Engagement Roles & Responsibilities

govrn at MCG/Kelly — three core roles. One engagement, clear accountability.

Overview

Every AI-governance engagement has three personas — Sales (win the deal), Solution Architect (design it), Delivery Lead (build it) — with clear handoffs and overlapping moments where they work together.

1. Sales

Who

Director of Sales / Technology or account executive leading the wedge motion.

Before the deal (pre-sales)

• Identify the wedge opening: where is governance urgent? (Fair housing? AI board mandate? Regulatory pressure?)
• Map the stakeholders: who's the champion (CTO/Chief AI Officer)? Who's the decision-maker (CIO/CISO)? Who pays (CFO)?
• Present the Kelly pitch: the vision (governance-as-a-wedge, discovery dividend, whole-account play).
• De-risk the conversation: "here's what it costs, here's the timeline, here's what you get on day one."
• Land the discovery baseline: a scoped, paid initial engagement (2–3 weeks, 2–3 people). This is the first contract, the proof, the foot in the door.

During the deal (kickoff)

• Lead the kickoff presentation (with Solution Architect) to the client's IT/Risk/AI stakeholders.
• Confirm success criteria: what does "done" look like for discovery? (Inventory count? Shadow-AI finding count? Risk posture baseline?)
• Establish the cadence: weekly syncs? Monthly boards? Who reports to whom?
• Introduce Delivery Lead to the client's engineering team. Make it warm.

After discovery (transition to build)

• Socialization: take the discovery readout to the client's board. Frame it as "here's the problem we found + here's the investment to fix it."
• Expansion deal: convert the discovery into a Phase 2 (assessment) and Phase 3 (implementation) contract. Bigger deal, longer timeline.
• Nurture the sponsor: keep the champion (CTO/Chief AI) engaged. They're the one who keeps the lights on when the work gets hard.

Ongoing (operations & expansion)

• Quarterly business review: posture trend, remediation progress, new risks discovered, next phase of work.
• Expansion hooks: "while we're in here governing AI, we noticed your X is broken — want to add Y to the scope?" (Discovery dividend in action.)

Success metrics (Sales owns)

• Deal velocity: discovery → assessment → implementation contracts all signed within 6 months.
• Deal size: starts at $X (discovery), grows to $5X (full engagement) by month 3.
• Sponsor health: champion is happy, engaged, expanding scope.

2. Solution Architect

Who

Senior technical architect (Principal Engineer, Staff Engineer) with systems design + vendor knowledge.

Before the deal (pre-sales)

• Due diligence: tech stack audit. (What's their cloud? On-prem? Nexthink already deployed? Other telemetry?) This shapes the discovery approach and the cost.
• Design sketch: rough architecture for discovery + assessment + implementation, tailored to their setup.
• Technical Q&A: answer the CTO/Chief AI Officer's questions. ("How long does discovery take?" "Can we integrate with Nexthink?" "Will you touch our prod?")

During discovery (kickoff + execution)

• Lead discovery design: what tools, what data sources, what queries will we run? (SA owns this.)
• Oversee discovery execution: make sure the discovery team is getting clean data, not junk.
• Daily standups with Delivery Lead (the engineer doing the work) — unblock fast.
• Readout prep: help the engagement lead turn findings into a narrative. ("Here's the architecture, here's why we found what we found.")

During assessment (SA designs the scoring)

• Control library curation: which standards matter for this client? NIST AI RMF? EU AI Act? Industry-specific (health/finance/energy)? Build the control set to match their risk profile.
• Scoring methodology: how do we map discoveries to control gaps? What's a critical finding vs. a nice-to-have?
• Findings narrative: why does this client have this gap? (Is it architectural? Process? Compliance-driven?)
• Roadmap design: if we prioritize fixes by criticality, what's the 90-day plan? The 6-month plan?

During implementation (SA designs, Delivery Lead builds)

• Architecture review: control registry schema, APIs, monitoring hooks. Make sure the system is extensible, not brittle.
• Technical decisions: Nexthink vs. custom discovery? Postgres vs. another DB? Lambda vs. ECS for monitoring agents?
• Integration design: how does the client's existing stack (ServiceNow? Datadog?) talk to our platform?
• Go-live readiness: testing plan, cutover plan, rollback plan.

