Big-SI Cortex Positioning and the Mid-Market Gap phData's CAF Can Own
The question
How do major Snowflake SI partners (Slalom, Accenture, Deloitte) position their Cortex AI/Intelligence offerings for enterprise clients, and what differentiable mid-market gap can phData's CAF own? Context: phData's CAF (Catalyst Assessment Framework — the assessment-to-build catalog/agentic framework) is the founder's claimed differentiator in Snowflake AI consulting, but the large SIs' actual pitch has never been mapped — so we are differentiating against an unknown target while the KnowBe4 workshop goes live this week.
What we already know (from the vault)
- CAF is an assessment-and-productization framework spanning the sales→delivery seam, not a demo. It is a library of ~106 prescriptive plan-only skills (being collapsed to ~15-20 objective skills + orchestrators + on-disk engagement manifests) that walks a consultant through assessing an engagement and emits a build manifest. First half accelerates pre-sales discovery/scoping; second half productizes delivery. See [[2026-06-09-caf-restructure-proposal]] and [[use-case-niche-ledger]].
- phData's delivery spine is the skills-first Cortex Agent workflow (10-step build+eval:
setup→create-views→create-semantic-view→create-agent→deploy+ golden-set/TruLens eval), wrapping Snowflake-native primitives (Cortex Search + Cortex Analyst + Cortex Agents) with phData IP. Typical shape: 8-12 week PoC-to-production, est. $150-400k PoC / $500k-$1.5M follow-on. See [[2026-05-20-phdata-cortex-agents-practice]]. - phData is an Elite Snowflake Consulting Partner / 2025 Partner of the Year but is NOT in top-10 enterprise-AI consultancy surveys — its path to agentic revenue runs through Snowflake-partnership leverage and the mid-market band, not head-to-head with the Big-4. The vault already names the four deployment archetypes and places phData in "specialist consultancy ($150-500k, mid-market + F1000)." See [[2026-05-21-enterprise-ai-agent-deployment-paths]].
- The KnowBe4 workshop IS the CAF pattern made tangible: Cortex-Agent-generated semantic view → custom Snowflake MCP (RBAC-governed) → Claude Cowork, built fast with Cortex Code; medallion (bronze/silver/gold) spine teaching "model the ontology once, activate everywhere." See [[2026-06-26-knowbe4-snowflake-workshop-pitch]].
- phData's public accelerator IP (Customer Data Platform Accelerator, Clinical Cohort Creation Accelerator, the "phData Forge" AI-native delivery system) is the named, productized layer — the same category the Big-3 sell, but narrower and stack-deeper. See [[2026-06-26-knowbe4-snowflake-workshop-pitch]].
What the web says
- Accenture = the enterprise-scale, C-suite-reinvention play. The Accenture Snowflake Business Group (formed Dec 2025) is positioned as Snowflake's "leading global partner for Cortex Code and Snowflake Intelligence," anchored by the AI Refinery™ platform, a jointly-funded global Center of Excellence, and 5,000+ SnowPro-certified people — the largest certified talent pool in the ecosystem. Pitch language is reinvention/transformation aimed at executives ("85% of C-suite plan to increase AI investment"), with marquee logos (Caterpillar). Differentiation = scale, industry breadth, and AI Refinery as the orchestration substrate. (Snowflake/Accenture press release, Accenture page)
- Deloitte = packaged, vertical/function-specific AI products with hard ROI numbers. Lead asset is "AI Advantage for CFOs" — a finance-ops platform with "14+ modules and 30+ intelligent AI agents" on Snowflake's Cortex suite, sold on quantified outcomes ("10-30% forecast-accuracy improvement," "6-month ROI," "$100K-$500K+ annual recurring value per enterprise client"). Differentiation = deep industry/function expertise pre-built into productized solutions for a named buyer (the CFO), horizontal across enterprise finance. (Deloitte AI Advantage for Snowflake, Deloitte/Snowflake gen-AI)
- Slalom = "fiercely human," outcome/value-modeling consulting. Seven-time Snowflake Partner of the Year (incl. 2025 Global AI Partner of the Year). Named offerings: Enterprise AI on Snowflake (Cortex Code + Intelligence), an AI Value Platform (forecasts TCO/ROI/risk before build), Zero Legacy (AI-assisted modernization), and a Snowflake Delivery CoE for "repeatable best practices." Differentiation = consultative change-management + value-case rigor + industry solution patterns, brand-led on the human-centered transformation angle. (Slalom Snowflake Summit, Slalom on Cortex Code)
- The platform itself is racing up the stack to the "control plane for the agentic enterprise" (Snowflake Intelligence + Cortex Code expansion, Apr 2026; Snowflake Intelligence hit 2,500+ accounts in three months). The SIs are all repositioning their delivery on top of a substrate that is itself moving from "answer questions" to "take action." (Snowflake control-plane release)
- All three big SIs converge on the same three pillars: (1) a named platform/accelerator (AI Refinery / AI Advantage / AI Value Platform), (2) a certified-talent + CoE scale story, and (3) executive-outcome language (reinvention, ROI, value). None publicly expose a repeatable assessment-to-build engagement engine as the product — accelerators are pre-built solution endpoints, not a methodology that travels client-to-client.
