Amperity Pér AI Agent Turns Customer Data Into Approved Action

Amperity Pér is a new AI agent designed to close the gap between what a brand knows about its customers and what it does next. According to the company announcement, Amperity Pér takes a team from a question about customers, campaigns, or audiences to an answer, a recommendation, and, with human approval, action inside the tools that team already uses.

The launch, dated October 26, 2026, arrives as marketing and analytics teams wrestle with a familiar bottleneck. Dashboards can surface a trend, but turning that trend into a decision still tends to pass through data requests, manual analysis, meetings, and separate activation tools. Pér is presented as a single governed workflow that replaces that chain, rather than as another dashboard to check.

At the centre of the product is Amperity's customer data platform. The company says its patented identity resolution builds a trusted foundation by unifying fragmented customer records into governed profiles. Pér works from those profiles, which matters for the kind of agent it claims to be. A generic chatbot can draft a plausible answer from a prompt. An agent grounded in resolved identity can, in principle, say which customers changed, why a recommendation fits them, and what should happen next.

From insight to approved action

According to Amperity, Pér can investigate what changed in a customer base, identify the customers affected, explain the evidence behind a recommendation, and prepare an audience or an action for approval. That sequence is deliberate. The agent does not simply act. It builds a case, surfaces the signals and business rules behind it, and then waits for a person to review, refine, and approve the next step.

That approval step is the product's clearest answer to the trust problem facing agentic software in the enterprise. An agent that can move audiences into marketing tools is powerful precisely because its mistakes travel fast. Keeping a human in the loop before activation limits that blast radius, while still removing the slower handoffs that used to sit between analysis and execution.

Once an action is approved, Pér moves it into enabled marketing and business workflows rather than asking teams to copy results between systems. It can also monitor execution results and recommend mitigation strategies as outcomes change, according to the announcement. The pitch, in short, is not autonomy for its own sake. It is a shorter, auditable path from a question to a decision that a human has signed off.

Why grounded agents are the next test

The broader agent market has spent the past year racing toward tools that can act, not just answer. Recent launches on this beat show the same pattern from different angles: agents that work through messaging apps, agents that run locally, and agents built for regulated enterprise workflows. Each one raises the same question for buyers. What data is the agent allowed to trust, and who checks its work before it touches a customer?

Pér answers that question with governance rather than speed alone. For data and technology teams, the agent works within identity-resolved customer data, configured permissions, and workflows the enterprise already manages, according to Amperity. It does not create a separate AI layer that sits apart from the customer record. It extends that record into decisions about campaigns, audiences, and experiences.

That distinction may matter more than any single feature. Many AI tools can summarise a segment. Fewer can show their working, respect existing permissions, and hand a decision back for approval in a form a marketer can actually use. Amperity argues that completeness and governance are what let an agent move from interesting demo to measurable economic outcome, because the data underneath it can immediately be put into action.

Early users point to proactive auditing

Early customer voices in the announcement stress proactivity over reactivity. Elise Keefe, Senior Manager of Business Intelligence at MacKenzie-Childs, said in reported remarks that Pér helps the team move beyond broad segments toward individual customer journeys, cutting the grunt work of searching for answers so staff can spend more time on decisions. For retention work, she pointed to identifying the moments that matter and moving from insight to approved action through one connected workflow.

Geoff Clark, CRM Manager at Johnnie-O, pointed to a different benefit: real-time auditing. According to his reported remarks, Pér helped identify overlaps and inefficiencies across segments and campaigns and act on them at scale, with changes made on the fly improving the accuracy and performance of long-standing segments. The contrast he drew was blunt. Most AI tools react to a prompt, he said. Pér, in his account, acts proactively on what it finds.

Those testimonials should be read as launch claims, not independent results. They do, however, reveal where Amperity expects the product to earn its keep: not in answering one-off questions faster, but in continuously watching segments and campaigns for waste, overlap, and drift, then bringing a fix forward for approval.

What to watch next

Three questions will decide whether Pér becomes a staple or another agent experiment. First, how well does the approval workflow hold up under volume, when dozens of recommendations arrive each week and approvers start rubber-stamping? Second, how transparent are the explanations when a recommendation is wrong, and can a team trace it back to the profile data that produced it? Third, how cleanly does the agent fit into tools beyond Amperity's own ecosystem, where most enterprise work still happens?

Amperity's background gives it a head start on the data side. The company says more than 400 brands worldwide rely on its platform, including retailers, airlines, and hotel groups, and it was founded in 2016 with offices in Seattle, New York City, London, and Melbourne. That installed base means Amperity Pér does not have to win the customer-data argument from scratch. It has to prove that an agent on top of that data is trustworthy enough to approve, day after day.

For teams evaluating agentic tools, the launch is a useful benchmark. Ask not only what the agent can do, but what it is grounded in, what it must get approved, and what audit trail it leaves behind. On those three tests, Amperity is betting that a governed agent beats a merely clever one. Readers following the agent ecosystem can also see how the same approval and grounding questions play out in other recent launches, including AI agents moving into banking and insurance workflows, a personal agent that works through text messages, and a voice model that runs locally rather than in the cloud. More detail on the launch itself is available in the Amperity announcement as reported by LiveNews.