AI Agents and Assistants
What is an AI agent? A plain-English guide for business owners
What an AI agent is, how it works, where it helps a small business, the risks, and a 7-point check to see if you are ready for one.
Business Agents
AI Executive Assistant
Inbox triaged, calendar managed, briefings prepared and follow-ups chased, so nothing slips through.
If your inbox decides how your day goes, important messages get buried, meetings arrive without preparation and follow-ups quietly slip. This assistant sorts your inbox by what matters, manages your calendar within your rules, prepares a short briefing before each meeting and chases the follow-ups you would otherwise chase yourself. It only acts where you allow, and logs everything.
- Inbox sorted so the important messages reach you first
- A short briefing ready before every meeting
- Follow-ups chased on time, with every action logged
$4,400
20–30 working days
Optional maintenance $249/mo
What it is
AI Executive Assistant is a delegate that triages your inbox, manages your calendar, prepares meeting briefings and chases the follow-ups you would otherwise chase yourself. It reads only the systems you connect, acts inside the rules you set, and logs everything it does. It remembers prior context within your workspace.
Why businesses buy it
Leaders lose hours each week to inbox triage, calendar juggling and reminding people about things already agreed. A human assistant helps but is a big commitment, and generic tools need your time to configure. This assistant is set up around your own rules, so routine coordination happens without you and nothing important slips. It can only use the access you approve, and every action is logged. The Scope Card lists systems, workflows and limits before you order.
Problems this solves
Important emails get buried under newsletters, notices and routine requests
The assistant triages your inbox inside the rules you set, sorting what needs you from what can wait. It works only in the systems you connect, and every action it takes is logged.
Walking into meetings without the background you needed
Before each meeting it prepares a briefing from your connected systems and documents. It remembers earlier context within your workspace, so the briefing reflects what was already agreed.
Follow-ups you meant to send slip for days or weeks
It chases the follow-ups you would otherwise chase yourself, using up to six workflows with your own approval rules. You decide which messages it may send and which wait for your yes.
Calendar juggling eats time that should go to real work
The assistant manages your calendar through the connected calendar system, handling routine scheduling within the limits you approve. Anything outside those limits comes back to you to decide.
A good fit if
You spend real time each day sorting an inbox that mixes urgent client emails with newsletters and internal noise.
Your calendar needs active management: moving meetings, finding slots across time zones, and declining things that should not be there.
You want a one-page briefing before a meeting, not a scramble through old threads to remember what was last agreed.
You have follow-ups that regularly get dropped because nobody owns chasing them until someone finally does.
Not the right fit if
You mainly need task and deadline tracking across a team project, not one person's inbox. See the AI Project Manager Agent, built for that.
You want notes and action items captured from meetings themselves, not the calendar around them. See the AI Meeting Agent.
You are one person managing your own day rather than delegating for someone else. The AI Personal Productivity Agent is a lighter build for individual use.
You need more than six workflows or more than two connected systems. Add a note at checkout for a written Change Quote, or use the Custom AI request form for larger builds.
You only need email replies drafted for approval, not inbox triage or calendar work. AI Email Drafting Agent is the entry-level build.
How it works
- Your inbox is triaged as it arrives. Messages are sorted into what needs you now, what can wait, and what the assistant can answer or file on its own, using the rules you set during intake.
- Your calendar is kept honest. New requests are checked against your real availability and preferences, meetings are moved or declined inside your rules, and time zone conflicts are caught before they become a problem.
- Briefings are prepared ahead of meetings. Before a call, the assistant pulls the last relevant thread, notes and open items from your connected systems into a short brief, so you walk in knowing where things stand.
- Follow-ups are chased. When a task or reply is owed to you or by you, the assistant tracks it and sends a polite nudge on the schedule you approve, instead of letting it quietly go stale.
- Anything outside its approval limits is flagged to you rather than sent. It drafts; it never sends on your behalf unless you have explicitly allowed that exact action.
- Every action is logged. You get a running record of what was triaged, moved, drafted or chased, so nothing happens invisibly.
What you provide within 72 hours
Your portal lists these the moment you order. Delivery starts the working day after everything is in and your access is checked.
- Access to the inbox, calendar and any other systems the assistant will work in, set up as sandbox or test accounts where you have them.
- Your ground rules: what the assistant can act on directly, what needs your sign-off first, and any limits on tone or timing.
- Ten real emails or threads that show the kind of decisions and replies you actually deal with.
- The name of whoever approves actions that need a person, so nothing sits waiting on the wrong desk.
- Admin access and API keys for every connected system, shared by adding our team as a user or creating a key you control, never sent by email.
- Your business information (company basics, hours and brand voice), shared during intake, so the assistant understands your context.
- One decision-maker on your side who can sign off changes during the build.
- An API key for your own AI model account (for example OpenAI or Google Gemini), created for this build with a spending limit.
The 72-hour rule. Send everything on your checklist, including access, within 72 hours of ordering so your build starts on time.
Timeline
Start
Intake verified, approval rules agreed.
Weeks 1–2
Inbox and calendar workflows built.
Week 3
Connected systems linked and permissions set.
Weeks 3–4
50 test scenarios run, results sent to you.
Week 4
Your review and two rounds of changes.
