Last year, conversations with executives were about what AI capabilities you have. This year the opening line has converged on three questions. We reviewed hundreds of customer conversations from the first half of the year, and almost every enquiry maps to one of these three.
Question one: when will we see a return on this spend
Customers used to ask whether something was possible; now they ask how long until payback. The wording changed because more buyers have been burned before. The standard line about results in three months no longer convinces anyone.
Our habit is to work the numbers in detail. For a support scenario we look at how many conversations the current team handles each month and what that costs, estimate the share AI can take over, and then give a range rather than a single figure. A conservative six to nine months, with four in an aggressive case, earns more trust than promises of quick results.
Question two: whose hands is the data in
Data sovereignty was the most frequent concern this year. The typical wording is: my customer records and business data go into your system, so how is security guaranteed?
There is no sales line that gets around this; only policy and product can answer it. Encrypted storage, tiered permissions, a private deployment option and a contract that states data ownership stays with the customer. Our contract template has a dedicated page of data ownership clauses, and once customers read it everything else is easier.
Question three: who maintains this after you leave
Two years ago a project ended at delivery. This year customers immediately ask what happens next. Models age, knowledge bases need updates, business changes require prompt adjustments, and an AI project without operations degrades within three months.
That is why our proposals now include an operations package: knowledge base maintenance, model performance checks and monthly review reports after launch. Some customers ask whether we are simply selling long-term revenue, and we admit we are. More importantly, an AI project without operations wastes the customer's money and damages our reputation.
Rational procurement is good for teams that do the work
A rational market shrinks the space for selling concepts and rewards teams that can deliver results. Our strategy at ZClaw has stayed simple: make one scenario work thoroughly, then expand. If you are evaluating AI vendors, ask these three questions exactly as written; the quality of the answers will filter out about half the candidates.