Give frontline support an AI helpdesk that checks the systems for them: the agent digs across account, order and network tools, proposes the fix, and a person confirms before anything changes on a live customer: AI agent prompt
AWS published a Nov 4, 2026 customer story on U Mobile's AiDaH (AI-Driven Agentic Helpdesk) on Amazon Bedrock AgentCore. Staff ask in Microsoft Teams; specialist agents cover CRM, problem management, number portability, orders, infrastructure and analytics across 150+ scenarios. Humans still approve any change on a live subscriber. Internal measurements showed provisioning and network tickets down up to 84.5%, general inquiries 81.6%, and billing 48.7%.
The prompt
1. List the top 10 support issues that force an agent to open 3 or more systems (for example: roaming data failure, failed add-on purchase, port-in status, pending order blocking a change). 2. For each issue, write: which systems to check, what a good answer looks like, and which actions need a human confirm before anything changes on a live account. 3. Connect an AI agent platform your company allows (Amazon Bedrock agents, Claude, ChatGPT enterprise, or similar) to those systems with the smallest read permissions that work, plus write actions only behind an approval step. 4. Put the agent where staff already work (Teams, Slack or your ticketing UI) and give them short prompt templates, not a blank box. 5. Ask: "For this customer issue, check each system on the checklist, return the root cause in plain English, list the evidence with source, and propose the remediation steps. Do not change anything until I confirm." 6. Track ticket volume and time-to-first-useful-answer by category for 30 days, and only expand to a new domain when accuracy and adoption look solid.
How to use it
- Copy the prompt above.
- Replace every [bracket] with your own details (or fill in the blanks on AgentPlaybook Real Estate, which remembers your answers for the next prompt).
- Paste it into Amazon Bedrock AgentCore or Claude/ChatGPT enterprise agents, Microsoft Teams or Slack, your CRM/BSS APIs and review the answer before you use it.
Works with: Amazon Bedrock AgentCore or Claude/ChatGPT enterprise agents, Microsoft Teams or Slack, your CRM/BSS APIs, knowledge base
Real-world example: aws.amazon.com/blogs/machine-learning/how-u-mobile-built-agentic-telco
Before you use it: what to check
Already built into this prompt
- For each issue, write: which systems to check, what a good answer looks like, and which actions need a human confirm before anything changes on a live account.
- Do not change anything until I confirm." 6.
- Ask: "For this customer issue, check each system on the checklist, return the root cause in plain English, list the evidence with source, and propose the remediation steps.
Check yourself
- This prompt works on text other people wrote (emails, reviews, messages or web pages). That text can hide instructions aimed at AI, such as "ignore your rules and…". Paste it as material to work on, and never let the AI act on requests inside it, like paying, sharing logins or changing records. How prompt injection works →
- Remove names, contact details, account numbers and other personal data the task doesn't need before you paste, and check your AI tool's data and training settings.
- Check names, dates, numbers and any claim against your own source material before you use the answer. AI can misread or invent details.
Using an AI agent? Download this prompt as a SKILL.md agent skill with these checks built in, or get the .zip for Claude.ai. How to install skills.
Want to make this prompt even better? Grade it with the free prompt grader or build your own with the AI prompt generator.
32 copy-paste agent prompts (PDF) plus 5 new prompts every week.
Get it free