Agents run on your Monton data when a record changes, on a schedule or when you ask. They read, use AI to write and decide, then create tasks, notify people and send emails. Every email waits for your approval.
Replaces Manual follow-ups
Free up to 5 people · No card- TriggerOn a scheduleFridays at 16:00
- ReadRead dataA project's tasks
- Model · AIWrite textProgress report for the client
- ControlApproval gateThe agent's owner approves
- ActionSend emailTo the client contact
Three ways to build an agent.
Start from a template, describe it in plain words or draw it step by step. New agents are created switched off: nothing runs until you enable it.
From a template
18 ready-made agents for projects, sales, team, billing and reports. Choose who gets notified and switch it on.
Build with chat
Describe what the agent should do and the assistant builds the workflow on the canvas. Nothing is saved until you press Save.
In the visual editor
Drag steps onto the canvas: triggers, data reads, AI steps, branches, approvals and actions.
Nothing gets emailed without your OK.
Emails always wait for approval. You see exactly what will be sent and to whom, fix the wording if you need to, and approve or reject, one by one or as a batch. If nobody decides before the deadline, the run is cancelled, so nothing goes out just because nobody looked.
- ✓Approve, reject, or edit and approve
- ✓Approve a whole batch of emails at once
- ✓What runs is exactly what you approved
- ✓A pending-approvals badge, so nothing gets stuck
- To
- accounts@waynetech.com
- Subject
- Invoices WT-0042 and WT-0045 are overdue
Hi Lucius, a quick reminder that two invoices totalling €21,300 are now more than 7 days overdue…
Start with 18 templates.
Each one is a working agent. Switch it on as it is, or open it in the editor and change it.
New project kick-off
When a project is created, adds the kick-off phase with its first tasks and notifies the person you choose.
Kick-off when a deal is won
AIWhen a work order is approved, creates its project with a kick-off phase and task, summarises the brief with AI and notifies the person you choose.
Weekly client report
AIApprovalEvery Friday reads a project's tasks, writes the progress report with AI and, once you approve it, emails it to the client.
Budget watcher
Every weekday morning finds projects that have used 80% or more of their budgeted hours and tells each owner which of theirs.
Stuck tasks
Every weekday morning finds the open tasks past their end date and sends each assignee one notification listing theirs.
Work order assigned alert
When a work order is created with someone responsible, or is reassigned, tells that person straight away.
New project alert
Notifies one person every time a project is created.
Qualify inbound prospects
AIWhen a prospect comes in, AI decides whether it is a priority and notifies the person you choose for each kind.
Cold prospect follow-up
AIApprovalEvery Monday finds the prospects idle for 7 to 13 days, drafts a follow-up for each with AI and sends them once you approve the batch.
New client alert
Notifies the accounts team every time a client is added.
Timesheet reminder
On Friday afternoon finds who has logged less than their weekly hours and reminds each of them how many are missing.
Welcome a new team member
When someone activates their account, creates their onboarding tasks and welcomes them.
Absence request alert
Tells the person you choose whenever someone requests leave or an absence, with who and which days.
Overdue payment reminders
ApprovalEvery Monday drafts one reminder per contact listing all their invoices at least 7 days overdue, and sends them once you approve the batch.
Billing due today
Every morning tells each project owner which planned billing items are due that day, so the invoice goes out on time.
Hour bank running low
AIApprovalWhen a client's hour bank crosses its alert threshold, AI drafts a renewal email to the client, sent once you approve it.
Monthly profitability for management
AIApprovalOn the 1st reads each project's profitability, highlights what is off track with AI and emails it to management after your approval.
Projects digest on demand
AIApprovalWhen you run it, reads the projects, summarises them with AI and emails the summary after your approval.
Included in every paid plan.
Agents run for organizations on a paid plan. AI usage is part of the plan price: no credits, no add-ons, no per-run fees.
See plans →Runs when something changes, on a schedule, or when you ask.
Start an agent when a project, work order, client, person, absence, allocation or prospect is created or updated, with conditions like "changed to" or "is after". Or run it on a schedule, or by hand with Run now.
- ✓When a record is created or updated
- ✓On a schedule, with a preview of the next runs
- ✓By hand, with Run now
- ✓Conditions: changed to, is after, is empty…
trigger schedule · weekdays 08:00
read budget usage ≥ 80%
notify each project owner
"3 of your projects are over 80%"AI that drafts. Steps you can see.
The AI never writes to your data
An agent is a workflow with clear steps. AI writes text, extracts data or picks a path; creating records and sending emails are separate steps you can see, limit and approve.
The AI can't pick the recipients
The model drafts the email, but recipients only come from your data. Agents read with their owner's permissions, and guests can't see or approve agents.
Every run on record
A read-only run history shows who or what started each run, every step's input and output, how long it took and what it cost.
Your first agent in three steps.
- 1
Pick a template
Open Agents, press New agent and choose a template, or start from a blank canvas or the chat.
- 2
Choose the people
Decide who gets notified, who approves and when it runs.
- 3
Switch it on
Enable it and follow each run in the run history. Disable it at any time.
Why agents, and not another automation tool?
Generic automation tools see your agency through webhooks and spreadsheets. Monton agents work where the data already is: budgets, logged hours, profitability, billing plans, hour banks and prospects. A budget watcher doesn't need a chain of connectors; it reads budget usage and tells the right people.
Agents also follow your organization's roles. Admins, directors and managers build them, agents read with their owner's permissions, and every run keeps a full history. That's what lets you hand them real work, like client reports and payment reminders, without losing control of what goes out under your name.
Questions about agents
- What are Monton agents?
- Automations that run on your behalf inside Monton. Each agent has a trigger (a record change, a schedule or a manual run) and a set of steps: read data, use AI to write, extract or decide, create projects, phases and tasks, notify people and send emails.
- Which plans include agents?
- Every paid plan. Agents don't run on the free Starter plan, because every run uses AI. There are no AI credits or per-run fees on top of the plan price.
- Which AI model do agents use?
- Agents use Anthropic's Claude models. In AI steps you can choose between the default model and a more capable one for harder tasks.
- Can an agent email my clients without me knowing?
- No. Email steps always require approval, and that rule is enforced while the agent runs, not only when you save it. If nobody decides before the deadline, the run is cancelled.
- Who can create and approve agents?
- Admins, directors and managers can create, edit and enable agents. Approvals go to the agent's owner, a specific person or whoever started the run. Guests can't see agents or approve anything.
- Can agents change or delete my data?
- Agents can create projects, phases and tasks, notify people and send emails. They can't update or delete existing records, and they can't call external services.
Features that work with this one.
What if you just asked Claude?
Connect Monton to Claude, Cursor or any MCP-compatible assistant.
What about the hours from all those meetings?
Turn your calendar events into logged hours in one click.
Log hours without leaving your editor?
Browse projects and tasks and run the timer from VS Code or Cursor.
One flat price. AI agents included. The whole team.
No per-seat fees. Every core feature on every plan, and AI agents on every paid one. When your team grows, your bill doesn't explode.
View plansPrice per plan, not per user · billed annually