OFFICIUMoh-FIK-ee-um
Workforce Blueprint & Starter Agent
Officium is Latin for a duty, the root of “office.” We write down every duty in the system: what each agent does, what operators run, and what your people decide.
The whole system scoped and priced in writing, with one working agent from the start.
A fixed fee, agreed in writing before work begins. Already ran a Role Diagnostic? Its code and Role Card are the starting material.
Included: the system design, acceptance criteria, a priced operations plan, and one starter agent running in an agreed mode.
Is this for you?
- Businesses ready to move one role, or one team’s repeatable work, to managed agents
- Owners who want a fixed price for the full system, and a running agent to judge before they commit
- Work that crosses several systems, such as a CRM, an accounting system, and a shared inbox
- Teams that want people kept in the loop, and that written into the design
The problem
Moving a role to agents is an operating decision, not a software purchase.
You have seen that agents can carry part of a role. The next question is the whole picture: which tasks move, who watches them, and who still decides.
Those answers decide the price, the risk, and how your team experiences the change. Written down, they can be reviewed, priced, and approved before anything larger is built.
You also want to see an agent working on your own material before you commit. A plan is easier to trust when part of it is already running.
How we solve it
Every duty written down, priced, and one agent already working.
We start with the people who do the work today. Each task in the role is listed and assigned to an agent, an Ergon operator, or someone on your team, with the reason for the choice.
While the design is written, we build one starter agent for a bounded task and run it on your work. It runs in shadow beside your team, or with approval on every action, so nothing changes without a person.
You receive the design, the acceptance criteria, and two prices in writing: a fixed price to build the rest, and a monthly price to operate it. The decision to go ahead is yours.
Teams by industry
Where a Workforce Blueprint fits.
Typical roles and starter agents across industries. Illustrative examples, not client results.
- Accounting firm
Client intake, scoped across the season
The Blueprint maps the intake coordinator’s duties from document requests to filing. The starter agent sorts uploads in shadow mode while the team keeps filing. Staff accountants confirm every sort.
- Distributor
Order entry across email, phone notes, and the portal
Order intake is split into agent, operator, and sales duties. The starter agent drafts order lines from emailed orders for a person to approve. Pricing exceptions stay with the sales lead.
- Medical practice
Referral and scheduling coordination
The design covers referrals, insurance checks, and appointment requests. The starter agent prepares referral summaries, and front-desk staff approve each one. Clinical decisions stay with clinicians.
- Property management
Maintenance coordination for a portfolio
Tenant requests, vendor dispatch, and follow-up are mapped. The starter agent triages requests and proposes a vendor. The property manager approves every dispatch and every spend.
- Insurance agency
Renewal preparation for account managers
The Blueprint assigns renewal tasks: gathering policy data, drafting summaries, and chasing documents. The starter agent drafts renewal summaries. Account managers approve every client communication.
- Logistics
Shipment exception follow-up
Exceptions from carriers are mapped by type and owner. The starter agent drafts customer notices for delayed shipments. Operators send routine notices within policy; your team handles claims.
- Professional services
Time entry and billing preparation
The design covers time capture, billing rules, and invoice drafts. The starter agent prepares draft invoices from time entries. A partner approves every invoice before it is sent.
- Home services
Booking, dispatch, and follow-up
Calls, bookings, reminders, and reviews are assigned. The starter agent proposes technician and slot for each request. The dispatcher confirms bookings until a policy is approved.
What you receive
The duty map
Every task in the role, assigned to an agent, an operator, or one of your people, with the reason for each.
The operating model
Who runs the agents day to day, what operators review, how exceptions reach your team, and how often you receive a report.
Architecture and integrations
The agents, the systems they read and write, and one authoritative source of truth for each kind of record.
Permissions and approval points
What each agent may do at each permission level, and the named person who approves each consequential action.
Acceptance criteria
Each outcome with its test, threshold, test data, and who runs it, agreed before the rest of the system is built.
A starter agent
One basic agent built for a bounded task in the role and running on your work, in shadow beside your team or with approval on every action.
A priced plan
A fixed price to build the rest of the system, and a monthly price to operate it, so the full cost is known in writing.
How it works
Discover the role
We interview the people who do the work, sit in on a working session, and map the systems, volumes, and rules involved.
Assign the duties
Each task goes to an agent, an operator, or your team. Your manager for the role reviews and approves the duty map.
Build the starter agent
One bounded task is built first and runs on your work in the agreed mode, so your team sees real output early.
Design and price
The architecture, permissions, acceptance criteria, build price, and monthly operating price are written up together.
Review and decide
Walk through the plan and the starter agent’s run records. You decide whether to build, operate, or pause.
A sample duty map
Every task has an owner before anything is built.
An excerpt for an accounts payable role. Yours is written for your work, with your systems and your approvers.
| Task | Agent | Ergon operator | Your team |
|---|---|---|---|
| Read incoming invoices | Extracts vendor, amount, dates, and lines | Reviews low-confidence reads | None |
| Match to purchase orders | Proposes a match and flags differences | Resolves differences within policy | Decides differences outside policy |
| Enter in the accounting system | Drafts the entry for posting | Posts entries the policy allows | Approves the posting policy |
| Schedule payment | Prepares the payment batch | Checks the batch against the rules | Approves every payment |
| Answer vendor questions | Drafts replies from the records | Sends routine replies, escalates disputes | Handles disputes |
Payments always wait for a named person on your team. The duty map records that in writing.
How the Blueprint is checked
The measures your approver agrees.
Each item can be checked against the written design, the starter agent’s run records, or the agreed acceptance criteria.
