AI agents · Cost

How much does an AI agent cost? Setup, monthly care and model usage

What an AI agent costs comes down to three parts: a one-time setup for each workflow, a fixed monthly care fee that keeps it working, and model usage you pay directly to the provider. You get the exact figure after a Discovery.

Discovery is a fixed fee, credited to your first months of monthly care

You pay
A one-time setup and monthly care per workflow, plus model usage direct to the provider
It depends on
Mainly the number of systems, the state of your data and the approval flow
Your figure
Comes from a Discovery, whose fee is credited toward your first months of monthly care

Cost structure

What makes up the cost of an AI agent?

An AI agent is priced per workflow in three parts: a one-time setup, monthly care, and language-model usage that you pay the provider directly. Before all three comes the Discovery, which produces the estimate.

First: a fixed-fee Discovery
The step that produces the estimate: process map, baseline, agent design and a fixed quote. If you continue, the fee is credited toward your first months of monthly care. It's a first step, not a fourth cost layer.
Setup: one-time, per workflow
Baseline, building the agent, connecting systems, the test set and shadow mode, until the agent meets the agreed criteria and goes live.
Fixed monthly care
Hosting and running, monitoring, prompt and model updates, small changes and a monthly report, so the agent keeps working as models and systems around it change.
Model usage: paid by you directly
The model provider bills your own account by volume, and you see every charge. With a local model on your own servers, the cost is hardware instead of usage.

Pricing

How it's priced: per workflow, not by the hour

Billing by the hour is unpredictable for you and penalises efficient work. A fixed price per workflow gives you a known figure for a defined result, and every path starts with a Discovery.

Start here

Discovery

A Discovery of one workflow: we map it, measure it, design the agent and estimate the return on investment.

  • Process map and baseline
  • Agent flow design and approval points
  • Return-on-investment estimate
  • Fixed quote for setup and monthly care

Credited toward your first months of monthly care

Start with a Discovery

Fixed fee, agreed up front in a written proposal

  • 01

    Starter

    1 workflow

    A business handing one daily workflow to an agent, such as quotes, invoices or follow-ups

    • Email and spreadsheets, such as Excel or Google Sheets
    • An approval step before every key action
    • An action log, measured against the baseline
  • 02

    Growth

    2–3 workflows + ERP

    A business with several related workflows that needs the agent to work with its ERP and a knowledge base

    • An ERP connection, such as Priority or SAP Business One, via its API
    • A knowledge base built from your documents and past jobs
    • Approval steps and an action log for every workflow
  • 03

    Industrial

    Scoped per site

    A plant with several lines that wants agents between the floor, quality, maintenance and the ERP, in phases

    • A custom LangGraph agent, or one running on your own servers
    • Read-only access to machine data, never writing to PLCs or setpoints
    • Hosting on servers in the plant or in a private cloud
    • Phased work, each phase paying back before the next

Monthly care covers

  • Hosting and running the agent
  • Monitoring and the action log
  • Prompt and model updates
  • Small changes to the workflow
  • A monthly report on the agent's work

How billing works

  • Per workflow: a one-time setup plus a fixed monthly care fee
  • A fixed price per workflow, agreed after Discovery and before any build work
  • On every tier, model usage is billed by the provider to your own account, so you see every charge

Cost drivers

What drives the cost of an AI agent?

Eight factors account for most of the difference between one workflow and another. Some of them are your choice, and that's how you bring the cost down.

How many systems, and whether they have an API
Every extra system adds integration and testing work. A system with a clean API is simpler to connect than one you can only reach through exports.
Data quality and past examples
Well-kept past examples bring the cost down. A messy history means clean-up work in Discovery and during setup.
Approval points and roles
Every approval point and every extra role is another path to build and test. A simple approval flow brings the cost down.
Document types
Plain text and emails are simpler. Scanned drawings, handwriting or voice recordings add work and testing.
Cloud or local model
A cloud model is quicker to set up. A local model on your own servers adds hardware and installation, and keeps the data with you.
Volume
High volume mainly affects the model-usage cost, and much less the setup.
Accuracy needed before go-live
The stricter the go-live criteria, the larger the test set and the longer the shadow mode.
Security requirements
Secure sign-in, a full audit log or a data-processing agreement add work, and are sometimes a must.

Model usage

How much does the language model itself cost?

It depends on how many tasks there are each month and how large the documents in each task are. The volume is estimated in Discovery, before you start.

Each task the agent runs sends text and documents to the model and gets an answer back, and the provider bills by volume. So the monthly cost depends on two things: how many tasks there are each month, and how large each document is.

Simple tasks, like sorting an email or extracting lines from an invoice, can run on small, inexpensive models. Complex tasks, like reading a drawing, get a stronger model. The model is chosen per task for the best price-performance.

The charges go straight to your own account with the provider, with no mark-up from us, and you see every charge. Discovery estimates the expected usage, so there are no surprises. With a local model on your own servers there's no usage billing; the cost is hardware instead.

Choosing a model by cost per task

What you get back

What do you get back for the cost?

More capacity and more speed, not fewer people: quotes out the same day, fewer forgotten follow-ups and less paperwork for quality and maintenance. Two illustrative scenarios:

  • Illustrative example

    Small CNC machine shop

    Quotes from drawings, approved by the owner

    Every request for a quote means reading a drawing, finding similar jobs and calculating, usually in the evening. A slow reply loses the job.

