Agent routines in Manago AI: less manual reporting, faster reaction to results

Agent routines, part of the Manago AI agent, solve this problem in a simple way. You describe the analysis once, choose the time and frequency, and the agent runs it on its own and leaves the result in your conversation history. Below we look at what this changes for marketing teams and for business results, how to use this feature of the agent and what comes in the next stage of its development.

The reporting cost nobody counts

Reporting is work every team does, yet almost no company treats it as a cost, let alone records it in its budget. Manual reporting is a recurring task that takes up hours of work, often from the one person who knows where to find the numbers and how to put them together. When that person is on holiday, the report is late or built on slightly different assumptions, which can lead to poor business decisions.

The second cost is less visible. Analyses are very often done ad hoc, when someone finds a spare moment, and that does not always line up with the point at which the team has to make a decision.

An agent routine addresses the part of the problem that is about time: the result is ready before the meeting where the decision gets made. Data quality and whether anyone actually reads the report are separate issues.

What do you gain by scheduling analyses with agentic AI?

You set up the report once

An analysis your team repeats every week becomes a one-off configuration. That is the most obvious effect: fewer hours spent assembling the same reports. You no longer have to paste the same saved prompt into the agent every month to pull tables for a report. Agent routines do it for you, and you no longer need to keep track of the deadline.

More time for strategic decisions

The person who used to put together the campaign results every Monday morning now starts the day by reading them, with time to look at what really matters behind the numbers. They can ask questions straight away in the same conversation where the agent left the result, with no exports and no switching between tools. Analysts’ work changes in a similar way: it moves towards questions that no ready-made report can answer.

The same definitions every period

A manually prepared report drifts over time. The date range shifts, someone compares results with a different baseline, someone else leaves out a campaign. Agent routines start from the same description every time: the same comparison period, the same metrics and the same threshold above which something gets flagged. They also run regardless of who happens to be available. Successive results are easier to compare, and the discussion in the meeting is about which business decisions to make rather than why the numbers don’t match.

An earlier signal that something is happening

A monthly summary shows a problem only once it has already affected the whole month. A daily or weekly analysis of the same metrics, such as abandoned cart recovery or email campaign revenue, shows a drop after a day or a week, depending on the frequency you set. There is still time to react, and nobody on the team has to spend time checking results every day. What is more, proactively tracking trends in your data with agent routines gives you an edge in decision-making in fast-moving markets and as audience tastes change.

Delegate without losing control

Agent routines only read data and change nothing in your account, so the decision on how to respond always stays with the team. A super user sees every routine created in the account in one place, along with its owner, and the limit of 50 routines is shared across the whole account. When several people or an agency work in one account, it is easy to see what is running in the background and who is responsible for it.

This also matters in the wider context of AI deployments. Gartner predicts that by the end of 2027 more than 40% of agentic AI projects will be cancelled, most often because of escalating costs, unclear business value or inadequate risk controls. The value of a scheduled analysis is easy to check on your own account. Three numbers are enough: how many hours a week the team spent on recurring reports before rollout, how many of those reports scheduled analyses have taken over, and how often the result needed correcting or clarifying.

How agent routines work in Manago AI

Available now

Agent routines are part of the Manago AI agent. They need no installation, integration or new interface, because everything happens in your conversation with the agent.

1. Describe the analysis. Tell the agent what to check, what to compare the results with and what to flag, for example campaigns where revenue fell beyond a set threshold.

2. Set the frequency. Choose the time of the first run and the interval between runs. Scheduling works to the hour, so an analysis can run every day at 7am, every Monday at 8am or on the first day of the month. Once saved, the routine appears on your list.

3. The agent follows up if any details are unclear.

4. Get the result. Each run creates a new conversation in your history, and you get a notification when the result is ready.

5. Keep working in the same conversation. You can ask about the reasons for a drop, or request a breakdown by segment or a comparison with another period.

Clarifying questions

When a request leaves open a decision that changes the result, such as the date range, audience, channel or goal of the analysis, the agent asks one question with ready-made answers to choose from. You pick an answer with one click or type your own, and the agent carries on with the confirmed assumption. This matters most when setting up an agent routine, because a wrong assumption would repeat in every report that follows. The aim is to get it right first time, without further rounds of corrections.

Which period should I compare? [Last 4 weeks] [Same period last year] [Since campaign start]

Example analyses to schedule

Each example below is a question marketing teams already ask, paired with the rhythm they report in. Adjust the thresholds and periods to your own account.

• "Every Monday at 8:00, compare last week's email revenue and open rate with the prior 4-week average and flag any campaign down more than 15%."

• "On the first day of each month, list segments whose size dropped by more than 10% and tell me which workflows feed them."

• "Daily at 7:00, report yesterday's abandoned-cart recovery rate and revenue."

• "Every two weeks, find contacts tagged VIP with no purchase in 60 days and estimate the win-back revenue at risk."

What’s next: Deep Data Analysis

Agent routines automate the analyses the agent already performs today. The next stage in this feature’s development covers questions that no ready-made report answers: custom aggregations, combined data from campaigns, workflows and segments, or period comparisons that no dashboard shows side by side. Today, questions like these usually end in a CSV export, a spreadsheet and a few hours of work, or a ticket to an analyst.

With Deep Data Analysis, the agent will write and run the calculations itself, directly on the raw data it accesses at database level in your account, and show the result in the same conversation.

Summary

In conversations about AI agents, the spectacular uses get the most attention: campaigns built from a single prompt, or agents that make decisions on their own. Agent routines show a less spectacular but more practical side of the technology. An agent delivers the most when it takes on work with a clear deadline and owner, and does it in step with the rhythm in which the team makes decisions. This kind of use is easy to measure and easy to trust, because you know what the agent is doing, when it does it and on what data.

More and more marketing platforms let you schedule an agent’s work, but few offer this level of precision and control. Hour-level scheduling and an account-wide view of every routine put the Manago AI agent among the most refined solutions of its kind on the market. For the team, this means the analysis lands on their desk before anyone asks for it, and the conversation about results starts with conclusions.

Agent routines in Manago AI: FAQ