AI Employees · Agent workforce

A role you would otherwise hire for

An AI Employee joins your business the way a person does. It has a name and a position description, sits in your Teams or Slack, gets access to your systems, reads years of company history in days and is doing real work in the same week.

It picks up its next task every thirty minutes around the clock, works to a performance contract and is measured on KPIs. Correct it once and the correction sticks. It does the work. You make the call.

In #accounts-receivable, AR-02 (Accounts Receivable Analyst) reports: Morning Dana. Three invoices went past thirty days overnight: Hartley & Co (AUD 4,180 incl. GST), Marsden Group (AUD 12,650 incl. GST) and Coastal Fitouts (AUD 2,300 incl. GST). Reminders are drafted in each customer's tone from their payment history. Queued at the gate; nothing goes out without your sign-off. Two files are attached, then Approve & send and Edit drafts buttons. Dana approves at 9:12. AR-02 replies: Sent. Logged to the audit trail as #F-20A917. Dana then asks: Thanks. Flag Marsden for a call if nothing lands by Friday, and update the cash forecast. AR-02 replies: Reminder set for Friday and the cash forecast updated with the three. The forecast note is queued at the gate and goes to the team on your sign-off. Same audit ID, #F-20A917, above Send note and Edit note buttons. In #people, HR-01 (HR Officer) reports: Morning Sam. Two new starters on Monday. Contracts are drafted from the approved templates, the onboarding checklists are built and the IT access requests are prepared for the position each one holds. Nothing issues without your sign-off. Two files are attached, then Approve & issue and Review contracts buttons. Sam approves at 8:40. HR-01 replies: Issued to both starters. Logged to the audit trail as #P-44C102. Sam then asks: Book their inductions with the head of department and send the welcome notes. HR-01 replies: Inductions booked for Monday at 9:00 and the welcome notes drafted. The notes are queued at the gate and go out on your sign-off. Same audit ID, #P-44C102, above Send notes and Edit notes buttons. In #projects, CA-03 (Contract Administrator) reports: Morning Priya. Progress claim 7 for the Wyloo camp works is assembled: quantities reconciled to the site diary, variations 12 to 14 attached, AUD 486,200 excl. GST. Held at the gate; nothing is submitted without your sign-off. Two files are attached, then Approve & submit and Open claim buttons. Priya approves at 8:20. CA-03 replies: Submitted to the principal's representative. Logged to the audit trail as #C-88D310. Priya then asks: Chase the unpaid claim 6 and put the retention figure in Friday's report. CA-03 replies: Reminder on claim 6 drafted and the retention line added to Friday's report. The reminder is queued at the gate and goes out on your sign-off. Same audit ID, #C-88D310, above Send reminder and Edit reminder buttons.

Pick a channel on the left to follow that agent. Illustrative. Three roles in three channels, each stopping at the same gate. Businesses, people and figures are fictional.

The bench, in full

Every role on the platform today, shaped to a job your business already hires for. Each one runs to a position description, stops at a sign-off gate, and writes to the decision log. Open any role to meet the agent on the Legare site.

67 specialist roles

13 disciplines

Any role that is missing is staffed to a brief.

01Executive & Strategy5

02Finance10

03Actuarial & Modelling7

04Risk, Compliance & Legal4

05Client Coverage & Sales2

06Property & Real Estate5

07Architecture & Construction11

08IT, Data & Engineering11

09Marketing & Creative4

10HR & Ops3

11Communications & Document2

12Research2

13Knowledge1

A role that is not on the bench. Every role here starts from a position description, and any role that is missing is staffed to a brief. Writing that brief is the part we do with you: the workflows the role owns, its success measures, the systems it reaches and the point where a decision comes back to a person. Bring us the role.

Hired like a person, started like a person

The fastest way to understand the workforce is to notice how little of it is new. Every step is one you already run for a new starter.

