Operations line
One line, two jobs. The Client Agent turns meetings and delivery notes into updates, actions, and health checks. The Report Agent turns the week into one brief: what moved, what is behind, what needs a decision. Nothing leaves the business without a person.
Summary
Runs
Client updates, meeting recaps, onboarding steps, account health checks, weekly briefs, and claim checks
Reads
Meeting notes, delivery files, the KPI ledger, pipeline exports, and the client workspace
Writes
Drafted updates, action lists, health flags, the weekly brief, and dashboards with source paths
Guardrail
No client sees a message and no number reaches a report without a person
Three lines
Two jobs in this line
Direct answer
An Operations Agent runs the two jobs that keep a service business honest: client work (onboarding, meeting capture, updates, health checks, proof requests) and reporting (KPI memory, weekly reviews, dashboards, claim checks). It prepares and files the work; people approve anything a client or the public sees.
Price
One build price and one run price, named before any work starts. No hourly billing, no retainer. Model and tool fees are separate and published on the pricing index.
Build, fixed
$18,000
About six weeks
Run, monthly
$1,200/mo
Month to month, stop any time
A real artifact
We do not publish client numbers without a name, a figure, and a date window. What we can show is our own operating output: this is one real artifact from our rollup routine, redacted, exactly as it was produced.
Coverage
9 contacts total, 0 attributed; 1,689 events total, 0 attributed. Attribution still broken; sessions are the only signal this week.
What moved
What needs a decision
The attribution read-back is missing, so the kill-and-scale review cannot rank by pipeline. Fix the read-back before the next review, or rank by sessions and say so.
Where it helps
The first agent should have a clear job, known inputs, and a review path the team can keep using.
What it can run
These capabilities come from the public Conversion Skills patterns. The names under each card are the documented skills the build can draw from.
It captures a call or a meeting into decisions, owners, dates, and open questions, then files the recap where the team looks.
It drafts the status update from delivery notes and the last recap, in the client's language, and holds it for a person to send.
It runs the onboarding checklist, names the next step, and flags what the client still owes.
It reads activity, delivery, and tone across accounts and raises a flag with the reason before a quiet client becomes a lost one.
It turns the ledger and the week's notes into one page: what moved, what is behind, what needs a decision.
It builds the dashboard from the ledger and checks claims against sources, marking anything missing instead of inventing it.
How it works
Step 01
The agent reads the client files, the KPI ledger, and the review rules first, so every draft starts from the business instead of a blank page.
Step 02
It drafts updates, recaps, flags, and the brief, with sources attached wherever a number appears.
Step 03
Each artifact names the human move: send, fix, ask, or decide. Nothing goes out on its own.
Human gates
The agent prepares work and keeps the trail clean. People still approve the moves that can affect buyers, customers, money, permissions, or public claims.
Outputs
The useful output is not a magic answer. It is a reviewable artifact, a clearer next move, and a source trail for anything factual.
Client status updates, drafted and dated
Meeting recaps with decisions and owners
Onboarding checklists with the next step named
Account health flags with the reason
The weekly brief: moved, behind, decide
Dashboards and scorecards with source paths
Quick answers
Short answers for buyers comparing agent work, service scope, and human review boundaries.
Related agents
Most useful AI systems start with one agent, then connect the next repeated job once the first one earns trust.
Public standard
Conversion Skills is the public reference for these patterns: plain files, documented skills, proof-aware outputs, and review gates. The agent pages show how that public method turns into useful business systems.
