I need a self-contained AI agent that specialises in technical troubleshooting for our internal IT projects. Its core job is to resolve software errors and identify hardware problems on Windows machines. The agent should analyse logs, event viewer data and user-supplied screenshots, then recommend clear remediation steps or automatically trigger predefined scripts. The workflow I envision is simple: a user opens a ticket in our existing help-desk portal( Servicenow), the agent ingests the ticket content plus any attached diagnostics, runs its reasoning model, and replies with a ranked list of fixes. If it cannot find a match in the knowledge base, it should request the missing details or escalate to a human technician with a concise summary of what it tried and why it failed. I am comfortable hosting the solution on our on-prem Windows Server, so feel free to leverage Python,, or containerised micro-services—whichever produces the fastest, most maintainable result. Integrations with OpenAI or similar LLM APIs are welcome as long as sensitive data remain on our side; sanitisation steps must therefore be built in. Deliverables • A working AI agent (service or executable) capable of live ticket interaction • A curated troubleshooting knowledge base seeded with at least 50 common Windows error patterns covering software faults and hardware misreports • Setup scripts, deployment guide, and clear documentation (including API endpoints, model training approach, and escalation logic) • A short test report demonstrating successful resolution of ten sample issues and one graceful escalation case Acceptance is complete once the agent runs on our staging server, passes the provided test set, and teams confirm that responses are actionable and technically sound.