Public documentation for governed AI labor
SDKs/Governance/Connectors
Arx / Docs / ARX × McKinsey & Company — Channel partner brief

Documentation

ARX × McKinsey & Company — Channel partner brief

arxsec-site / partners/variants/mckinsey-ope.md

arxsec-site repo-root partners/variants/mckinsey-ope.md

To: [Senior Partner name], McKinsey & Company · Operations Practice / QuantumBlack From: Mershard Frierson · Founder, ARX Date: May 2026 Length: 2 pages, 5-minute read

> *We're proposing a 90-day co-pilot engagement: McKinsey commits two senior partners + one QuantumBlack technical SME and one named F500 prospect; ARX commits a free pilot license, a dedicated solutions-engineering pair, and co-branded delivery surfaces. If the pilot lands the prospect as a paying customer, we move to a co-sell agreement with 20% rev share on closed deals plus joint go-to-market on the resulting reference. The artifact you'd take into the next boardroom is concrete, not slideware.*

---

1. The boardroom problem you have today

Every F500 transformation engagement McKinsey OPE runs in 2026 ends in the same conversation. The customer's CEO asks: "what's our AI workforce plan, and when do we report measurable productivity to the board?" Your team has the strategy answer — you've been the firm that defines AI-workforce strategy at the F500 level, with QuantumBlack on the AI side and OPE on the operating-model side. The operational answer — *the deployable infrastructure that delivers 10,000 governed agents across the customer's actual platform mix, with audit-ready evidence the board can defend* — does not exist on the customer's side, and it does not exist as something McKinsey currently sells.

Three risks sink the deal at procurement every time:

  1. Governance. "Who's accountable when an agent does something it shouldn't?" The customer's CISO needs a real answer. Today's answer — "we'll define a governance framework in workstream 3" — is a slide, not a system.
  2. Audit defensibility. "Show me what our auditor sees." Without a tamper-evident, externally-verifiable audit chain, the customer's audit committee won't sign off, and the engagement stalls in legal review.
  3. Kill switch. "If something goes wrong, how fast can we turn it off?" A scope-of-work line item that says "kill switch capability" doesn't pass the CISO's first question.

These are not strategy gaps. They are infrastructure gaps. They are why the typical $30–80M McKinsey AI-transformation engagement closes as a $8–15M strategy + diagnostic engagement instead, with the implementation handed to a partnered SI who captures the multi-year deployment economics.

---

2. What ARX is

A vendor-neutral control plane that issues per-agent cryptographic identities, enforces action-level policy in-process at every agent runtime, generates auditor-verifiable compliance evidence, and propagates kill switches across every system the agent has touched. Across the seven commercial agent platforms (Salesforce Agentforce, Microsoft Foundry, Google Gemini Enterprise, AWS Bedrock AgentCore, ServiceNow AI Agents, UiPath, IBM watsonx Orchestrate) and the four open frameworks (LangChain, CrewAI, AutoGen, OpenAI Agents SDK).

ARX is not a model vendor and is not a competitor to McKinsey's strategy work. It is the infrastructure layer beneath it. Today, the customer's procurement team has no third-party choice for cross-platform agent governance — Microsoft only governs Microsoft, AWS only governs AWS, Salesforce only governs Salesforce. The customer with a multi-platform agent estate has no single answer. ARX is that answer. Where QuantumBlack provides the AI talent and OPE provides the operating-model rigor, ARX provides the substrate that makes the rigor enforceable.

---

3. The boardroom moment we want to enable

In your next F500 boardroom — say, [the engagement you're scoping with the [Customer] CFO this quarter] — the working session looks different.

Today: you walk in with a 60-slide deck, hypothetical productivity numbers, and a 24-month roadmap. The board asks two clarifying questions and tables the decision for the next quarter.

With ARX: you open the Engagement Canvas live in front of the customer's exec team, drag in their actual functions (Sales, Engineering, Customer Success, Finance, Operations) parameterized with their actual headcount and current systems, and the cost / productivity / connector-availability model recomputes in real time. You toggle conservative / balanced / aggressive scenarios with a single click. The customer's CFO can see, on screen, that 19,900 governed agents at the balanced scenario yield $1.92B in annualized recovery against $84M in platform + operating cost — sourced from the customer's own numbers, not from a McKinsey template.