  1. 1Position description
  2. 2A seat in Teams or Slack
  3. 3Access to your systems
  4. 4Real work the same week

We write the position description with you: the objectives, the KPIs, the systems the role touches and the actions that need a person to sign off. The agent is hired against that document and measured against it. It joins the channels the role would sit in, so your people brief it and correct it the way they would a colleague, and it reports back in the same place.

It gets the access the position description says it needs and nothing more. Read access comes first; anything that writes to a record is gated until you have seen it work. In its first days it reads the policies, the contracts, the past correspondence and the shared drive nobody has tidied, so it starts with the context a long serving employee would have. Then the first tasks are narrow and every output is checked, and the scope widens as trust is earned.

Every thirty minutes, it asks what is next

An AI Employee is proactive against its objectives. On a thirty minute cycle it looks at what has come in, what it owes, what is overdue and what the KPIs say, then picks up the next task. Nobody has to prompt it. Overnight and over the weekend it keeps going, and on Monday the queue is clear.

  1. 01

    Layered memory

    Memory of your business, your people and its own work. It knows who approves what, which supplier is always late and how you like a report laid out, and it carries that forward from one week to the next.

  2. 02

    The best model for each task

    Chosen behind the scenes, so a routine reconciliation and a difficult piece of drafting are each handled by whatever does that job best. Nobody in your business has to pick a model.

  3. 03

    Correct it once

    A correction becomes part of how the role is done from then on. You never repeat yourself, and the agent does not make the same mistake twice.

  4. 04

    Your data stays yours

    Skills travel between deployments so every client benefits from the roles maturing. Your data stays inside your deployment and is never used to train external models.

It does the work. You make the call.

Reading, drafting, analysis and preparation run on their own. Anything that changes a record, sends money, reaches a client or leaves the building is proposed and held for a named person to approve, in the channel they already use. Every action is logged with the person who approved it, so the audit trail exists before anyone asks for it.

The gate is a design choice, and it is the reason a board can adopt an AI Employee with the same confidence it has in its existing controls. The full position, including the standards we build to, is on the ethics and governance page.

Never without a person

  • Sending money or changing a bank detail
  • Anything that reaches a client, a supplier or a regulator
  • Changing a record of account, a contract or a payroll entry
  • Reaching a system its position description does not name

Legare, delivered in Australia

Legare is the enterprise agentic platform the roles run on. MyBizz Solutions has been selected as one of a very small number of partners able to deliver it in Australia, and we sit beside the platform for the whole engagement: writing the position description, running the induction, tuning the role and taking the call when something needs a person.

The platform is built with input from over 750 researchers and a leading figure from TIME's Top 100 in AI. It is in production in over 60 companies globally, with around 250 AI Employees in action.

Australian deployments run on infrastructure in Australia. The roster, the role finder and the technical detail live on the Legare site.

A position description for every role

Objectives, KPIs, systems and the actions that need sign-off, written before anything is deployed.

In the channel you already use

Briefs, corrections and approvals happen in Teams or Slack, in the thread the work is in.

Auditable by default

Every action is recorded with what was done, when and on whose approval.

How an AI Employee is priced

Per role, per month

Quoted in writing after a thirty minute conversation

Each AI Employee is priced per role per month, with a once off activation fee that covers the position description, the induction and the first weeks of tuning. Billing is monthly, in Australian dollars, with GST stated on every invoice, and a role can be stood down at the end of any month.

No figure is published here because the activation work varies with the systems the role touches. The written quote sets out the monthly amount, the activation fee and exactly what the role will do, before anything starts.

What an AI Employee will not do

  1. 01An agent works to its position description. Ask it to do something outside that document and it says so and stops.
  2. 02Anything that changes a record, sends money, reaches a client or leaves the building is proposed and held for a person to approve. That is a design choice.
  3. 03Judgement calls stay with your people. The agent prepares the decision and the person makes it.
  4. 04It will be wrong sometimes, early on especially. A correction is permanent, so the second week is better than the first.
  5. 05It is a member of staff on a monthly arrangement. It can be stood down at the end of any month, and what it has learned about your business stays with you.