Convergences and contradictions
- Convergence: Vault and web agree the Big-3 compete on scale + named solution accelerators + executive ROI framing, and that phData is deliberately a tier below them (Elite partner, mid-market band, deeper stack credibility per Snowflake $). The vault's "specialist consultancy" archetype is confirmed by the web's scale gap (Accenture's 5,000 SnowPro-certified people vs phData's boutique footprint).
- Contradiction / sharpening: The Big-3's accelerators are solution endpoints (a CFO finance platform, an industry CoE deliverable). phData's CAF is a different object class — an assessment-to-build engine that decides which agents to build and productizes the delivery of building them. The big SIs sell the destination; CAF sells the repeatable road to the right destination. That is not a smaller version of what Accenture/Deloitte pitch — it is a different layer, and that is the whitespace.
- Watch-item: Accenture's AI Refinery is the one big-SI asset that could drift toward CAF's territory (it is positioned as orchestration, not just a solution). Monitor whether AI Refinery starts being sold as an assessment/build methodology vs. a runtime.
Synthesis for RDCO
The ownable gap is the "right-sized, repeatable, mid-market assessment-to-build engine" — sold as method, not as a pre-baked solution endpoint. The Big-3 all win the enterprise on scale and named solution accelerators, but every one of those accelerators is a destination product: Deloitte's CFO finance platform, Accenture's industry CoE deliverable, Slalom's value-case-plus-modernization. None of them publicly productize the front of the funnel — the disciplined "assess the estate, classify the use cases, route by autonomy, and emit a build manifest" motion that decides what to build before anyone builds it. CAF is exactly that engine, and it spans the sales→delivery seam (assessment artifacts become the delivery resume-point via the engagement manifest). For a mid-market buyer ($150-500k, KnowBe4-sized 2,500-3,000-employee shops) the Big-4's $300k-$2M floor and 6-12 month pre-production cycle are the wrong shape; they want a fast, governed, repeatable path that a small expert team can run again next quarter. That is CAF's natural home.
The sharpest differentiation to put in the ledger and the demo is "semantic-layer-once, activate-everywhere, governed by RBAC, built in a weekend" — which is precisely the KnowBe4 build (Cortex-Agent-accelerated semantic view → custom MCP → Cowork, RBAC in place). Against Deloitte's "14+ modules, 30+ agents" packaged platform, phData's counter is not "we have more modules" (it would lose that arms race) — it is "we install a repeatable factory in YOUR environment that produces exactly the agents your ontology needs, and you can run it again without us." Against Accenture's scale, the counter is time-to-value and right-sizing: the weekend-built demo is itself the proof that a small expert team beats a 5,000-person CoE on a mid-market clock. The KnowBe4 medallion talk-track (model events/entities once at silver, expose the gold semantic layer everywhere including MCPs) is the consumable shape of CAF's underlying thesis, so the workshop doubles as live evidence of the wedge.
This sharpens the Use-Case → Ownable-Niche Ledger along two axes. First, it confirms the ledger's filter (recurs / carries budget / wedge a big SI won't serve well): the "assessment-to-build engine for mid-market Cortex adoption" is structurally a slice the Big-3 won't chase because their economics depend on large named-solution engagements, not on selling a repeatable method to sub-$500k accounts. Second, it gives the ledger a positioning vocabulary — every use-case CAF surfaces should be scored on whether it is a destination product (Big-3 territory, avoid) or a repeatable build pattern (CAF/RDCO territory, pursue). The owned-product wedge RDCO is hunting is more likely to live inside CAF's recurring use-case archetypes (the same agent shape showing up across mid-market clients) than inside any single client solution.
