On October 13, 2026, Amperity introduced Pér, an AI agent built directly into its customer data platform, designed, in the company's words, to help brands move from customer questions to approved actions inside their existing software. The launch was reported the same day by IT Brief and covered widely across the marketing technology press. It is Amperity's most explicit move beyond identity resolution, the capability the CDP was built on, and it lands on a question every marketing team recognizes: we have the unified customer view. Now what?
The boring version of that question has been the industry's default for years: a dashboard gets refreshed, a manual data request gets queued, an analyst writes a memo, a meeting happens, and an activation tool eventually runs something. Pér compresses that entire relay into a single process, and the details of how it does that say a lot about where marketing automation is going.
What Pér actually does
According to IT Brief's October 13 report, users can ask Pér natural-language questions about campaigns, audiences, and customers. The agent then investigates changes, identifies which customers are affected, explains the basis for its recommendation, and prepares an audience or action for approval before any step is taken. Two design choices deserve attention:
- Investigation before recommendation. Pér is framed as explaining why it is suggesting something, positioned against the black-box auto-optimization pattern that has made marketers rightly skeptical of "AI-powered" buttons.
- Approval as a first-class step. Proposed actions go to human review, refinement, and approval before being sent into enabled marketing and business tools. The agent prepares the action; people release it.
The agent is also designed to monitor execution results and suggest mitigation strategies while campaigns are live, letting teams adjust course mid-flight rather than waiting for the post-mortem. Amperity cites early uses in retail and apparel marketing operations; Elise Keefe, Senior Manager of Business Intelligence at MacKenzie-Childs, described moving "beyond broad segments," using Pér to "reach trusted, actionable findings faster, so we can spend more time making decisions and less time searching for answers" for the company's retention work.
The insight-to-action gap, closed by an agent
Every company that has ever bought a CDP, a BI tool, or an analytics suite has lived the same sequence: the data layer gets fixed first, and the action layer never quite catches up. Amperity's launch implicitly concedes that assembling a reliable customer view was never the bottleneck. The bottleneck is everything that happens between the insight and the campaign, a stretch of work that spans dashboards, data requests, analysis, meetings, and activation tools, and that is precisely the work Pér is built to replace.
For technology and data teams, the product works within identity-resolved customer data, configured permissions, and workflows the organization already manages. The stated goal is to avoid creating a separate AI or data layer, extending unified profiles into campaign and audience decisions instead. That is the right instinct. A marketing agent that requires its own parallel data stack would recreate the exact fragmentation CDPs exist to remove.
| The old relay | With an agent in the CDP |
|---|---|
| Dashboard refresh | Question asked in natural language |
| Manual data request | Agent investigates changes itself |
| Analyst memo | Recommendation with its basis explained |
| Meeting to decide | Human review and approval gate |
| Activation tool setup | Prepared audience sent into existing tools |
| Post-campaign analysis | Mid-flight monitoring with mitigation options |
What transfers to the SMB: the pattern, not the product
Amperity is an enterprise CDP; most of our readers will never buy one. But the launch pattern is directly transferable, and it is the pattern, not the vendor, that matters:
- One source of truth first. Pér can be trusted to investigate precisely because it queries one identity-resolved view with configured permissions. If your customer data lives in five tools that disagree, no agent fixes that; it just argues with itself faster. Consolidate before delegating.
- Approval gates, everywhere. The most encouraging design choice in this whole launch is that Amperity kept humans at the release point. For SMBs, the equivalent is simple discipline: an agent can prepare email sends, audience changes, and budget shifts, but a named person approves anything that touches customers or money.
- Mid-flight adjustment beats post-mortem. Most small marketing teams measure exactly once, at the end. Live monitoring with suggested mitigations is now a product-level expectation, not an analyst luxury. Even lightweight agent setups can check in on a campaign daily and recommend changes.
The reduction of marketing-to-analytics handoffs is a real productivity claim here. IT Brief notes the tool is intended to take over routine work so analysts keep more time for specialist judgment, which matches the pattern we have argued all along (and again this week): pay for verified deliverables and saved decision time, not software seats.
The skeptical read
Three caveats worth holding onto the same day the announcement makes the rounds. First, this is a launch announcement; the same-day press has no independently verified performance results, and we do not expect any for months. Treat every "could reduce handoffs" claim as a hypothesis your own pilot must test. Second, agent mediation cuts both ways: when an agent summarizes the analysis and preps the action, teams tend to stop reading the underlying data. Keep the interfaces open so at least one human still looks at the raw numbers on a schedule. Third, the consolidation wave continues, Amperity joining ZoomInfo Agent Teams and every major martech vendor bundling agents this quarter, which means buyers should weigh exit costs before adopting any agent layer that learns their workflows. An agent that accumulates your procedural knowledge is valuable precisely because it is hard to replace.
The marketers who win the next cycle will be the ones who pair agent speed with the stubborn habit of asking the agent to show its work. If you want to build that kind of insight-to-action loop, with approval gates your team actually trusts, we design exactly that.
Read next:
