Managed AI Implementation Roadmap

AI Maturity Assessment

Establish the company’s current AI maturity baseline before implementation begins.

Discovery Assessment

AI Business Analyst meets with company leadership and key stakeholders to confirm business goals, intended users, approved use cases, security requirements, compliance requirements, and rollout priorities.

Department Discovery Conversations

AI Business Analyst meets with department leaders and AI champions to understand current AI use, pain points, repetitive work, opportunities, expected outcomes, and success criteria.

Select Solutions, Connections & Infrastructure

• Select the initial catalog solutions. Up to five included in tier.
• Determine MCP connections. Up to five included in tier.
• Determine supported non-MCP app connections. Up to six connections included in tier.
• Determine RAG sources and requirements.
• Determine Private LLM workloads and infrastructure requirements.
• Confirm users, business owners, and technical dependencies.

Provide Configuration Documentation

• Client provides approved source documents and data.
• Client defines access boundaries.
• Client provides documentation required to configure chosen solutions, connections, RAG sources, and Private LLM workloads.

Deploy & Configure the Managed AI Environment

• Set up the sanctioned AI workspace of choice.
• ITS provides an Acceptable Use Policy.
• Deploy Shadow AI tooling, enable blocking controls, and configure required reporting.
• Configure blocked-site review, suspected Shadow AI alerting, and prompt-level oversight.
• Deploy the selected pre-built catalog solutions.
• Configure instructions, skills, guardrails, approved source content, approvals, and technical dependencies for selected solutions.
• Deploy and configure up to 5 managed MCP connections.
• Configure up to 6 assisted non-MCP app connections.
• Deploy and configure RAG where purchased and scoped.
• Deploy the Private LLM where purchased and scoped.
• Configure least-privilege credentials and permissions.
• Configure SSO where applicable.
• Deploy MFA if not pre-established.
• Provision users and establish user roles and permissions.
• Complete employee orientation.
• Add client POCs to the monthly AI tip email campaign.
• ITS provides access to the ITS AI video library, online knowledge resources, and historical tip library.
• Invite client POCs to monthly AI BA office hours.
• Establish the monthly 60-minute POC and AI Business Analyst strategy touchpoint.
• Confirm user access, permissions, deployed solutions, RAG and Private LLM configuration, and Acceptable Use Policy rollout.
• Transition the deployed Managed AI environment to Service Desk support with documented escalation paths.

Initial Reporting

• Review initial usage and adoption reporting.
• Review initial Shadow AI reporting.
• Review initial user prompt insights.
• Review blocked-site activity and suspected Shadow AI alerts.
• Identify the next catalog opportunities based on initial adoption findings.

Re-score AI Maturity

Re-score the company’s current AI maturity level after deployment and initial adoption.

Continued Support & Engagement

• Monthly 60-minute AI Business Analyst strategy touchpoint.
• Monthly AI tips via email with access to historical tips through the password-gated webpage.
• Monthly office hours with the AI Business Analyst.
• Continued support and management of ITS-deployed Managed AI infrastructure through the existing Service Desk under ITS’s standard service level agreement.
• Support for ITS-deployed agents, the AI workspace, connections, and integrations, including remediation when a deployed solution breaks, misbehaves, or stops connecting.
• Monitoring for abnormal behavior such as failed runs, error spikes, or solutions that stop producing.
• Defined escalation paths for exceptions and approvals.
• Monthly usage and adoption report reviewed live with the AI Business Analyst.
• Monthly Shadow AI report showing what is in use and by whom.
• Monthly blocked-site activity review.
• Proactive alerting when suspected Shadow AI or policy violations appear between reports.