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.
Oversight
AI Escalation System
AI cases are escalated to a person only when a rule triggers, such as low confidence or high risk.
Reviewing every AI decision by hand defeats the point, but letting risky ones through is worse. This system lets routine cases run on their own and sends only the ones that trip a rule to a person: low confidence, a high-value decision or a flagged keyword. The right reviewer gets the case with its context, and you set and change the triggers.
- Only risky or uncertain AI cases reach a person
- Each escalation routed to the right reviewer with full context
- Every trigger and outcome logged for tuning the rules later
$2,400
20–30 working days
Optional maintenance $249/mo
What it is
AI Escalation System routes specific cases to a person based on rules you set, such as low model confidence, a high-value decision, or a flagged risk keyword, rather than stopping every action for review. It is reactive and conditional, escalating only what genuinely needs a human.
Why businesses buy it
A blanket review step slows every case down, while no review at all leaves you exposed on the few that matter: the large refund, the angry customer, the answer the model was unsure about. Escalation by rule fixes both problems. Your team spends its attention on the cases that need judgment, and the rest move at AI speed. The written scope lists the workflow, the connected systems and the trigger rules before you order, so you know exactly what will escalate and where it will go.
Problems this solves
Stopping every AI action for review slows the whole process down
The system lets routine cases run and sends only those matching your rules to a person. Triggers can include low model confidence, a high-value decision or a flagged risk keyword.
Risky cases slip through because escalation rules were never written down
Your escalation rules are built into the workflow's AI decision step and tested on twenty runs of your sample data. Each case that meets a rule goes straight to the person you choose.
Escalated cases arrive without context, so staff start from zero
Each escalation carries details from up to three connected systems, along with the rule that triggered it. The person picking it up can decide quickly without digging for background.
A good fit if
Most AI-handled cases are routine and only a minority genuinely need a person, so reviewing every single one would waste your team's time.
You can define the triggers that matter, such as low confidence, a dollar threshold, or specific flagged language.
You want AI to keep working independently on the cases that do not trigger a rule, without a standing bottleneck.
You need different triggers to route to different people or teams, not one queue for every kind of escalation.
Not the right fit if
You want every AI-proposed action reviewed by a person, not just the ones that trigger a rule. See Human Approval Workflow, our standing-gate build.
You want an AI that actively supervises another agent's ongoing work and can intervene on its own, not just route cases to a person. See AI Supervisor Agent.
You need multiple agents coordinated in one workflow rather than an escalation layer on top of one process. See AI Agent Orchestration System.
Your workflow needs more than 12 steps or more than one AI decision point. Use our Custom AI enquiry form instead.
How it works
- We map your process and agree the specific triggers worth escalating, such as low confidence, a dollar or risk threshold, or particular flagged language.
- The AI continues to handle routine cases on its own, without stopping for review, as long as no trigger fires during that case.
- When a trigger fires, the case, and the reasoning behind why it triggered, routes to the right person or team, not a generic shared inbox.
- The escalation connects to the tools your team already uses, such as Slack, Microsoft Teams, email, your helpdesk or your CRM.
- Every escalation, and the outcome once a person handles it, is logged, building a history you can use to refine the trigger rules later.
- We run 20 test cases on your sample data before handoff, then deliver a recorded session and an admin guide covering how to adjust the triggers.
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.
- Your business information (company basics, hours and brand voice), shared during intake, or confirmation we can reuse what you gave us on an earlier order.
- The systems involved in the workflow, with admin or API access for each.
- A written description or a screen recording of the current manual process.
- Five to ten real sample inputs, covering both routine and edge cases.
- The specific triggers that should cause an escalation, and who receives each one.
- Who should receive alerts and errors if the escalation routing itself fails.
- Admin access and API keys for every connected system, shared by adding our team as a user or creating a key you control, never by email.
- One decision-maker who can approve the final trigger rules and routing.
- 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
business information confirmed and the escalation triggers agreed.
Weeks 1–2
Trigger rules and escalation routing built and connected to your named systems.
Week 3
Routing to the right person or team per trigger configured.
Weeks 3–4
20 test runs completed on your sample data, covering routine and trigger cases.
Week 4
Your review and two rounds of changes to the trigger rules or routing.