You then click "Preview as board." The customer's CEO sees a projected board view of Q3 2027 — cost recovered, productivity gain, audit defensibility (78 of 113 SOC 2 controls hash-pinned to the actual ARX release), open risk register, plan-vs-actual against your modeled cohort deployments, peer benchmark against McKinsey's portfolio of similar engagements (anonymized; firm-private benchmark IP).

The board sees a tangible artifact, not a slide. Your engagement closes that quarter, not next.

The Engagement Canvas and Board View are real and demonstrable today. We can show them on a Zoom in the next 14 days.

---

4. The economic model

The structural reason this works for McKinsey is that an ARX-anchored engagement is a larger engagement, with longer-tail revenue, with outcome-tied upside that today's strategy-only engagement model cannot capture. McKinsey's typical deal-size band runs at the upper end of the F500 strategy market.

| Engagement component | Today (McKinsey strategy-only) | With ARX (anchored) | Δ | |---|---|---|---| | Initial strategy + design | $8–15M [¹] | $10–18M (richer outputs justify higher fees) [²] | +$2–3M | | Integration / deployment delivery | $0–3M (handed to SI partner) [³] | $20–45M (retained via QuantumBlack Labs / McKinsey Implementation; or revenue-shared with named SI) [⁴] | +$20–42M | | Outcome-tied success fees | $0 (no defensible measurement layer) | $8–20M / yr for 3 yrs (productivity tracked by ARX) [⁵] | +$24–60M over 3 yrs | | Post-deployment optimization retainer | $0 (transferred to customer's internal team) | $4–10M / yr (continuous optimization on ARX dashboards) [⁶] | +$12–30M over 3 yrs | | ARX revenue-share on platform ARR | $0 | 20% of ARX ARR (typical $300K–$1.2M / yr per F500 account) [⁷] | +$0.9–3.6M over 3 yrs | | Total per F500 engagement, 3-year basis | $8–18M | $70–160M | +$60–140M [⁸] |

Three engagements per year through this channel = $180–420M of incremental McKinsey revenue over a 3-year window [⁹], on top of an already-budgeted strategy practice. The infrastructure investment from McKinsey's side is two senior partners + one QuantumBlack SME for a 90-day pilot.

> *Footnotes [¹]–[⁹] — sourcing and confidence labels for each cell — are in partners/economic-model-assumptions.md. Every numeric range is open to revision based on McKinsey's view of actual engagement economics; the sidecar exists to make our math transparent, not to argue with the partner who's run more of these engagements than we have.*

The ARX product takes a fixed share of platform ARR. McKinsey takes the services upside, the success fees, the ongoing retainer, and the rev-share. The structural alignment matters: ARX wins when McKinsey's engagements actually deploy and produce measurable outcomes. We are not competing for the customer's wallet. We are sized to make McKinsey's engagements deliverable, not to capture services revenue.

---

5. The 90-day pilot — what we're asking, what we commit

We're asking McKinsey to:

  • Assign two named senior partners + one QuantumBlack technical SME to the pilot (rough order of magnitude: 0.3 FTE total over 90 days)
  • Engage one named F500 prospective customer where you have an active or imminent AI workforce conversation
  • Co-deliver one boardroom-grade engagement using the Engagement Canvas + Board View
  • Provide a written post-pilot evaluation we can use jointly to refine the channel motion

ARX commits:

  • Free pilot license for the named customer for the 90-day window
  • Dedicated solutions-engineering pair embedded with your delivery team
  • Co-branded Engagement Canvas + Board View ("ARX × McKinsey")
  • Reference rights for joint case study + co-marketing if the pilot lands
  • 20% revenue share on the customer's contracted ARX ARR if the pilot converts to a paying customer
  • Mutual non-poach + standard IP terms (your IP stays yours; ARX IP stays ARX; the engagement-specific configuration is jointly owned)

Pilot success criteria (we agree on these in week 1, both signatures):

  1. The customer's CFO or CHRO sees a live demo of the Engagement Canvas + Board View by day 30.
  2. The customer signs a paid trial (≥$50K, ≥6 months) by day 90.
  3. McKinsey internally validates the channel economic model with one practice leader by day 90.

