Engineering Portal

Staging and Pilot operating model

Source: deck slides 21–23 · Wrike AI & MCP Readiness Programme

Staging definition and Pilot operating model#

Apply proportionate controls as adoption expands, while keeping the initial Pilot tightly governed. · Slide 21

Pilot: up to 5 agents#

  • Small controlled group
  • Every use case manually reviewed by the Taskforce
  • Write actions allowed; Delete disabled
  • Manual controls acceptable where risk is contained

Rollout: up to 30 agents#

  • Extended controlled group
  • Central monitoring and lifecycle controls required
  • Manual oversight becomes less scalable
  • Close gaps that cannot be managed reliably by process

GA: general availability#

  • Broader availability and decentralised use
  • Scalable support, assurance and reporting
  • Commercial and consumption controls in place
  • Outstanding enterprise blockers resolved or formally accepted

Pilot operating model#

  • Maximum five registered agents
  • Named primary owner and approved use case
  • Taskforce review before activation
  • Approval is time-, scope- and volume-bounded
  • Material changes require fresh approval
  • Wrike write actions permitted
  • Delete technically disabled
  • Approved spaces, fields, tools and action types only
  • Bulk and high-impact actions require separate approval
  • Individual agent disablement and incident route available

Decision rule: a control may be deferred only when the Pilot design removes the exposure or a documented manual control reduces the risk to an acceptable level.

Pilot governance artefacts#

Operational documents required to approve, run, monitor and close the controlled Pilot. · Slide 22

Approval & governance#

  • Pilot submission form
  • Use-case risk and scope review
  • Agent and approval register
  • Time-bounded approval evidence

Readiness & operations#

  • Pilot readiness checklist
  • Access and data boundary review
  • Incident and disablement procedure
  • Pilot exit assessment

Monitoring & assurance#

  • Daily monitoring log
  • Weekly Taskforce review
  • Incident and exception log
  • Activity and decision evidence

Operating sequence#

  1. Submit and assess the use case
  2. Confirm boundaries, controls and approval expiry
  3. Activate only after checklist completion
  4. Monitor daily and review weekly
  5. Modify, disable, escalate or progress based on evidence

Evidence is retained in the Pilot governance workbook and governance pack.

Embedded files on the slide: Pilot Governance Workbook (Excel) and Pilot Governance Pack (Word).

Daily and weekly Pilot monitoring process#

Manual oversight is proportionate for up to five Pilot agents; monitoring becomes more automated at Rollout. · Slide 23

Daily operational review#

  • Confirm approval remains valid
  • Review executions and write actions
  • Review bulk/high-impact actions
  • Check failures, permission errors and unexpected changes
  • Normal: continue
  • Warning: contact owner and review controls
  • Critical: disable agent and escalate

Weekly Taskforce review#

  • Review adoption and benefits
  • Review incidents and exceptions
  • Review bulk, retrieval and cross-system activity
  • Assess control effectiveness and approval expiry
  • Decide for each agent: continue, modify, disable, escalate, prepare Rollout or retire
  • Record decisions, owners and due actions

Evidence is retained in the Pilot governance workbook and governance pack.