During operations (SA owns the roadmap)

• Monitoring hygiene: is the monitoring working? False-positive rate acceptable?
• Standards tracking: when NIST publishes new AI RMF guidance, we integrate it into the control library (SA drives this).
• Performance: is the dashboard fast? Is the API responsive? Can it handle scale?

Success metrics (SA owns)

• Technical debt avoided: clean architecture, extensible design, no major refactors needed in ops phase.
• Integration friction: the more seamlessly we integrate with their existing vendors (Nexthink, ServiceNow, etc.), the faster implementation.
• Reusability: the control library, scoring methodology, and monitoring patterns we design for Client A should be 80% reusable for Client B.

3. Delivery Lead

Who

Full-stack engineer (mid-level+) or engineering manager who owns the execution sprint-by-sprint.

Discovery phase (Delivery Lead = the executor)

• Data collection: run cloud SDKs, query APIs, parse logs. Own the harness that fetches data.
• Normalization: map vendor APIs into our unified schema. (If they have Nexthink, we ingest its API; if they have Datadog, we ingest that.)
• Quality checks: is the inventory complete? Any obvious gaps? Flag them to SA.
• Time tracking: log hours against the discovery budget. If we're overrunning, flag early.

Assessment phase (Delivery Lead = the analyst + coder)

• Control scoring: for each asset/model, score it against each control. Delivery Lead runs the assessment engine (SA designed it).
• Evidence gathering: for each finding, fetch the source (log entry, config, policy document). Build the audit trail.
• QA: test the scoring logic. Do the scores make sense? Sanity-check the high-risk findings.
• Readout prep: prepare data exports (CSVs, JSON) for SA to turn into narratives.

Implementation phase (Delivery Lead = the builder)

• Registry implementation: build the control database, APIs, intake workflows. (SA designed, Delivery Lead codes.)
• Monitoring agents: write the Lambda/Functions that watch for drift. Deploy them. Test them.
• Dashboard integration: consume the registry APIs, build the operational views.
• Testing & hardening: unit tests, integration tests, load tests. Make sure the system is production-ready.
• Deployment: run through the cutover plan, deploy to client infrastructure, train their ops team to run it.

Operations phase (Delivery Lead = the SRE)

• Day 1 support: be on-call for the first week. Fix bugs, handle escalations.
• Monitoring: are the agents running 24/7? Any failures? Are alerts firing correctly?
• Performance: is the dashboard snappy? Are API response times acceptable?
• Handoff: document the system. Train the client's ops team to manage it. Step back.

Success metrics (Delivery Lead owns)

• On-time, on-budget: discovery ships in week 3, assessment in week 7, implementation in week 15. Stays within headcount budget.
• Zero production incidents: the system we deploy is stable. No "emergency fixes" needed in ops.
• Team velocity: we learn from Client A, so Client B is 30% faster to deliver. Reusable code, patterns, processes.

Handoff moments (where roles overlap)

A note on ownership

• Sales owns the relationship. If the client is unhappy, Sales fixes it (sometimes by getting SA/Delivery to move fast, sometimes by managing expectations).
• SA owns the design. If the system is too slow or too brittle, SA did something wrong. SA fixes it or owns the credit for great decisions.
• Delivery Lead owns the execution. If we're over budget or behind schedule, Delivery Lead is the one who escalates. They own the build quality.
• All three own the success. A great engagement is Sales landing the deal, SA designing it right, and Delivery building it fast. Miss any one, and the engagement suffers.

Typical staffing for a standard engagement (mid-market)

Questions for your team

• Who's your Sales for this motion? (Is it Richard? A sales leader? An account executive?)
• Who's the SA? (Is there a principal architect at Kelly/MCG who owns the design?)
• Who's the Delivery Lead? (Do you have a staff engineer ready to own the build?)
• How do you handle handoffs? (Weekly sync of the three? Daily standup? Async Slack?)

These three roles scale from one engagement to a hundred.