Stage: warm prospect after a 2026-06-26 demo call. No commitment; Sophie will sync internally first and the right point of contact for a follow-up is their PMM, who was out at the time of the call and “back soon.”
Who they are: EU-focused business-banking fintech. Released their MCP server the week before the call (~2026-06-19), rebuilt from a beta stefano-amorelli had running since summer 2025, with a proper security layer added for launch (authentication with credentials, tracking who made calls, call logging).
MCP posture: 52 tools — on the high end vs other MCP servers. Primary goal is serving existing users first; Sophie said they do want to compete versus other MCPs and at some point use it as a discoverability/acquisition channel, but not yet.
Their telemetry (self-reported): no prompt logging, but tool-level visibility. Most-called tool is list-transactions; file-upload tools added the week of the call immediately became among the most used. Claude dominates client share by a large margin (Claude Desktop largest, Claude Code second — a positive surprise to them); ChatGPT share is very small.
[idea] Qonto’s own read (Stefano, 2026-06-26): users are mostly using the MCP to build financial dashboards — inferred from list-transactions dominance, not measured intent.
Our data shown on the call: Qonto ranks ~8th in the broad Finance & Accounting Ops category (30-day view); 9% share on a sample corporate-card-review prompt. [idea] Elliot’s read: they’d likely rank top-5 in a narrower V2 category, given they only just launched.
Their two substantive product objections: (1) geographic relevance — Brex ranked first in a query despite not operating in Europe; our runs are US-based, geo-targeted runs (France/Italy) offered as the answer; (2) memory/personalization — real users have highly customized Claude contexts, our incognito approach removes that; memory profiles (target personas, incl. connected-competitor setups) are on our roadmap for enterprise.
Stakeholder fit: Sophie says they sit at the intersection of product and marketing; decision needs marketing’s buy-in on how much they want to leverage MCP discoverability. We pitched the Lucidchart model (biweekly sync with product marketing + lead engineer).
Pricing quoted: $149/mo core tier; enterprise custom; 50% private-beta discount in exchange for feedback + case-study participation.
Open questions
Who is Qonto’s PMM (name/contact) and when is she back?
How much does Qonto marketing want to leverage MCP discoverability vs. Stefano/Sophie’s product-side interest?
Does Brex still outrank Qonto on EU-geo runs (France/Italy)? Untested at call time.
Timeline
[2026-06-26] (call) Demo call: elliot-garreffa + vincent-mcleese with sophie-cornay and stefano-amorelli. Qonto shared MCP context (52 tools, launched ~Jun 19, Claude-dominant usage, tool-level-only telemetry). Elliot walked through the platform: ~30k prompts run in Claude, four score components (Reachability, Relevance, Recall, Rank), Qonto ~8th in Finance & Accounting Ops, DirectBooker 42%→63% case study, prompt types (discovery/learning/action), incognito runs. Stefano raised geo relevance (Brex in EU queries); Sophie raised memory/personalization and stakeholder fit; pricing quoted ($149/mo, 50% beta discount for case-study partners). Sophie’s takeaway: interesting but very early internally. Commitments: Elliot to send updated deck with slides tailored for their PMM/marketing team; re-engage when the PMM returns from leave.