Go-live
Live, recorded handoff session, and the 30 days of support begins.
| Option | Delivery time | Price |
|---|---|---|
| Standard | 20–30 working days | $2,400 |
| Express | 15–25 working days | $2,900 |
| Rush | 10–20 working days | $3,350 |
What you receive at handover
- The escalation workflow running in the background, up to twelve steps across up to three systems
- Your escalation rules, prompts and guardrails configured in the AI decision step
- A test report from twenty runs on your sample data, showing which cases escalated
- An admin guide and the recorded handoff session
- 30 days of support after launch
Expected benefits and how to measure them
- Routine work can keep moving while people focus on cases that need judgment.
- Helps make escalation decisions consistent instead of depending on who is on shift.
- Makes it easier to adjust thresholds as you learn what truly needs a human.
- Escalation volume
- Count escalated cases each week from the workflow's records and set them against total cases handled.
- Missed risk cases
- Review a monthly sample of cases that were not escalated and count any that should have reached a person.
- Time to resolve escalations
- Compare escalation-to-decision times after launch with how long similar manual reviews took before.
Before you start: Count how many cases a person reviews today and how long each takes. Note any recent cases that should have been escalated but were not.
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 escalation system afterwards costs roughly $20 to $100 a month in your own accounts, covering the automation platform plus AI model usage for evaluating each case against the trigger rules.
AI model usage during build and testing is also billed to your own accounts, so ownership of every provider relationship stays with your team.
Ask about running costsCustom requirements
If your workflow needs more than 12 steps, more than one AI decision point, or more than three connected systems, add a note at checkout. You get a written Change Quote, and your original order keeps its normal timeline. Larger or bespoke projects use the Custom AI request form instead.
Request a custom versionKeep it running
Monthly maintenance covers change hours for adjusting a trigger threshold or the routing as your process evolves. It starts after the 30 days of support and can be cancelled any month.
$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 case-routing tool charges a per-seat monthly fee | A custom escalation build is quoted after discovery | Ours is one fixed price, shown on this page before you buy |
| Scope | A routing tool leaves the trigger logic for you to configure | An agency agrees scope during the project | Ours is written to your exact triggers before you order |
| Delivery | A routing tool still needs real setup before triggers are trustworthy | An agency build usually takes several weeks, set per project | Ours has a fixed delivery time shown before you buy, with faster options |
| Coverage | A generic tool routes cases but does not evaluate them for a trigger | An agency deliverable depends on what was agreed | Ours evaluates each case and only escalates when a rule fires |
Questions about this product
How much does AI Escalation System cost?
The build is one fixed price, shown with each delivery speed in the price and delivery panel on this page. It covers everything on the Scope Card: one workflow with up to twelve steps, the trigger rules that decide what escalates, routing to the right reviewer, and up to three connected systems. Running costs, mainly AI usage to check each case, are paid in your own accounts.
When exactly does a human get pulled into AI Escalation System?
Only when a rule you defined fires, such as low model confidence, a dollar threshold, or specific flagged language, so most routine cases are handled by the AI without a person ever seeing them. If you want every single AI-proposed action reviewed instead, Human Approval Workflow is built as a standing gate for that.
What does AI Escalation System cost to run each month?
Running it afterwards costs roughly $20 to $100 a month in your own accounts, covering the automation platform plus AI model usage for evaluating each case against the trigger rules. There is no separate platform fee on top of that figure. The exact cost depends on case volume and how often triggers fire.
What access does AI Escalation System need to my systems?
We need admin or API access to every system involved in the workflow, shared by adding our team as a user or creating a key you control, never by email, plus a written description or a screen recording of your current manual process. Five to ten real sample inputs, covering both routine and edge cases, help us set triggers that reflect genuine risk.
Can we change the escalation triggers or routing later?
Yes. Adjusting a trigger threshold, or changing who a specific type of case routes to, usually fits inside the monthly change hours included with maintenance. Adding an entirely new workflow, or a fourth connected system beyond the three in scope, goes through a Change Quote agreed in writing before that work starts.
Can we cancel AI Escalation System maintenance later?
Yes. Maintenance runs month to month with no lock-in, so you can cancel any month once the 30 days of included support has ended. Unused hours do not roll over. If your trigger rules need a bigger rework, that goes through a Change Quote.
Who owns the escalation history AI Escalation System records?
You do. Every escalation and its outcome is logged inside your own connected systems, not a platform we control, so the full history is available for refining trigger rules later. We hand over an admin guide and a recorded session covering how the triggers and routing work, with nothing locked to us.
Guides on this topic
All guides Escalate the cases that need a person, not all of them.
Order AI Escalation System and rules-based escalation to the right person is live on a fixed delivery date.
Works well with
Oversight
Agent Exception Handler
Catches technical failures in AI runs, such as errors, timeouts and unusual runs, and routes them for resolution.