- Every duty has an owner
- Each task in scope is assigned to an agent, an operator, or a named person, and the role’s manager approves the duty map.
- The starter agent runs as agreed
- It runs on the agreed task, in the agreed mode, and its run records show input, output, corrections, and cost.
- Approval points are named
- Each consequential action has a named approver, and each agent has a written permission level.
- Acceptance is written as tests
- Each outcome has a test, a threshold, the data it is tested on, and the person who runs it.
- Both prices are in writing
- A fixed build price and a monthly operating price are stated together, with what each includes.
Common concerns
What owners ask before they sign.
- Your team’s experience
- People whose work changes are interviewed first and see the duty map before it is approved. Their knowledge shapes the rules the agents follow.
- Systems access
- Access is requested per system, at the lowest permission that does the job, and agreed in writing before any connection is made.
- The starter agent’s mode
- It runs in shadow or with approval on every action. Moving it to a higher permission level is a later decision your approver makes.
- Estimates versus measures
- Volumes and time savings in the design are labelled as estimates until the starter agent’s run records measure them.
- Scope changes
- New roles or tasks discovered along the way are listed separately. Adding them goes through a change order, so the fee stays clear.
Who does what
Who builds, who runs, and who approves.
Your team
- Name the role or team in scope and its manager
- Make the people who do the work available for interviews
- Grant agreed, limited access to the systems involved
- Name the approver for consequential actions
- Approve the duty map, the acceptance criteria, and any next step
Ergon
- Interview the team and map every task in the role
- Design the agents, integrations, and permissions
- Build and run one starter agent on your work
- Write the acceptance criteria with you
- Price the build and the monthly operations in writing
Afterward
Where clients go next.
Most take one of these routes. You choose, and nothing starts without your approval.
Build and operate
Approve the build against the agreed criteria, then move into Managed Agent Operations at the monthly price in the plan.
Operate the starter agent first
Keep the starter agent running under managed operations while you decide on the rest of the system.
Run it with your own team
Use the design and documentation to build and operate the system internally.
Pause here
Keep the written design for later. Handover terms for the starter agent are set out in the agreement.
Before you start
What to have ready.
- The role or team in scope, and the person who manages it
- A list of the role’s weekly tasks, even a rough one
- The systems the role uses, and who administers each
- Approximate volumes: how much work arrives in a typical week
- Your Role Card and Diagnostic code, if you have them
Related services
Who approves what
Operators run the agents. Your approver decides.
Ergon engineers build and test each change, and Ergon operators run the agents inside these levels. Your named approver authorizes consequential actions and releases.
| Level | What the agent may do | Who approves | Example |
|---|---|---|---|
| Read | The agent reads the sources it was granted, and nothing else. | The access is approved once, when the system is scoped. | Reading open invoices in the accounting system. |
| Draft | The agent prepares work, and a person decides whether it is used or sent. | An Ergon operator or someone on your team reviews each draft before it leaves the system. | A reply to a customer, waiting in the review queue. |
| Pre-authorized routine write | The agent makes a bounded, reversible change that a written policy already allows. Every change is logged. | A named owner on your team approves the policy in advance. Operators watch the changes, and anything outside the policy waits for a person. | Correcting a duplicate customer address under the approved conflict rule. |
| Consequential action | Payments, commitments to customers, changes to a record of truth outside an approved policy, and anything that cannot be undone. | A named person approves each one, every time. If nobody approves, nothing happens. | Paying a supplier, or changing a price a customer was quoted. |
| Release | A new model, agent, prompt, or rule version goes live. | The client’s named approver approves each release after the agreed tests pass. Contractual acceptance of the work is a separate step. | Moving an improved extraction model from testing into daily use. |
What “accepted” means
Measured against criteria you agree to in advance.
- Every task in scope is assigned to an agent, an operator, or a named person on your team.
- The starter agent runs on the agreed task in the agreed mode, and its run records are shared with you.
- Each outcome in the acceptance criteria has a test, a threshold, test data, and a named evaluator.
- You receive a fixed build price and a monthly operating price in writing. New or changed scope goes through a change order.
Request a Workforce Blueprint
Tell us about the role.
We reply with the questions that shape scope, then a proposal with the fixed fee agreed before work begins.
- An engineer reads every request
- A reply by email within two business days of receipt
- No obligation until scope and price are agreed
Not ready to scope this? Ask an engineer first: a free 15-minute call about which parts of the role agents could take on.
Where this fits
- Learn How it works
- Ask free Ask an engineer
- Trial one role Agent Role Diagnostic
- Scope the system Workforce Blueprint & Starter Agent
- Run every day Managed Agent Operations
Not ready yet? Agent Role Diagnostic. After this: Managed Agent Operations.
Questions
What is a starter agent?
A basic agent for one bounded task in the role, such as sorting intake or drafting follow-ups. It runs on your work in shadow mode or with approval on every action, so your team judges real output before the full build.
How is the Blueprint priced?
A fixed fee, agreed in writing before work begins. It depends on the roles, systems, and integrations in scope.
Do we have to buy managed operations afterwards?
No. The Blueprint prices it, and the decision is yours. You can also run the system with your own team from the documentation.
What happens to the starter agent if we pause?
What is delivered with the agent, including its runbook, is stated in the Blueprint. Ownership and handover terms are set out in the Master Services Agreement.
How long does it take?
It depends on scope. The proposal gives a schedule alongside the fee, before you commit.
Can the Blueprint cover more than one role?
Yes. Scope is set in the proposal, and the fee reflects the roles and systems included. One role is the usual first step.
Do we need to have done a Role Diagnostic?
No. It helps, because its code and Role Card become starting material, but the Blueprint begins with its own discovery.