    1. RFQ arrives by email
    2. Drawing read
    3. Similar past jobs
    4. Draft quote
    5. Owner approves(human approval)
    6. Sent, follow-up set

    The aim: Quotes go out the same day instead of after several days, and no price is sent without approval

    Typical plan: Growth

  • Illustrative example

    Plant with several production lines

    From a floor report to an NCR draft

    Non-conformance reports get written late or incomplete, from a note or a phone photo. Quality staff spend their time on paperwork instead of the problem.

    1. Text or photo report
    2. Structured NCR
    3. SOPs and past CAPAs
    4. Draft root cause
    5. QA signs off(human approval)

    The aim: Less paperwork for quality and maintenance, and NCRs opened on time and complete

    Typical plan: Industrial

Illustrative scenarios based on typical processes, not client projects.

The machine-shop scenario in full

Comparison

An AI agent vs a ready-made bot, a SaaS tool or a new hire

A qualitative comparison, with no prices: what you pay for, who maintains it, and when each option fits.

  • A ready-made bot: you pay a subscription, and the vendor maintains it. It answers common questions but doesn't work with your systems and documents. A fit for simple customer service.
  • A ready-made SaaS tool: you pay a subscription per user or by volume, and the vendor maintains it. It works from a generic model, and your data sits with the vendor. A fit when your workflow looks like everyone else's.
  • A new hire: you pay a salary and invest in recruiting and training. A fit when the work needs broad judgement, personal relationships with customers or physical presence.
  • A custom AI agent: you pay setup and monthly care per workflow, and the model usage directly to the provider. Conceptech maintains it. It works on your own data and systems, and a person approves the key steps. A fit for a workflow that repeats, is specific to your business and takes real time.

An agent doesn't replace your team. It gives them time back, so quotes go out faster and fewer things get missed. In a plant the work is phased, and each phase should pay back before the next. Phased AI implementation in a plant

Getting an estimate

How do you get a cost estimate for your workflow?

Three steps, and at the end a written figure for your workflow.

  • Write to us about one workflow: what it is, how often it repeats and which systems it touches.
  • A short call with an engineer, to understand the workflow and the systems.
  • A written proposal for a Discovery, at a fixed fee.

At the end of the Discovery you get an ROI estimate and a fixed quote for setup and monthly care. If you continue, the Discovery fee is credited toward your first months of monthly care. One cost driver worth checking early is the ERP connection: connecting an AI agent to Priority and your ERP

Discovery, credited toward monthly care

Discovery

What do you get from a Discovery?

A fixed-fee, fixed-scope first step: one process, one report you can act on, and the fee is credited toward your first months of monthly care if you continue.

What you bring

  • One process owner who knows the work and can go through examples with us
  • 20–100 real past examples, such as emails, quotes or invoices, sanitised if needed
  • Read access to, or exports from, the documents and systems the process uses
Fee
Fixed and agreed up front, in a written proposal
Credit
Credited toward your first months of care, if you go ahead with setup
Data
NDA on request; examples can be sanitised
Monthly care
Cancel any time with one month's notice, no minimum period

Discovery is a fixed fee, credited to your first months of monthly care

Discovery report
  1. 01Process mapHow the process runs today, step by step: who does what, in which systems
  2. 02BaselineTime per task, weekly volume, errors and response time to customers
  3. 03Agent flow designSteps, approval points, systems, model, and whether it runs in the cloud or locally
  4. 04ROI estimateHow much time the agent is expected to save, against the cost of setup and care
  5. 05Draft go-live criteriaWhat must hold in shadow mode before the agent starts acting
  6. 06Fixed quote for setup and careA fixed price for the workflow, set before any build work starts

AI agent cost: FAQ

How much does an AI agent cost a business?

The cost has three parts: a one-time setup per workflow, a fixed monthly care fee, and model usage you pay directly to the provider. You get the exact figure for your workflow in writing after a Discovery, and the Discovery fee is credited toward your first months of monthly care.

Why is there no price list on the site?

Because two businesses with "the same" workflow differ in systems, data and approval flow, and those differences set the cost. Instead of a generic price list, every workflow gets a fixed price in writing after the Discovery, before any build work.

What does monthly care include?

Hosting and running the agent, ongoing monitoring and the action log, prompt and model updates, small changes to the workflow, and a monthly report: tasks done, time saved and errors caught at approval. Larger changes are quoted up front, before any work on them.

Why do we pay the model provider directly?

For transparency: you see every charge, there's no mark-up from us, and the data agreement is between you and the model provider. The account settings, including usage caps, stay under your control.

Can we start with one workflow and add more later?

Yes, and that's the path we recommend. Each workflow is priced separately, and each phase should pay back before the next one starts. The decision on the next workflow is made from the monthly report, and if you decide it isn't for you, you can cancel the monthly plan any time with one month's notice, with no minimum period.

How does the Discovery credit work?

If you go ahead with setup, the Discovery fee is credited toward your first months of monthly care, so those first months are covered by what you've already paid. The Discovery fee itself is agreed up front, in a written proposal.

Which is cheaper: a custom agent or an off-the-shelf tool?

For generic tasks, usually the off-the-shelf tool. A custom agent pays off when it works on your own data, documents and systems, in a workflow that repeats every week. The Discovery tells you which fits, even if the answer is a ready-made tool.

Are there hidden costs?

No. There are only three lines: setup, monthly care and model usage. Hardware for a local model, if needed, is quoted separately and up front. Changes bigger than a "small change" are quoted before any work on them.

Want a real figure? It starts with a Discovery of one workflow

Tell us about one workflow. An engineer will reply with a written Discovery proposal, which ends with a fixed price for setup and monthly care.

Usually a reply within 3 working hours, by email or phone

Talk to an engineerWhatsApp(opens in a new tab)