A few customers complain that an electronic device is overheating. Each report lands in customer service under a slightly different description. Then a journalist asks about the same problem. Only when someone looks back across the records does a potential safety issue become clear.
It is a hypothetical example, but the failure it illustrates is familiar to communications teams: warning signs can sit in different parts of an organisation for weeks before anyone connects them. Another alert on its own would not necessarily solve that problem.
Jeremy Seow, president and co-founder of Sequencr AI, a communications intelligence technology and consulting company, wants PR teams to move beyond periodic monitoring reports. In a written interview with ContentGrip, Jeremy outlined how AI agents could investigate emerging issues continuously, and why people still need to decide what deserves action.
Table of contents
- An overheating complaint can hide in plain sight
- AI can keep investigating between briefings
- One product change, several different warning signs
- A correct rebuttal can still be the wrong move
- Twenty mentions can be noise
- Test AI on the daily issues briefing
An overheating complaint can hide in plain sight
In Jeremy’s illustrative overheating scenario, customer service initially handles each complaint as an isolated case. The wording varies, so the reports are not necessarily tagged as the same fault. A journalist’s question finally prompts the business to search its own records more carefully.
The issue has been visible in pieces. Customer service has complaints; the communications team has a media inquiry. Neither team initially sees the shared pattern. By the time the organisation recognises a possible product safety problem, it may have lost weeks in which it could have checked the evidence and considered its options.
Jeremy said, “When a problem is identified late, those options tend to narrow considerably.” Late recognition leaves fewer choices, especially when safety concerns demand a response from teams beyond PR.
AI can keep investigating between briefings
Most monitoring still begins with somebody deciding what to search for, an analyst reading the results and a report being delivered. The next inquiry waits for another request. Jeremy believes agentic AI could help keep the investigative work moving between those reporting cycles.
He imagines a system that retains relevant information about stakeholders, previous findings and organisational vulnerabilities. When new information appears, it could check earlier evidence, ask follow-up questions and flag a change in the risk picture. Analysts would then spend more time testing what the evidence means and advising colleagues, rather than restarting research each morning.
The proposal is more ambitious than faster media summaries. As Jeremy put it, agentic AI offers the possibility of “investigative work continuing without someone having to initiate every step.”
He also draws a firm limit. Well-defined research and analysis workflows are already feasible when appropriate information is available and outputs are evaluated. Decisions in ambiguous or high-stakes situations remain harder to delegate, particularly when the facts are incomplete.
One product change, several different warning signs
A second hypothetical example shows why an issue’s significance can change as more people become involved. Imagine a technology company making a routine platform update. The product team sees an improvement. A civil society group worries about its effect on a particular set of users. Separately, a policymaker discusses platform responsibilities, while journalists in another market investigate similar practices.
Each development could look manageable in isolation. Together, they raise a different question: does the change conflict with commitments the company previously made to regulators or to the people using its product?
Jeremy’s suggested response moves the discussion beyond a holding statement. Communications would need to bring in policy, legal and product colleagues, review what the company promised, and examine how the update could affect different stakeholders. Depending on the findings, the organisation might need to change the product decision itself.
A monitoring system that tracks mentions without connecting those earlier commitments could miss the development that matters most.
A correct rebuttal can still be the wrong move
Jeremy offers another hypothetical: a stakeholder publicly criticises an organisation with which it once had a close relationship. The records appear to support the organisation’s position. A public rebuttal looks defensible.
An AI agent could investigate the past relationship, earlier disagreements and the political or regulatory climate. That context could help the team understand how a response might be received. Yet a seasoned adviser may know about private conversations, personalities and lost trust that the system cannot see.
A forceful public answer could make reconciliation impossible. Silence could let an inaccurate allegation spread. Jeremy suggests a third path might be better: engage the stakeholder privately first, while preparing a factual public response if needed.
Jeremy said, “Sometimes, it means recognising that leadership is asking the wrong question.” The decision is about the relationship and consequences, not simply whether the facts permit a rebuttal.
Twenty mentions can be noise
Communications teams can feel pressure to react when a board member forwards an article, an executive shares a screenshot or negative mentions spike. Jeremy warns against treating the volume of activity as the measure of danger.
Jeremy said, “Twenty mentions can be noise, while one comment from the right stakeholder can materially change the situation.”
A better alert would ask whether an issue has reached a new audience, whether an influential stakeholder has become involved, or whether fresh evidence changes the company’s exposure. Safety, misconduct and regulatory obligations may require immediate attention even without any visible surge in discussion.
Those criteria should guide what an AI agent investigates and escalates. They also give advisers grounds to tell leadership when an issue has been examined and no intervention is warranted. More dashboards and alerts are useful only when the team can explain which development changes the decision.
Test AI on the daily issues briefing
Jeremy’s recommended starting point is the daily issues briefing, a familiar task that can be compared with an existing process. He would begin with three to five priority issues, identify what leadership needs to know about each, and agree on the stakeholders, sources and circumstances that warrant deeper investigation.
AI would undertake initial research and analysis. The communications team would check the findings before distribution, insisting on concise evidence and a clear account of what remains uncertain. For a few weeks, it would run the new briefing alongside the old one.
The comparison would cover time spent, relevance and accuracy, developments missed, unnecessary escalations, and whether leadership found the output more useful. Jeremy would look especially for evidence that the team investigated more issues, gave sharper advice or caught a warning it previously would have overlooked. Saving time alone would be an incomplete result.
Outside communications, Jeremy collects records and makes playlists. His musical hobby stretches back decades, and he now curates SIGNAL, a series of carefully sequenced 13-song selections. On his Substack, The Judgement Layer, he writes about music, life and work, returning to songs whose meanings change as he grows older.
