AI Agents

Ten agents run the Observe, Plan, Act, Report loop. Each one has a fixed scope, fixed inputs, and a fixed output type.

The ten agents

Acera Labs runs ten specialist agents. Each one has a narrow, defined job. They do not overlap.

Agent	What it does
Analytics Agent	Monitors customer segments, detects anomalies, generates weekly intelligence reports
Data Science Agent	Manages model training, monitors drift, retrains propensity and CLV models
Media Strategy Agent	Analyses FFM results, identifies saturation points, recommends budget reallocation
SEO Agent	Monitors organic search performance, identifies content gaps, tracks ranking changes
Engineering Agent	Monitors system health, data pipeline performance, and infrastructure metrics
Insights Agent	Synthesises outputs from all agents into executive summaries and board-ready reports
Campaign Operations Agent	Ingests, normalises, and analyses campaign performance data from all channels in real time
Audience Agent	Manages audience health, detects fatigue, and optimises targeting across channels
Anomaly and Alert Agent	Dedicated watchdog using Bayesian change-point detection across all metrics
Financial Impact Agent	Translates every marketing metric into financial language for CFO-ready reporting

What each agent does in detail

Analytics Agent

Runs on a scheduled cycle. Reads your customer segment data and CLV records. Detects statistical anomalies in segment size, trend, and health. Produces weekly intelligence reports and segment dashboards. Outputs are visible in the Command Centre agent activity feed.

Data Science Agent

Monitors your propensity and CLV models for performance degradation. When model accuracy drops below the configured threshold, it triggers a retrain. Produces model reports and drift alerts. You can see training history and cost per run in AI Agents then Data Science Agent.

Media Strategy Agent

Reads your FFM (full-funnel model) results and saturation curves. Identifies which channels are above or below their saturation half-point. Produces budget reallocation recommendations ranked by expected incremental return. Works closely with the BYOMM connector if you bring your own Marketing Mix Model.

SEO Agent

Monitors your organic search performance against configured keywords and topic clusters. Identifies content gaps where paid spend is covering ground organic could own. Tracks ranking changes week over week.

Engineering Agent

Watches system health, data pipeline freshness, and infrastructure metrics. Fires alerts if a connector sync fails, if a model run exceeds expected duration, or if an API endpoint degrades. Outputs are incident reports and health dashboards.

Insights Agent

Runs last in the cycle. Reads outputs from all other agents and synthesises them into a single executive summary. Produces board-ready reports that combine media performance, customer intelligence, and financial impact in one narrative. Most teams use this as the opening slide in their weekly review.

Campaign Operations Agent

Ingests campaign performance data from all connected ad platforms in near real time. Normalises metrics across platforms into a single schema. Produces the pacing alerts and optimisation recommendations you see in the Command Centre channel health table.

Audience Agent

Monitors the health of your customer segments. Detects audience fatigue (rising frequency, falling CTR). Identifies lookalike opportunities seeded from high-value segments. Flags cross-channel audience overlap that is wasting budget.

Anomaly and Alert Agent

A dedicated watchdog that runs independently from the other agents. Uses Bayesian change-point detection to identify statistical breaks in any metric. Classifies each anomaly by severity (critical, warning, informational). Generates a root-cause hypothesis with the alert. This is the agent most likely to send you a notification outside the morning cycle.

Financial Impact Agent

Translates marketing metrics into P&L language. Calculates CAC payback period, EBITDA contribution from media spend, and lifetime value impact of current campaigns. Produces CFO-ready summaries that do not require marketing interpretation.

Where to see agent output

All agent activity surfaces in the Command Centre activity feed on the main dashboard. Each entry shows which agent ran, what it found, and a link to the full output.

You can also browse individual agents in AI Agents. Each agent card shows:

Current status (active or idle)
Last run time
Cost this month (AI token spend)
Capabilities and key outputs

Agent run costs

Each agent run consumes AI tokens. The cost is shown on each agent card and in the billing section. Costs vary by agent. The Campaign Operations Agent and Insights Agent typically have the highest token spend because they process the most data.

Token costs are passed through at cost. There is no markup on model spend.

// Next steps

Core concepts