Go-live
Live, recorded handoff session, support starts.
| Option | Delivery time | Price |
|---|---|---|
| Standard | 20–30 working days | $4,400 |
| Express | 15–25 working days | $5,150 |
| Rush | 10–20 working days | $5,950 |
What you receive at handover
- The AI Executive Assistant, configured and running in your own environment
- Your workflow, prompt and guardrail settings, including the approval rules for each workflow
- Working connections to up to two agreed systems, such as your email and calendar
- A test report covering the agreed test scenarios
- An admin guide and a recording of the handoff session
- 30 days of post-launch support
Expected benefits and how to measure them
- Can give you back time now spent sorting email and arranging meetings
- Helps you arrive at meetings prepared, with the context already gathered
- Makes it easier to see that promised follow-ups actually went out
- Time spent on email each week
- Log your email time for two weeks before launch. Repeat the same log a month after launch and compare.
- Follow-ups sent on time
- Compare your sent folder for a month before launch with the assistant's action log for a month after.
- Meetings with a briefing ready
- Count calendar meetings that had a briefing prepared beforehand, for a month before launch and a month after.
Before you start: Note how many hours a week go on email and scheduling, and how many follow-ups slipped last month. Keep a copy of a typical week's calendar.
These are things to track, not promised results. Outcomes depend on your data, your team and your market.
Running costs
The build price is one-time. Running the assistant is billed in your own accounts, mostly AI model usage plus light hosting, and typically lands around $50 to $250 a month depending on inbox volume and how many workflows are active day to day. You keep your existing Google Workspace, Microsoft 365 or CRM plan on top of that.
AI usage during the build and testing is billed the same way, in your own accounts, so every provider relationship stays yours.
Ask about running costsCustom requirements
Add a note at checkout if you need something specific. If it fits within the Scope Card it is simply included. If it goes beyond it, you get a written Change Quote within two working days, and that work starts only once you accept it, with its time added to your delivery date.
Request a custom versionKeep it running
Monthly maintenance covers monitoring, fixes and change hours for new workflows, tweaked approval rules or a new connected tool. Bigger changes go through a Change Quote. Maintenance starts once the included support period ends.
$249/mo (the minimum fee, includes 2 change hours)
Optional. Starts after the post-launch support period. Cancel any month.
How maintenance worksHow this compares
| Do it yourself with a SaaS tool | Hire a custom agency | Upmark Labs | |
|---|---|---|---|
| Price | A scheduling or inbox tool charges a monthly seat fee whether or not it is configured for you | A virtual-assistant agency charges a monthly retainer | Ours is one fixed price, shown before you buy |
| Scope | A SaaS tool ships generic automations you configure yourself | An agency's scope is agreed informally and can drift | Ours is written on the Scope Card before purchase |
| Delivery | A SaaS tool is live once you finish setting it up, which can take weeks of your own time | A VA agency usually takes one to several weeks to ramp a new hire | Ours has a fixed delivery time shown before you buy, with faster options |
| Approval rules | A SaaS tool applies the same automation to everyone | An agency's judgment varies by who is assigned | Ours acts only inside the approval rules you set at intake |
| Records | A SaaS tool logs what it automated, not why | An agency's notes depend on the person | Ours logs every triage, move, draft and chase in one place |
Questions about this product
Can it send emails on my behalf?
Only where you explicitly allow it, for specific low-risk replies you approve in advance during intake. Everywhere else, it drafts a reply and leaves it sitting for you to review and send yourself. You set that boundary once, and it never quietly widens beyond what you approved.
What happens with requests outside its approval rules?
It flags them to you instead of guessing at an answer. Anything not covered by the rules you set, such as an unusual meeting request or a reply that would commit you to something, gets queued for your decision rather than acted on automatically, with the reasoning shown alongside it.
What will it cost me each month to run?
Roughly $50 to $250 a month in your own accounts, covering AI model usage and light hosting, on top of the Google Workspace, Microsoft 365 or CRM plan you already pay for. Cost tracks inbox volume and how many workflows stay active day to day. There is no separate fee to us.
How much of my team's own time does onboarding take?
About one to two hours for intake, mostly gathering access and writing down your approval rules, spread across the 72 hours after you order. After that, expect roughly an hour to review the 50 test scenarios, plus the recorded 60-minute handoff session once it goes live.
Can it work across time zones?
Yes. Calendar handling accounts for the time zones on both your calendar and your contacts' calendars, and every meeting suggestion is checked against both before anything is proposed, moved or declined, so nothing gets scheduled at an hour that does not actually work.
Can I add more connected systems later?
Small adjustments to an existing connection fit inside the monthly change hours included with maintenance, such as a new folder rule or an extra calendar. A brand new system beyond the two in scope comes through a written Change Quote, priced and agreed before the work starts.
What if the assistant gets something wrong at launch?
You get two rounds of changes during the build, after reviewing the 50 test scenarios together, so most issues surface and get fixed before launch. After delivery you also have five working days to flag anything against the Scope Card, corrected at no extra cost to you.
Who owns the data and accounts it works in?
You do, entirely. Every connected account, API key and piece of context stays in systems you control, shared by adding our team as a user or creating a key you control rather than by email. We configure the assistant inside your environment, and nothing about it routes through accounts that belong to us.
Guides on this topic
All guides Stop losing hours to an inbox that runs your day.
Order the AI Executive Assistant and it starts triaging your inbox and managing your calendar on a fixed delivery date, always inside the rules you set.
Works well with
Business Agents
AI Project Manager Agent for Linear
Tracks tasks, deadlines and blockers and posts status updates to the team, connected natively to Linear so it reads and updates issues,…
Business Agents
AI Project Manager Agent for Monday.com
Tracks tasks, deadlines and blockers and posts status updates to the team, connected natively to Monday.com so it reads and updates boards,…
Business Agents
AI Project Manager Agent for Asana
Tracks tasks, deadlines and blockers and posts status updates to the team, connected natively to Asana so it reads and updates tasks,…