# Fin-archetype *The horizontal incumbent that published its benchmarks, re-priced to outcomes, and certified for AI governance. Named for Fin by Intercom; used as a reference class for DD on vertical AI agent challengers.* ## Defining properties The Fin-archetype is a horizontal customer-experience AI agent that has set the category's proof bar across six dimensions simultaneously. Any one of these is copyable in isolation; what makes it an archetype is the bundle. 1. **Published model metrics at scale.** Customer-averaged resolution rate (67% global average, 80–84% top decile), hallucination rate (~0.01%), uptime (99.97%), customer count (7,000+), conversation volume (>1M/week). Published means: in the marketing site, in the docs, defensible under scrutiny. 2. **Outcome-based pricing with a performance guarantee.** $0.99 per resolution, no seat fee, and a $1M refund if the average resolution rate falls below a published threshold. Converts pricing from a negotiation to a commitment. 3. **Published model papers on the custom stack.** `fin-cx-retrieval` (custom retrieval model) and `fin-cx-reranker` (custom reranker model) treated as IP-grade artefacts the company stands behind in public. This is the durable model-IP layer — not the LLM generation step, but what sits around it. 4. **Native multilingual LLM generation across 45+ languages** with no translation layers. Attacks any regional challenger's language moat directly. 5. **Full AI-governance certification stack.** SOC 2 Type II, ISO 27001 / 27701 / 27018, HIPAA, **ISO 42001** (AI management system), **AIUC-1** (quarterly adversarial red-team testing). The last two are the new category-defining certifications. 6. **Self-managed by the customer's CX team.** No engineering required to configure, update, or iterate — the product ships "Procedures" (multi-step workflows with backend integrations) that a CX lead edits same-day. ## Adjacent capabilities that widen the bundle - **Voice channel** as an AI-delivered capability (Fin Voice), not an integration story. Few vertical challengers carry voice. - **Procedures library** covering integrations with the horizontal systems of record (Stripe, Salesforce, Linear, Shopify). A vertical challenger's "playbooks" are the same pattern; the breadth of coverage determines whether the challenger's vertical lock-in outruns the horizontal's horizontal lock-in. ## Attack surface on a vertical challenger The moment the Fin-archetype ships native quality in the challenger's primary **language + channel + workflow** at the Fin-archetype resolution rate, *three liabilities crystallise at once*: - Unpublished accuracy claims look weak against published ones. - Usage-based or seat-based pricing looks expensive at the resolution-rate cross-over point (see [[AI Agent Vertical SaaS DD MOC]] §6 for the cross-over math). - Missing ISO 42001 / AIUC-1 becomes a procurement blocker in AI-governance-literate enterprise buyers. ## Defence for a vertical challenger Five defences that compound: - **Vertical data depth.** Commerce / legal / clinical / field signals the horizontal does not maintain. This is the single most important structural defence. - **Vertical systems-of-record integration depth.** Platform-level access (e.g., storefront, EHR, PLM) with source-of-truth posture, not shallow APIs. - **Faster CX-team iteration inside the vertical's workflow conventions.** If the horizontal's Procedures are general and the challenger's playbooks are vertical-specific, the vertical player ships change faster inside the customer. - **Regional regulatory and data-residency posture.** Matters more in regions with strong data-protection regimes. - **Publish benchmarks first** in the challenger's strongest vertical / language. This is a product act, not a marketing one — it moves the artefact from CLAIMED to SUPPORTED in the investor and enterprise buyer's head simultaneously. See [[Evidence Hierarchy]]. ## DD test > When a horizontal Fin-archetype ships native resolution in the challenger's primary language / channel at 70% average next quarter, what are the three tactical moves the challenger makes in the following 90 days — and which of them does the company already have in flight today? Related: [[Haptik-archetype]] · [[AI Agent Vertical SaaS DD MOC]] · [[Incumbent Bundling Risk]] · [[The Age of Vertical Models]] · [[AI era Defensibility]] · [[Outcome-Based Pricing]] --- Tags: #verticalAI #competitive #AIstrategy #investing #systems