We built an AI reporting agent that reads their tools and writes the client report every week.

The setup

Weekly client reporting is the operating heartbeat of a PR firm. It is also the part of the week that grows linearly with the client list — every additional client is another set of analytics dashboards to pull from, another batch of coverage clips to log, another deck to format and send.

At Leverage With Media, the team was spending most of Sunday and Monday morning preparing the week's reports. The work was high-skill enough that it could not be handed off to junior staff without quality dropping, and low-leverage enough that it kept the senior team away from the work they were actually hired to do.

What we built

We built an AI reporting agent that sits across the team's tools. It pulls coverage from their media monitoring stack, performance numbers from analytics platforms, social signals from each channel, and brand mentions from the press-tracking service. Then it drafts the weekly client report in the firm's own template — narrative, highlights, the gaps worth flagging, the data tables — and prepares the email that ships it.

The team reviews each report before it goes out. The agent has learned the difference between a coverage win that deserves a callout and a passing mention that should sit in the appendix; over the first few weeks the senior team trained it on the calls it was getting wrong. Now the edits are mostly cosmetic.

Nothing leaves the agency without a human approving it. The agent does the assembly. The team does the framing.

What changed

Roughly twelve hours each week are no longer spent stitching reports together. Every client gets their report on schedule — the Monday-morning scramble simply does not happen anymore. The senior team gets that time back for the actual PR work: relationship building, pitch craft, strategy with clients.

The first month was bumpy. The agent did not yet know which client cared about share-of-voice vs reach, which one wanted competitor mentions called out, which one wanted the data flat and quiet. By month two it had absorbed enough of the team's judgment that the drafts needed less work than the originals had.

Where it goes from here

The same agent now handles monthly executive summaries on top of the weekly cadence, and is being extended to draft the proactive recommendations that used to live in the senior team's heads. The pattern keeps compounding because the firm is no longer hiring against client growth — they are hiring against the work that actually requires senior judgment.