Questions we get asked about AI Employees

A software agent that joins your business the way a person does. It has a name and a written position description, sits in your Teams or Slack, gets the system access the role needs, reads years of company history in its first days and is doing real work in the same week. It runs on the Legare platform, which MyBizz Solutions delivers in Australia.

There are 67 roles on the bench across 13 disciplines, from a Bookkeeper and a Management Accountant to a Contract Administrator, a Bid Coordinator and a Chief of Staff. Any role that is missing is staffed to a brief, and every deployment names a Chief of Staff alongside the specialists.

The case we make is more output from the team you have. An AI Employee takes the recurring load, the re-keying, the month-end scramble and the inbox that never empties, and the work left to your people is the work that needs a person. We will not write a proposal whose case rests on a smaller team.

Reading, drafting and analysis run on their own. Anything that changes a record, sends money, reaches a client or leaves the building is proposed and held for a named person to approve, in the channel they already use. Every action is logged with the person who approved it.

Per role per month, with a once off activation fee that covers the position description, the induction and the first weeks of tuning. Billing is monthly, in Australian dollars, with GST stated on every invoice, and a role can be stood down at the end of any month. We quote in writing after a thirty minute conversation about the role.

Australian deployments run on infrastructure in Australia. Your data stays inside your deployment and is never used to train external models. Personal information is handled under the Australian Privacy Principles.

The first tasks are narrow and every output is checked, usually within the first week. The scope widens as trust is earned, which is the same way you would treat a new starter.

What agents build

Four roles, and what each one sets up

Inside its role an AI Employee builds the systems the job needs: on its own initiative where the work sits in the efficiency scope of its position description, or when someone in the business asks in Teams or Slack. These are the four roles we place most often, what each sets up, and a working example of each on sample data.

  1. 01Quoting and proposals
  2. 02Invoices and admin
  3. 03CRM and accounting
  4. 04Reporting and dashboards
  5. 05Beyond the four

Role 01 · Sales Coordinator AI Employee

Quoting and proposals

Works from your price book, your terms and the job details already sitting in your systems, and assembles a consistent, accurate quote or proposal for a person to check and send. Quoting is one of the first things the role takes on, because it sits squarely in the efficiency scope of the job.

Works with. Xero, common CRMs and job management systems, Google Workspace and Microsoft 365, with the system access its position description grants.

What it sets up

  • The pricing rules you already apply by hand: margins, options, inclusions, applied the same way every time
  • A structured intake, or a feed straight off the CRM or job system, so quotes assemble from records you already keep with no retyping
  • Your proposal format, and any other format you ask for in Teams or Slack the way you would ask a colleague
  • The order that follows: invoice, stock allocation and freight booking written from one entry and reconciled

What it fixes

Slow turnaround, inconsistent pricing and quotes that depend on one person being available. If a quote takes days because it waits for whoever knows the pricing, work is lost to faster competitors. The Sales Coordinator keeps the turnaround predictable and removes the retyping that causes pricing errors. Every quote still waits for a person to approve it before it reaches the customer.

Working example · Sales Coordinator

One entry instead of four systems

The order is typed once. The Sales Coordinator writes the invoice, the stock allocation and the freight booking from it, reconciles the three, and holds everything until a person signs off.

Demonstration · sample data

Try it: create the order, then sign off what the Sales Coordinator prepared.

One entry, three systems

The order is entered once and fanned out. A person releases it, software does the typing

Demo

New order

Customer
Sample customer 12
Item
Bracket kit, galvanised
Quantity
40
Value
$1,860.00
Ship
Thursday

Accounting

No invoice yet

Written from the one entry

Waiting

Inventory

No allocation yet

Written from the one entry

Waiting

Freight

No booking yet

Written from the one entry

Waiting

Reconciliation: runs once the systems are written

The order is typed once, here. The three systems on the right get written for you.

Role 02 · Accounts Payable Officer AI Employee

Invoices and admin

Captures supplier invoices, matches them to purchase orders, prepares the reconciliation and works through expense claims, and puts every payment in front of a person to approve. This is the repetitive, rule-based computer work owners most often lose their evenings to, and it sits at the centre of the role.

