Dolfin
TLDR: An AI-native compensation platform in a crowded field. No claimed moat is evidenced; AI-native is a head start incumbents can copy. A distribution-and-execution bet. Pass today. One fact reopens it: a cross-customer data effect.
What they claim, and what the evidence saysdata: Asserted only
What the company asserts, held against each type's proof bar.
data
Dolfin makes no explicit network-effect claim. The only network-adjacent candidate is a comp-benchmarking data effect, and there is no evidence it exists yet, so it is capped at Asserted only. The company's real barrier, if any, is switching cost (compensation is payroll-critical and deeply integrated), handled in diagnostics. The 'AI-native structural advantage' headline is tested in the erosion section, where it reads as a head start rather than a moat.
- Compensation logic is customer-specific and drawn from each customer's own systems, so it does not obviously aggregate into a cross-customer advantage.
- 'AI-native' is a product posture, not a data moat; a competitor can also be AI-native.
Is the network effect real? Seven tests3 Partial · 3 Weak · 1 None
| Diagnostic | Rating | Barrier test | What would change this |
|---|---|---|---|
| Atomic network unit | No network barrier and no chicken-and-egg for a rival. | A cross-customer layer such as compensation benchmarking or a shared plan-template library that improves with adoption. | |
| Cold-start status | Early traction compounds into references and data, not a barrier. | Evidence that each new customer lowers the cost or raises the value of serving the next beyond ordinary operating leverage. | |
| Density and clustering | Density is irrelevant to the current model. | A benchmarking effect that improves with density in a segment (for example, all SaaS sales teams in one market). | |
| Multi-tenanting exposure | Single-system-of-record dynamics cap multi-tenanting later; not yet proven at this stage. | Evidence customers make Dolfin the sole system of record rather than a parallel trial. | |
| Disintermediation risk | The barrier is whether Dolfin's depth and integration outrun incumbents adding AI. | Win/loss evidence against incumbents once they ship AI features. | |
| Value curve | An asymptotic efficiency gain is matchable once competitors reach similar AI capability. | A cross-customer data effect that keeps improving with scale. | |
| Switching-cost decomposition | Passes if ripping Dolfin out means rebuilding comp logic and integrations mid-cycle, which is costly; unproven at seed. | Net revenue retention above 100% and customers relying on Dolfin to run live payroll-affecting cycles. |
Does the moat survive AI?2 intact-but-thinner · 1 dissolves
Each moat component against LLM-era commoditisation. Every call carries its falsifier.
| Component | Outcome | Flips if… |
|---|---|---|
| The AI-native plan-design and reconfiguration experience (the 'structural advantage') | medium confidence | This flips to dissolves if incumbents ship comparable AI reconfiguration and Dolfin has no other edge; it flips to intact if the comp-modelling depth proves genuinely hard to replicate. |
| CRM/ERP/HRIS integration and payroll-critical embedding (switching costs) | medium confidence | This flips to intact if net retention proves high and customers treat switching as a multi-cycle project. |
| Compensation and performance data compounding across customers (benchmarking) | low confidence | This flips to strengthens if Dolfin demonstrates a benchmarking or plan-template product that measurably improves as more customers contribute data. |
What to ask the founder, and what is missing3 questions · 3 gaps
Three questions for the founder
- Does compensation or performance data from one customer improve the product for others, or is each customer's comp logic fully siloed? Is there a data effect, or only per-customer configuration?
- What is net revenue retention, and are you the sole system of record for compensation, or a layer running alongside spreadsheets and incumbents?
- As Xactly, CaptivateIQ, Spiff and Everstage add AI, what is your durable advantage beyond being AI-native first?
Evidence missing
- Whether any compensation or performance data compounds across customers. This is the line between a durable data effect and a head start that saturates. It is the decisive question for whether 'AI-native' becomes a moat.
- Net revenue retention and system-of-record status. Switching cost is the only real barrier candidate, and only retention plus sole-vendor status show whether it holds.
- Win/loss against incumbents (Xactly, CaptivateIQ, Spiff, Everstage). Sales-comp is a crowded category; beating incumbents as they add AI is the core defensibility test.
What must be true by Series A3 conditions, each with proof and kill
Derived from evidenced types only.
Switching costs are real: Dolfin becomes the system of record for compensation, not a parallel trial.
Proof: Net revenue retention above 100% with customers running live payroll-affecting cycles, cohorts retaining horizontally across successive comp cycles, and CAC payback shortening.
Kill: Customers keep comp in spreadsheets or incumbents alongside Dolfin and churn after a cycle.
The AI-native edge stays ahead of incumbents adding AI.
Proof: Documented wins against Xactly, CaptivateIQ, Spiff and Everstage after they ship AI features, closed on pricing power rather than discounting.
Kill: Incumbents match the reconfiguration speed and win on distribution and integrations.
A cross-customer data effect emerges to turn a head start into a compounding advantage.
Proof: A benchmarking or learning-template product whose accuracy improves with customer count, with engagement cohorts improving vertically.
Kill: Compensation logic stays siloed per customer and no data effect materialises.
What would kill the thesis3 ways this dies
- The thesis breaks if incumbents match the AI-native reconfiguration, since 'AI-native' is a head start rather than a moat.
- The thesis breaks if compensation logic stays siloed per customer and no benchmarking or data effect emerges.
- The thesis breaks if a general LLM over a spreadsheet closes enough of the gap for smaller teams to not need a dedicated platform.
Every source, cited6 items, cited
Written before the report. If it is not cited, it does not exist.
| Supports | Source | Confidence |
|---|---|---|
| Founded 2023, Barcelona, Daniel Seror (CEO) and Antoni Bardina (CPO) | The AI Insider | high |
| €2.1M seed (May 2026), Swanlaab lead, with Archipelago Next, Inveready, Dozen; no Samaipata affiliation | The AI Insider | high |
| Product: AI for RevOps/finance/comp teams to design and manage incentive plans without spreadsheets or consultants; integrates CRM, ERP, HRIS | The AI Insider | high |
| Company-stated: onboarding 6 months to weeks; commission validation days to hours | The AI Insider | medium |
| Serves organisations with over €851M combined annual revenue; SOC 2 certified | The AI Insider | medium |
| Defensibility claim: AI-native from day one framed as a structural advantage over consultant-heavy legacy tools | EU-Startups | medium |
{ "verdict": null }
I would pass today. Not because the team is wrong about the AI-native head start, but because none of the claimed moat is evidenced in public and the one differentiator they lead with, being AI-native, is the easiest part for a funded incumbent to copy. Switching cost is the only real barrier here, and at seed there is no retention data to show it exists yet. One fact reopens this: a demonstrated cross-customer data effect, the durable version of the moat, which does not exist today. If Dolfin ships benchmarking that measurably improves with scale, I want the next meeting. Until then, pass.