Stop drowning a tiny platform-support team: put an AI agent hub in front of failed jobs and how-to questions so it troubleshoots, checks standards, and escalates with context—people only handle judgment calls: AI agent prompt
Dataiku published a Nov 4, 2026 customer story on ENGIE Supply & Energy Management. About 250+ Dataiku users (roughly 80% non-technical) across 25 servers were overwhelming a three-person support team with failed jobs, connection issues, admin requests and standards questions. ENGIE built an Agent Hub that coordinates four specialized agents—troubleshooting, guidance, knowledge retrieval and human escalation—behind ~24 purpose-built tools so users bring a problem once and the hub routes it. The company estimates about a 20% productivity gain for that three-person team (roughly a day a week back). Routine fixes can run automatically; a person joins when judgment matters.
The prompt
1. List the top 10 repeat tickets your internal data/BI platform support team gets (failed job, bad connection, “does this follow standards?”, access request). 2. For each, write what the agent may do alone vs what must escalate to a person, and which system tools it may call (logs, connection config, project standards check). 3. Stand up a single front door (chat or ticket form) that routes to specialist agents for troubleshoot / guidance / knowledge / escalate—users should not pick the agent. 4. Start read-only: agent diagnoses and proposes the fix; a support person confirms before any write. 5. Track time-to-first-response and tickets closed without a human, then unlock one safe write action at a time (restart job, reopen connection) with an audit log.
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 Dataiku Agent Hub (or similar platform agents), internal knowledge base, ~server/job/admin tools and review the answer before you use it.
Works with: Dataiku Agent Hub (or similar platform agents), internal knowledge base, ~server/job/admin tools, human escalation channel
Real-world example: dataiku.com/blog/engie
Before you use it: what to check
Already built into this prompt
- For each, write what the agent may do alone vs what must escalate to a person, and which system tools it may call (logs, connection config, project standards check).
- Stand up a single front door (chat or ticket form) that routes to specialist agents for troubleshoot / guidance / knowledge / escalate—users should not pick the agent.
- Track time-to-first-response and tickets closed without a human, then unlock one safe write action at a time (restart job, reopen connection) with an audit log.
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