Business Profile, phone, and your online store. This is the operating design that pulls them into one flow of intake, classification, drafting, and approval.
Open pageOperations assessment
Working from the channels and files you use today, we settle which work should be handled first, where a person has to approve, and which metric will show whether it worked.
Get an assessment