El Registro
Dated, falsifiable predictions about real companies. The point is not any single call. It is the loop. The system periodically rescans its own predictions against what actually happened, and every resolved one recalibrates the rubric the next case runs on. The judgment gets sharper the longer it runs.
Predictions
Each is a closed toggle. Open one for the reasoning and the criteria that will settle it.
Dolfin (AI-native sales-compensation SaaS)open · resolves 2027-12-31 · conviction
By 2027-12-31, at least two of Xactly, CaptivateIQ, Spiff and Everstage will have shipped AI-driven compensation-plan design or reconfiguration comparable to Dolfin's "AI-native" pitch.
El Foso rated the AI-native reconfiguration advantage "intact-but-thinner": a head start, not a moat. Incumbents have distribution and every incentive to close the gap. If they do, Dolfin's headline differentiator commoditizes.
- Resolution criteria
- CORRECT if, by the date, at least two of the four named incumbents have a public product announcement, changelog or documented feature describing AI compensation-plan design/reconfiguration. INCORRECT otherwise.
- Made on
- 2026-07-03
Dolfinopen · resolves 2028-06-30 · lean
By 2028-06-30, Dolfin will not have publicly demonstrated a cross-customer data effect (a benchmarking or template product whose value measurably improves as more customers join).
El Foso found zero evidence of a data effect; compensation logic is siloed per customer. The durable version of the moat requires exactly this, and it does not yet exist. Predicting its continued absence is the honest read.
- Resolution criteria
- INCORRECT if Dolfin ships or publicly documents a benchmarking/cross-customer product that improves with scale. CORRECT if no such capability is public by the date.
- Made on
- 2026-07-03
Dolfinopen · resolves 2028-06-30 · lean
If Dolfin raises a Series A on or before 2028-06-30, the diligence-legible basis will be switching-cost / system-of-record evidence (net revenue retention, payroll-critical embedding), not a demonstrated data network effect.
El Foso identified switching cost as the only surviving barrier candidate. A round that clears would most plausibly be underwritten on retention and system-of-record status.
- Resolution criteria
- VOID if no Series A is announced by the date. Otherwise CORRECT if the publicly reported basis emphasises retention/system-of-record, INCORRECT if it centres on a demonstrated cross-customer data network effect.
- Made on
- 2026-07-03
Kabilio (Spanish accounting / invoice AI)open · resolves 2028-06-30 · conviction
By 2028-06-30, no horizontal document-AI will have commoditized Verifactu-compliant Spanish invoice and ledger processing to the point that a Spanish accounting firm would drop a specialist tool like Kabilio for a generic one.
El Foso rated Spanish regulatory compliance plus bank-integration depth as intact against LLM commoditization; three of four moat components survive because the barrier is regulatory complexity and access-constrained local data, which AI does not erode.
- Resolution criteria
- INCORRECT if a horizontal/general document-AI ships turnkey Verifactu-compliant Spanish processing that documented firms adopt in place of specialist tools. CORRECT if the specialist regulatory layer still holds by the date.
- Made on
- 2026-07-03
Kabilioopen · resolves 2027-12-31 · lean
By 2027-12-31, at least one incumbent Spanish accounting platform (Sage, A3 / Wolters Kluwer, or Holded) will ship native Verifactu-compliant AI invoice processing.
El Foso named incumbent absorption, not startups, as the real disintermediation risk. Incumbents own distribution and the compliance obligation is universal. Note this can be true while R-4 is also true: shipping AI is not the same as overcoming switching costs.
- Resolution criteria
- CORRECT if at least one named incumbent publicly releases Verifactu-compliant AI invoice processing by the date. INCORRECT otherwise.
- Made on
- 2026-07-03
Kabilioopen · resolves 2028-12-31 · lean
By 2028-12-31, Kabilio will not have entered a second regulated geography outside Spain without substantially rebuilding its compliance and bank-integration layer.
El Foso flagged that the same Spanish regulatory specificity that defends the moat may make it non-portable, capping a fragmented low-ACV market. Portability without a rebuild is the bull case; its absence is the base case.
- Resolution criteria
- VOID if Kabilio announces no second-country launch by the date. Otherwise CORRECT if the expansion required a substantially rebuilt compliance/integration stack, INCORRECT if the existing stack ported with only minor adaptation.
- Made on
- 2026-07-03
European AI-application-layer seed companies (market-level)open · resolves 2028-12-31 · conviction
Among European AI-application-layer seed companies that publicly claim a "data network effect", the majority that raise a Series A before 2028-12-31 will do so on switching-cost, regulatory, or proprietary-access moats rather than on a demonstrated cross-customer data network effect.
Both El Foso cases rated the app-layer "data network effect" claim as narrative (asserted or contradicted), while the durable barriers were switching costs and regulated/local-data access. Generalising: at the app layer, the graduating moat is usually not the claimed data flywheel. This is a direct, falsifiable answer to the fund's own open question.
- Resolution criteria
- Against a pre-registered sample of European app-layer seed companies with a public "data network effect" claim (list frozen in this repo at resolution), CORRECT if, among those raising a Series A by the date, most disclose a basis other than a demonstrated data network effect; INCORRECT if most graduate on a demonstrated data network effect.
- Made on
- 2026-07-03
Predictions derive from the case analyses and from one market-level question. Each carries adjudication criteria with no wiggle room. Two conviction levels, no probability theatre.