Works with. Xero and the other systems Australian businesses already use, with credentials held in managed secrets and your data excluded from model training under the platform's commercial terms. What it can reach is written into its position description. More on the FAQ page.

What it sets up

  • Incoming invoices captured, categorised, matched to the purchase order and entered as draft bills
  • Your own invoices raised from where the work is recorded, sent on schedule and followed up when overdue
  • Data entry between systems, supplier and customer follow-up, document routing, expense claim checks, recurring finance reporting
  • A spending cap it applies as a proposal is built, so a breach is caught on the day rather than at month end

What it fixes

Evenings spent on bills and claims, payments that wait on whoever has the login, and duplicates and mismatches that surface at reconciliation. The Accounts Payable Officer prepares the work and moves it between systems; you approve what goes out and what gets paid. Nothing is paid without a person.

Working example · Accounts Payable Officer

A budget the system enforces

Spending proposals roll up against rules that are on screen and checked live. The Accounts Payable Officer applies the cap as the proposal is built, so a breach blocks submission on the day and a person approves what passes.

Demonstration · sample data

Try it: include the van and watch the cap the Accounts Payable Officer applies.

A budget the system holds

Managers propose, the rollup checks every rule live, and a breach blocks submission

Demo

Proposals · Q3 capital and subscriptions

Rules the system enforces

  • Quarter capped at $120,000$67,300 rolled up
  • Software capped at $30,000 of the quarter$14,800 rolled up
  • Single items over $25,000 need director sign-off1 item held for sign-off

Rollup

$67,300 of $120,000

Within budget. Items over the sign-off limit go to a director before anything is committed.

Role 03 · Systems Analyst AI Employee

CRM and accounting

Sits across the systems your business already runs on, so a record entered once turns up everywhere it is needed. The symptom it clears: the CRM, the accounting file and the job sheets each hold a slightly different version of the truth, and someone spends hours each week retyping between them.

Works with. Xero, common CRMs, Google Workspace and Microsoft 365. The process audit maps what is in use, what must stay and what must connect, and that becomes the role's position description before it starts.

What it sets up

  • Contacts, invoices and jobs synced between platforms, with one agreed place where each record is entered
  • An action in one system triggered by an event in another, or an interface that sits over several tools at once
  • A watch on the operational signals across those systems, flagging what needs a decision from a person
  • The clean data the other roles work from: the Data Analyst's dashboards, the Accounts Payable Officer's matching, the Sales Coordinator's quotes

What it fixes

Duplicate data entry, records that disagree with each other, and reports that take hours because the numbers live in three tools. Every other role is weaker when information is trapped in separate places.

You usually do not need to replace anything. The Systems Analyst starts with the existing stack, finds the friction inside its remit, and proposes a new system only where the business needs one. Any change to a live record follows the rules in its position description, with a person approving anything consequential.

Working example · Systems Analyst

Found on the day, not at month end

A job running long, a delivery that never scanned, stock quietly hitting zero. The Systems Analyst watches the signals across the systems it keeps in step and flags what needs a decision from a person.

Demonstration · sample data

Try it: open an alert to see the signal the Systems Analyst acted on.

Found today, not at month end

Rules watch the operational signals and flag what needs a decision on the day

Demo
  • The signal behind the flag

    • MonOn hand 96 · Job 214 drew 30
    • TueOn hand 66 · Job 219 drew 48
    • Today 10:42On hand 0 · no purchase order open
    • Open demandJobs 221 and 224 need 55 this week

    The reorder never happened because usage doubled inside one week. The flag fires on the zero crossing, with the open demand attached, so purchasing knows the size of the hole and not just that one exists.

None of these are financial reports. A job clock, a carrier feed and a stock ledger, each with a rule watching it. The month-end report would have found all three in five weeks.

Role 04 · Data Analyst AI Employee

Reporting and dashboards

Assembles the recurring reports and the dashboards from the systems where the data already lives, and keeps them current. The problem it takes on is familiar: reporting that eats hours every week because someone has to pull numbers out of three tools and stitch them together in a spreadsheet.

