Proposed approach
First year
I would use the first year to understand what already works, choose a small portfolio, prove value, stop weak projects and move the strongest tools into normal work.
Resource assumption: the Director coordinates a small cross-functional group using existing practice, IT, DataTech, Legal, Risk, Cybersecurity and quality capacity. Each phase depends on named staff time and reviewer availability. This plan assumes neither a confirmed organizational chart nor a budget; hiring and investment recommendations follow demonstrated need.
First 30 days
Understand before changing
- Meet senior leaders, practice leaders, IT, DataTech, Legal, Risk and project teams
- Inventory existing AI tools, pilots and unofficial usage
- Review data access and client restrictions
- Identify high friction workflows and establish baseline measures
- Map what is already working and should not be disrupted
Concrete deliverables
- Current state inventory covering tools, owners, usage, contracts and overlapping efforts
- Baseline measurement plan with paired task timings and quality checks
- Source and access readiness assessment for candidate pilot data
- Proposed portfolio criteria and intake path
- Proposed owner roles
- Director of AI Strategy & Adoption, supported by IT, DataTech, practice leaders, Legal, Risk and Cybersecurity.
- Dependencies
- Access to existing inventories, contracts, project teams, source owners and enough staff time for interviews and baseline observation.
- Phase exit decision
- Proceed only when leadership agrees on portfolio criteria, candidate workflows have usable baselines, and source owners confirm that access and confidentiality can be assessed.
Days 31 to 90
Make the first choices
- Set the use case intake and review process
- Launch no more than three well scoped pilots with explicit go or no-go criteria
- Start an AI champion network
- Create the first executive scorecard
- Define approved, controlled and restricted use
Concrete deliverables
- A charter for each pilot naming the workflow, users, business owner, data owner, reviewer and measures
- A build, buy, partner or wait recommendation with full cost categories
- Documented permission tests, human review steps and a release decision
- An initial scorecard with baselines and targets agreed during discovery
- Proposed owner roles
- The Director coordinates. Practice sponsors own workflow value, IT owns platform readiness, DataTech owns technical delivery where assigned, and Legal, Risk and Cybersecurity review controls.
- Dependencies
- Named practice sponsors, available subject matter experts, approved environments, source permissions and vendor or procurement review where needed.
- Phase exit decision
- Each pilot starts only with an approved charter, accessible sources, a named reviewer, baseline measures and explicit go or no-go criteria. Otherwise, defer or rescope it.
Days 91 to 180
Evaluate and narrow
- Compare pilot results with the measures set at the start
- Use independent subject matter and risk reviewers who were not responsible for building the system
- Make a documented stop, fix or expand decision for each pilot
- Improve integration with normal workflows and expand role specific training
- Add technical staff only where the portfolio shows a real need
Concrete deliverables
- A fixed evaluation set with source supported answers and subject matter review
- Paired baseline and pilot results that include review and rework
- Permission, security and human review test results
- A decision record covering value, quality, risk, full cost and support readiness
- Proposed owner roles
- Practice and quality reviewers evaluate usefulness independently. IT and DataTech provide technical evidence. Risk, Legal and Cybersecurity evaluate controls. The Director recommends the portfolio decision.
- Dependencies
- Sufficient pilot usage, comparable baseline cases, reviewer capacity and complete cost and incident records.
- Phase exit decision
- Expand only when agreed value and quality measures hold, permission tests show zero unauthorized disclosures, and support ownership is credible. Fix a specific weakness or stop the effort.
Days 181 to 365
Move useful work into production
- Put proven systems into supported production and retire weak experiments
- Establish recurring leadership review
- Refresh the next year portfolio
- Coordinate client facing opportunities with DataTech
- Share external lessons only where Hazen has real evidence and experience
Concrete deliverables
- A service record for each production tool naming business, technical, monitoring and support owners
- Operating procedures for access changes, incidents, evaluation after material changes and vendor review
- A monitoring schedule and recurring executive scorecard
- A next year portfolio that continues, expands, fixes or stops each effort
- Proposed owner roles
- Business owners own workflow outcomes. IT owns enterprise support where assigned. DataTech owns technical services where assigned. Practice and quality owners retain professional review.
- Dependencies
- Support capacity, monitoring access, training, vendor commitments, change management and agreed operating ownership.
- Phase exit decision
- A tool becomes a supported service only when ownership, monitoring, support, access review and incident response are operating. The year closes with a funded or deferred recommendation, not an assumed budget.
Year one success is not the number of pilots. It is having a few useful systems in normal work, clear controls, evidence of value and a better process for choosing what comes next.