What should we get done? Ask about invoices, revenue, leads, or tasks.
Do anything
Ask first
Agaro
Agaro Agent is AI and can make mistakes. Please double check its work.
Static illustration of Agaro Agent with sample data.
Preview / Pilot
Agaro ERP is in Preview / Pilot.
We are working with a small group of pilot customers first. Leave your
email and we will reach out when a pilot spot fits your business.
Nothing else, ever.
Join the pilot waitlist.
You're on the list!
We'll be in touch when a pilot spot fits. Talk soon.
One note about the pilot. No spam.
PercyOllie
How Agaro Agent works
Ask. Watch. Approve.
Ask
Say the outcome in plain words. The signed in person and their workspace decide what is allowed.
Watch it work
Agaro Agent finds the real records, prices and terms, and prepares the change. Totals, tax and numbering come from the ERP, never from a guess.
Approve
By default you review the change before it runs. Money, payroll, deletions and anything sent outside always ask, and the outcome is recorded under your name.
Your workspace. Your permissions.
Agaro Agent helps you find records, prepare invoices, update tasks, customize your ERP, and understand your business. It works inside your workspace with your module access and permissions, and it asks before it changes anything.
01 · Prompt → Identity → AccessThe user states the outcome; the signed in user, workspace, role, modules, and grants decide what is allowed.
02 · Resolve data → EngineReal records and values are read, then validated and calculated by the engine that owns them.
03 · Action → AuditThe same product server action runs, and the outcome is recorded under the user who asked.
Sample workflow
Agaro Agent
Ask first
You
Create a draft invoice for Acme Corp for the implementation package. The total is $12,000, due in 30 days.
Agaro Agent
Acme Corp’s billing terms are Net 30. The implementation package is 1 × $12,000.00. I’ll prepare the invoice as a draft and leave it unsent. Please approve the draft first.
Draft invoiceINV-2026-0418 Draft
Customer
Acme Corp
Implementation package
$12,000.00
Terms
Net 30
You
Approved. Create the draft only. Do not send it.
Agaro Agent
Created draft INV-2026-0418 for Acme Corp, totaling $12,000.00 with Net 30 terms. The action is recorded under Acme. No email has been sent.
RemyPeep
What it works in
The best way to run the business.
One source of truth
Customer, price, hours, contracts, approvals, and payment history are already recorded.
Same authority
Agaro Agent runs supported actions within the permissions of the person who asked and the modules installed.
Deterministic engines
Invoices, payroll, stock, and journals are calculated by the ERP, never improvised.
Everything in one workspace
One workspace connects every record Agaro Agent needs to act, so revenue, operations, and workforce read from the same data instead of being reconciled later.
By default Agaro Agent asks before every change. You can switch to automatic for everyday work, and it still asks about anything risky.
Some things always ask
Moving money, running payroll, deleting records, sending anything outside the company and changing who can do what wait for a yes in every mode.
Same permissions as you
Agaro Agent can never see or change more than the person who asked. Downgrading a plan never widens anyone's access.
Every action on the record
Each change is recorded in your workspace activity history under the person who asked, the same as a click.
Bizzy
Pricing
Pricing that fits a growing business.
Monthly, in US dollars. Employee accounts are free. Payroll covers 10 people, then it is $4 per person a month. Agaro ERP is in Preview / Pilot, so we confirm your plan with you before anything is billed.
The questions teams ask before they move their operations.
Agaro ERP is an AI native ERP built by Agaro Technologies LLC for finance, customers, people, operations, and builders. It keeps one connected source of business truth: contacts, leads, orders, invoices, payments, stock, employees, attendance, payroll, projects, and journals live in the same workspace rather than in separate tools that have to be reconciled. Traditional ERP records what happened and then waits for a person to act: find the records, resolve conflicts, calculate the result, push the buttons. Agaro closes that gap by adding an execution layer. Agaro Agent, the AI assistant inside the ERP, reads real records, calls the same business logic the product interface calls, and records the outcome under the user who asked. One permission model covers people and AI. Deterministic engines handle the parts where correctness matters, such as invoice totals, tax, numbering, and payroll. It is in its Preview / Pilot phase.
Yes. Agaro Agent is the AI assistant inside the ERP. It finds records, answers questions about your business, prepares supported work such as invoices and task updates, and can propose changes to how your ERP looks and works. It works with the same roles and access as the person who asked, so it can never see or change more than that person can, and it asks before it changes anything.
Whatever the person who asked can do, and no more. It is not a search box or a summarizer bolted onto a report screen. It creates and updates records, runs workflows, and completes transactions inside the ERP. Ask it to invoice a customer and it resolves the real customer, the catalog item, the stored price, the workspace currency, tax, and payment terms, then calls the finance engine to validate the request, calculate totals, allocate the invoice number, create the invoice, and commit the journal. It can create and assign tasks, move projects, prepare a payroll run, raise a purchase, or record a stock movement, each through the same server action a person triggers by clicking. When something is missing it asks. When an action needs confirmation, such as sending a document to a customer, it presents that action and waits for you.