Works with. Whatever holds the numbers: Xero, the CRM, the job system, the point of sale, Google Workspace and Microsoft 365.

What it sets up

  • A dashboard built from the systems that already hold the numbers, kept to the handful that drive decisions
  • Recurring reports generated and sent on schedule, with the live position on the dashboard in between
  • Any other cut of the numbers you ask for in Teams or Slack, with a person reviewing anything that goes to the board, the bank or the wider team
  • Financial, operational, workflow, turnaround, stock and team measures, whatever your systems record

What it fixes

The Monday morning spreadsheet. The Data Analyst reads the source systems directly, so the report assembles itself each week or month. Where a report sits in the efficiency scope of the role, it sets it up on its own. This usually depends on the systems being in step first, which is the Systems Analyst's remit above; once the data flows, the reporting is straightforward.

Working example · Data Analyst

Four stores, one Monday morning screen

An executive view the Data Analyst assembles for a retail group on sample data: sales by store, the SKUs doing the work, stock that is about to run out, and margin by category.

Demonstration · sample data

Try it: switch stores and check the low-stock flags the Data Analyst surfaced.

The week's trading on one screen

Four stores reporting into one executive view, updated from the store systems

Demo

Sales this week

$72,370

vs last week

+1.9%

Gross margin

41%

Stock lines low

2

Sales by store

Click a store to filter the screen to it.

Margin by category

  • Kitchenware46% on $19,310
  • Small appliances33% on $26,350
  • Home audio40% on $16,910
  • Seasonal & gifts52% on $9,800

Bar length is category sales; the figure is its margin.

Top SKUs

SKUItemUnitsSales
SA-118Benchtop blender 1.2 L77$9,860
KW-204Stone frypan 28 cm121$8,330
HA-310Bookshelf speaker pair27$5,110
SE-402Scented candle trio129$5,030
HA-322Soundbar compact15$3,300

Stock on hand

LineOn handReorder atStatus
Stone frypan 28 cm8655OK
Benchtop blender 1.2 L3536Low stock
Bookshelf speaker pair1214Low stock
Scented candle trio8965OK
Espresso grinder118OK
Chef's knife 20 cm1710OK
Soundbar compact1110OK

The owner's Monday morning without the spreadsheet: every figure here is assembled from the store systems overnight, and the low-stock flags are the reorder list.

Beyond the four

Any role builds inside its remit

The four above are the roles we place most often. The same mechanic holds for every role on the bench: a position description, work set up inside it, and a person at the gate. Here is a Procurement Officer doing the same thing with supplier feeds.

Working example · Procurement Officer

The cheapest listed price is not the cheapest

A Procurement Officer AI Employee runs one search across several supplier feeds, merges the same item where two feeds carry it, and compares on landed cost. It keeps the comparison current inside its own remit, and a person approves any order that follows.

Demonstration · sample data

Try it: switch searches and see what the Procurement Officer compares on.

One search, every feed

Three supplier feeds searched at once, duplicates merged, compared on landed cost

Demo
M10 stainless hex bolts, box of 100
FeedSKUListed per boxFreightLanded per box
Feed AHXB-10-100$92.50$0.00(Free over $75)$92.50
Feed BST-HX10C$84.00lowest listed$18.50(Order below feed minimum)$102.50Same item · FTX-M10
Feed CC-88412$86.20$9.00(Flat rate)$95.20Same item · FTX-M10

Listings tagged with the same manufacturer code are one item appearing in more than one feed. The comparison treats them as one product with two prices, not two products.

Feed B shows the lowest listed price at $84.00, but its freight under the order minimum takes it to $102.50 landed. Feed A is $8.50 dearer on the sticker and the cheapest to actually buy.

Thirty minutes, one conversation

Tell us the role you would otherwise hire for. We will tell you whether an AI Employee fits it, what the position description would say, and what it would cost, in writing, before anything starts. If the honest answer is a person or a process change, that is what you will hear.

Book a 30 minute conversation