One honest caution carried from the vault: CAF is phData-proprietary and the cert/scale gates are real (Accenture's 5,000 certified people are not a vanity number — they are a procurement moat). The wedge holds only at the mid-market tier where scale stops being the deciding factor and right-sized repeatability starts winning. Above ~$1M engagements, the Big-3 own the field; below it, the method-as-product story is genuinely defensible.
Open follow-ups
- Is Accenture's AI Refinery sold as a runtime/orchestration substrate or as an assessment-to-build methodology? If the latter, it is the one big-SI asset that directly contests CAF's whitespace — needs a dedicated brief.
- What do mid-market-focused Snowflake SIs (Slalom's lower band, Hakkoda, Aimpoint Digital, Analytics8, Snowflake's own Professional Services) pitch at the $150-500k tier specifically? The Big-3 are the wrong comparison set for CAF's actual deal size — the real competitors are the other specialist shops.
- Does any competitor publicly expose an assessment/discovery framework as a named product (vs. a free pre-sales workshop)? If "assessment-as-product" is genuinely unclaimed, that is a naming/GTM opportunity for CAF beyond phData-internal use.
- What is the actual conversion rate from a Cortex HOL/workshop to a paid PoC in mid-market accounts? Directly sizes the KnowBe4-style motion as a repeatable funnel.
- How does Snowflake's own Professional Services position vs. its SI partners on Cortex — partner-enabler or quiet competitor at the mid-market tier?
Related
- [[2026-06-09-caf-restructure-proposal]] — what CAF actually is (assessment-to-build engine spanning sales→delivery seam)
- [[2026-05-20-phdata-cortex-agents-practice]] — phData's Cortex delivery spine, deal shape, and IP
- [[2026-05-21-enterprise-ai-agent-deployment-paths]] — the four deployment archetypes and phData's specialist-consultancy placement
- [[2026-06-26-knowbe4-snowflake-workshop-pitch]] — the live demo that embodies the CAF pattern and the mid-market wedge
- [[use-case-niche-ledger]] — the ledger this brief feeds; positioning vocabulary for scoring use-cases
Sources
Vault:
- [[2026-06-09-caf-restructure-proposal]] —
~/rdco-vault/01-projects/phdata/2026-06-09-caf-restructure-proposal.md - [[2026-05-20-phdata-cortex-agents-practice]] —
~/rdco-vault/06-reference/research/2026-05-20-phdata-cortex-agents-practice.md - [[2026-05-21-enterprise-ai-agent-deployment-paths]] —
~/rdco-vault/06-reference/research/2026-05-21-enterprise-ai-agent-deployment-paths.md - [[2026-06-26-knowbe4-snowflake-workshop-pitch]] —
~/rdco-vault/01-projects/phdata/2026-06-26-knowbe4-snowflake-workshop-pitch.md - [[use-case-niche-ledger]] —
~/rdco-vault/01-projects/phdata/use-case-niche-ledger.md
Web:
- Accenture + Snowflake press release: https://www.snowflake.com/en/news/press-releases/accenture-and-snowflake-drive-enterprise-reinvention-with-ai-and-data/
- Accenture Snowflake partner page: https://www.accenture.com/us-en/services/ecosystem-partners/snowflake
- Deloitte AI Advantage Solution for Snowflake: https://www.deloitte.com/us/en/alliances/snowflake/ai-advantage-solution-snowflake-data.html
- Deloitte + Snowflake gen-AI offerings: https://www.deloitte.com/us/en/alliances/deloitte-snowflake-gen-ai.html
- Slalom Snowflake Summit / offerings: https://www.slalom.com/us/en/who-we-are/events/snowflake-summit
- Slalom on Cortex Code: https://medium.com/slalom-blog/cortex-code-snowflakes-ai-coding-agent-for-modern-data-teams-fdc0619a8d40
- Snowflake "control plane for the agentic enterprise" release: https://www.snowflake.com/en/news/press-releases/snowflake-expands-snowflake-intelligence-and-cortex-code-to-power-the-control-plane-for-the-agentic-enterprise/