When AI agents collaborate, who owns the result?

Ali Alkhafaji, chief executive of Apply Digital, requires one named person to answer for every output of the firm's AI agent teams, named before deployment starts.

Ali Alkhafaji, chief executive of the customer experience firm Apply Digital, wrote a commentary in InformationWeek about who answers for work that several AI agents produce together. Each new client engagement at the firm starts with at least half of its team made up of AI agents. When one agent makes a poor call, the next may not have the information to catch it, and the error can compound before a person sees it.

• The article cites a report by the research firm Gartner in which 13 per cent of 360 survey respondents believe they have the right governance in place for AI agents.
• Responsibility attaches to one named person, not to a team, a framework or a review process, and that person must exist before deployment starts.
• A record at each handoff between AI agents captures what the agent was trying to do, what data it had, how confident it was and what action it took.
• The firm decides before deployment what gets logged at each boundary between agents, so that a review after an incident can trace the chain of reasoning.

An organisation running several AI agents in one workflow can check whether each workflow has one named person who answers for its output. The piece is a commentary by the head of a firm that sells this kind of work, and it reports no measured results.

Derived by Working Surface from the article; more in the Playbook. Source line: If you can't name them, you're not ready to deploy.

3 November 2026: teams working with AI agents are moving human review from each output to the plan, the shared instructions and the rules the agent follows beforehand, and naming in advance who answers for the result.