If we hit 2 of 3, we move to a co-sell agreement. If we hit 3 of 3, we move to a co-sell agreement with named-account exclusivity in your selected industry vertical for 12 months.

---

6. Why this conversation, why now

The window is bounded. Microsoft Defender, AWS Identity Center, and Google Agent Space all have native cross-platform agent governance roadmaps with first-party announcements expected at Microsoft Ignite October 2026 and AWS re:Invent December 2026. The 12–18 month window before hyperscaler bundling closes is the window in which the consulting channel for cross-platform governance gets established. McKinsey's QuantumBlack arm rode the analytics-platform wave with similar partnerships in 2014–2018 (Palantir, Anaplan, Dataiku) — the firm that established the channel position first earned durable share.

There is one cross-platform vendor-neutral agent governance platform that is shippable today. ARX is it. We are sequencing one to two consulting-firm channel partners through 2026 to ensure delivery quality on the first engagements; we are in active conversation with a small set of firms and expect to sign the first by end of Q3 2026. We would prefer the first to be McKinsey.

---

7. Next step

A 30-minute working session inside the next 14 days, with the goal of agreeing whether the 90-day pilot proceeds and identifying the candidate customer.

I have time blocks Monday/Wednesday/Friday next week, US Eastern. [Calendly link / direct calendar invite to follow.] If you'd prefer to bring a QuantumBlack technical lead to that conversation, we will bring our head of engineering to match. We can demonstrate the Engagement Canvas + Board View live; the artifact you'd take into [Customer]'s next board meeting will be on the screen.

Looking forward.

— Mershard

---

Appendix — Four things McKinsey partners will ask, with the answer

"How is this different from [hyperscaler]'s native governance?" Hyperscaler governance is platform-native (Microsoft governs Microsoft, AWS governs AWS, Google governs Google). The customer with a multi-platform agent estate — i.e., every F500 — has no single answer. ARX is vendor-neutral and built specifically for the cross-platform case. Hyperscalers will eventually compete on this, but they will not bundle each other's platforms; the cross-platform vendor-neutral position is structurally available only to a third party.

"What's the technical maturity? Can we put this in front of a CISO?" ARX has a working backend (~15K LOC FastAPI, 105 production connectors, real intercept layer, real audit, real compliance generator), is hosted on Aptible (SOC 2 Type II / HIPAA / ISO 27001 — fully inherited for infrastructure controls per docs/compliance/soc2-mapping.md), and is pursuing AIUC-1 Type II attestation for the platform layer with readiness assessment in progress and Type II target Q3 2026 (per docs/compliance/aiuc-1-readiness.md). SOC 2 control mapping for shared-responsibility controls is documented and live. The Engagement Canvas and Board View are live in-product surfaces. We are pre-Series-A — appropriate due-diligence applies — but the technical substrate is real, not slideware. We will sit through a CISO's technical review with QuantumBlack's SE in the room.

"McKinsey doesn't typically retain implementation. Doesn't ARX assume we do?" This is the most important question and we want to answer it directly. The integration / deployment row in the economic model ($20–45M) does not have to land at McKinsey directly. Two paths work:

  1. QuantumBlack Labs / McKinsey Implementation retains it. This is the natural fit if the engagement is McKinsey-led end-to-end. ARX as substrate makes the implementation more configuration than custom-build, which is closer to how QuantumBlack Labs is staffed than traditional SI work.
  2. McKinsey partners with one named SI on the integration and shares the revenue. McKinsey leads strategy + governance + outcome accountability; the named SI (Accenture / Deloitte / Cognizant) leads physical delivery; ARX is the substrate both sides build on. The economic split is per the partnership agreement; McKinsey typically retains 25–40% of the implementation work via PMO + governance + outcome-tracking roles. This still represents a 6–8× expansion vs. the strategy-only engagement.

The model spreadsheet supports both scenarios; we'll work through it together in the technical follow-up.

"What's the founder's commitment to the channel motion vs. direct sales?" The channel motion is the company strategy, not a side bet. We are not running a direct enterprise sales team in parallel that would compete with McKinsey's account leads. Every ARX deal in 2026–2027 routes through a consulting-firm channel partner or a designated systems-integrator. This is a structural commitment in our hiring plan and our cap table — we will share the operating plan in the technical follow-up.