Yes, through Studio. Describe the change in plain words and the assistant prepares it as one change set that you read as a diff, such as different summary cards, views and columns, grouping, field order, or which fields are required. You apply it or discard it, and nothing changes until you do. Customizing is limited to the people who already have Studio access, which means admins and managers, because the assistant never has more access than the person asking.
Yes, by default. Manually approve is the default mode, and the assistant asks before every change, so you stay in the loop on anything that changes. Automatically approve lets everyday changes run on their own while anything risky still asks. Skip all approvals is for admins, and even then moving money, running payroll, deleting records, sending anything outside the company, and changing who can do what or how people sign in still ask first. Whatever the mode, the outcome is recorded in your workspace activity history under the person who asked.
Yes. Every paid plan has full role-based access control (RBAC) and attribute-based access control (ABAC). That means 5 built in roles (Super Admin, Manager, Employee, Developer, and Viewer), custom access groups with inheritance, access levels for each module, separate read, write, create, and delete rights for each type of data, menu visibility, record rules that limit which records a person can see based on attributes like owner, team, or company, and field-level security that hides sensitive fields such as salaries. The same rules apply everywhere: in the app, through the API, over MCP, and to the AI assistant. The assistant acts with the exact permissions of the person asking, so it can never see or change more than that person can. Downgrading a plan never widens anyone's access.
No more than the person who asked. The signed in user is the identity and their workspace is the data boundary, so the assistant inherits the same role, the same module access, the same user grants, the same data permissions, and the same workspace. If the user cannot perform an action, the assistant cannot perform it. There is no service account with elevated rights sitting behind the chat window and no side channel into the database. Every request follows one controlled path: prompt, identity, access, resolve data, engine, action, audit. Access is evaluated before any record is read, engines validate and calculate before anything is written, and the outcome is recorded under the user who asked, not under the model. The model selects an action; the ERP decides whether and how that action executes. The audit history reads the same whether the work was done by clicking or by asking.
No. Payroll is calculated by the ERP, not improvised by the model. The payroll engine takes employment contracts, stored compensation, attendance and work entries, approved overtime, bonuses and adjustments, and benefits, deductions, and taxes as inputs. It produces the register deterministically: gross pay, minus taxes and deductions, plus approved reimbursements, equals net pay. Amounts are held in integer currency units so rounding cannot drift, the period is locked once it is run, and the same inputs always produce the same result. The outputs are a payroll register, a payslip snapshot, a payment file, a payroll journal entry, and an audit record. The assistant directs the workflow. It can gather what is missing, tell you which work entries are unapproved, and start the run, but it never estimates a number. The engine calculates and enforces correctness. The same rule applies to invoice totals, tax, numbering, and journals.
More than forty, grouped into seven app areas: Accounting, CRM, Productivity, Website, Human Resources, Marketing and Operations. Accounting covers quotes, invoices, payments, expenses, banking, loans, vendor bills, budgets, reports and taxes on one deterministic finance engine. CRM holds contacts, leads, pipeline, scheduling and events. Human Resources covers employees, attendance, time off, appraisals and the payroll register. Marketing covers campaigns, email and automation on the same contacts. Operations covers inventory, purchasing, warehouse and fulfillment. Productivity and Website cover tasks, projects, approvals, the module builder and the website builder, with the AI assistant working across all of them. Every module writes to the same workspace, so the assistant can find the real customer, price, and history before it acts.
Yes, with two builders that extend the product without breaking the core. The Module Builder is for custom operations: define your own fields and records, then deploy the module to your workspace. A module built this way inherits workspace isolation, permissions, record pages, and activity history, so a custom process is governed the same way a built-in one is. It is not a spreadsheet living outside the system. The Website Builder is for what customers see: landing pages, websites, storefronts, and portfolios, with forms connected to leads, products connected to the catalog, and checkout connected to Stripe. Neither builder creates a second source of truth. Custom operations stay inside the ERP and customer-facing experiences stay connected to the same records, so a form fill becomes a real lead and an order becomes a real order.
Through the same door as everything else. The product interface, the AI assistant, the API, and MCP are four entry points that reach one shared action layer. They do not each get their own path into the database. A request arriving over the API or over MCP is resolved to a signed in actor, a workspace, a role, a module set, and the data permissions that actor has, exactly as a button click is. From there it passes input validation, authorization, a transaction, and the activity log, and then into the finance, payroll, inventory, CRM, or workflow engine that owns the calculation. Data stays in workspace-isolated PostgreSQL with audit history attached. The practical effect is that an integration cannot do more than the account behind it, and a script cannot skip a validation a person could not skip. Same identity, same controls, same business actions, one audit trail.
It is in its Preview / Pilot phase. We are working with a small group of pilot customers rather than turning everything on at once, so the first teams get direct attention while the product matures. We are not publishing a general availability date. To join, tell us what is slowing you down in the form at the bottom of this page: what is not working in your current CRM, ERP or spreadsheets, and what you would want us to handle. We read every message and reply personally. You can also request a call with the people building it.
Putto work.
Tell us what is slowing you down. We read every message and reply personally.
