Event Magi ranks each delegate's counterparts by fit — who is looking for what someone else offers, not who looks similar — so the introductions actually convert. And you can show every sponsor exactly which meetings you delivered.
Every introduction is tracked from suggested to “we met” — so after the event you can show, meeting by meeting, exactly what the room produced.
The engine ranks a delegate's counterparts by what each side is seeking versus offering — a founder raising meets the fund with that thesis, not five lookalike founders.
Scoring is deterministic and pure (seeking↔offering complement weighted highest, then sector/stage/thesis/geography). Same inputs return the same ranking, so an agent can cache, diff, and reason about results.
Each match carries one plain-language line — "offers the enterprise distribution you're seeking" — so a delegate acts on it without reading a manual.
Every match exposes a structured reason plus the contributing factors, so an agent can surface, re-rank, or explain a match to its own principal without a second model call.
Show your board and your sponsors exactly which introductions happened and which converted — meeting by meeting, not a vanity attendance number.
Every match and introduction is tracked from suggested → met → converted, and that record is machine-checkable, so an evaluating agent can verify what happened instead of trusting a black box.
The event is the delivery surface; the durable product is the graph of people, intent, introductions, and outcomes that survives after everyone goes home.
Data is graph-shaped from day one (people, intent, introductions, outcomes). The relationship graph is queryable year-round, not a per-event export that goes stale.
For each delegate, the engine projects the whole room over stated intent and reciprocal fit, then ranks the counterparts who actually move their goal forward. Every match carries one plain-language line so a delegate acts without reading a manual.
Ranking is deterministic and pure — the same inputs return the same order — so an agent can cache, diff, and explain a match to its own principal without a second model call.
A match is only worth something if the meeting happens. Consented, double-opt-in introductions are tracked from interested to met, so nobody gets spammed and every accepted intro becomes part of the record the organizer can point to.
Delegates ask in plain language — “who should I meet before lunch?” — and the concierge finds matches and opens consented introductions through guarded tools. The same surface an AI agent drives on a delegate's behalf.
The concierge runs on the same MCP tools exposed publicly — find_matches, request_introduction, get_receipt — so an external agent gets exactly the capability a delegate does, behind the same policy boundary.
Relationship movement, introductions requested, introductions that converted — the analytics turn a two-day event into a record you can put in front of a board or a sponsor. Not booth scans, not vanity traffic: meetings that happened.
“I have two days and 300 strangers. Tell me the five who matter and get me in front of them.”
A ranked match feed with believable reasons and one-tap, consented introductions.
“Make my event feel like everyone met the right people, and let me prove ROI to sponsors.”
A receipted record of who was matched, why, and which introductions closed.
“Did my money buy me the meetings I wanted?”
Relationship movement against target accounts, not booth scans.
“Do the profiling, introductions, and follow-up a human chief-of-staff would.”
Guarded tools that find matches and request introductions, every action receipted.
Event Magi is built for an AI buyer. Discover it by vertical, evaluate it by capability, and integrate it through MCP, OpenAPI, or the CLI — then read receipts to verify what it did.
find_matchesPOST /api/matchesReturn the top-N complementarity matches for a delegate, each with a one-line reason and a tracked outcome id.
request_introductionPOST /api/introductionsOpen a consented, double-opt-in introduction between two delegates and track it to outcome.
get_receiptGET /api/receipts/:idRead the replayable outcome record for any match, introduction, or outcome.
# PROPOSED — endpoints are stubbed during the design-partner phase.
# 1. Point your MCP client at the Event Magi server:
# https://eventmagi.com/mcp (auth: Bearer <token>)
# 2. Call the matchmaking tool:
curl -s https://eventmagi.com/api/matches \
-H "Authorization: Bearer $EVENTMAGI_TOKEN" \
-H "Content-Type: application/json" \
-d '{ "eventId": "evt_...", "delegateId": "dlg_...", "limit": 5 }'
# -> returns ranked matches, each with a one-line reason and an rcp_ receipt id.
Install via CLI: npx @mogos/eventmagi@latest matches --event <eventId> --delegate <delegateId>
| Capability | Event Magi | Directory & networking apps | Matchmaking incumbents |
|---|---|---|---|
| Complementarity (seek↔offer), not lookalike similarity | Yes | — | partial |
| One-line reason a delegate believes | Yes | — | inconsistent |
| Receipted, replayable match provenance (rcp_) | Yes | — | — |
| Mutual-value (both-win) scoring | Yes | — | partial |
| Agent-native: MCP + OpenAPI + CLI from day one | Yes | — | partial |
| Relationship graph persists year-round | Yes | partial | Yes |
Reflects MOGOS's reading of public competitor positioning as of 2026-07 (Grip, Swapcard, Brella, b2match, Converve). Note: Grip already ships an MCP endpoint and an action-taking AI assistant — agent-nativeness is contested, but receipted match provenance is not yet offered by any of them.
A serious event ready to run the matching loop with the team and shape the roadmap.
Organizers running a conference, summit, or buyer-supplier programme.
Platforms and agents integrating matchmaking via API/MCP at scale.
Event Magi is already matching real people at a live Davos event (catalyst.mogos.ch). This is a running deployment, not a demo — the match feed, concierge, and introduction loop you see here are the product in use.
Who was suggested, why, and whether the meeting happened and converted — all of it is kept as an auditable record you can replay after the event. You prove the room worked instead of asking a board to take your word for it.
The matching engine, the shared Magi shell, and the analytics are engineered and owned by the MOGOS Collective — a team that ships the substrate under the product, not a thin wrapper over someone else's API.
Event Magi is an AI event matchmaking platform. For every confirmed delegate it surfaces the five people in the room they most need to meet, explains each match in one plain-language line, and tracks the introduction from interested to we met. It is built by the MOGOS Collective.
A directory shows you everyone and leaves the work to you. Event Magi ranks by complementarity — who is seeking what someone else is offering — gives a believable reason for each match, and emits a receipt proving who was suggested, why, and what happened. The matching is the product; the event is just where it is delivered.
Event Magi keeps an auditable record of every introduction — who was suggested, why, and whether the meeting happened and converted. After the event you can show, meeting by meeting, what the room produced against the accounts a sponsor cared about: relationship movement, not booth scans. This meeting-by-meeting proof is the part directory and networking apps do not offer.
Yes. Event Magi is agent-native: it exposes a Model Context Protocol (MCP) endpoint, an OpenAPI schema, and a CLI. An agent can find matches, request consented introductions, and read outcome receipts on behalf of a delegate. A machine-readable product manifest at /.well-known/magi-product.json lets a visiting agent self-onboard.
Sign in with LinkedIn, and public enrichment pre-fills roughly 80% of the profile before any form. The delegate answers three to five questions that actually move a match — what they are seeking and offering — and confirms. The target is a usable, match-ready profile in under a minute.
Event Magi is the intellectual property of the MOGOS Collective, designed and built on MOGOS-owned packages. Davos Catalyst is a branded single-tenant deployment (catalyst.mogos.ch) where Event Magi runs in production; it is a customer skin, not a co-creator. The unbranded product lives at eventmagi.com.
Pricing is per event for organizers, with a metered option for programmatic and agent-driven use. Early-access design partners work directly with the team. See the pricing section for current tiers.
Event Magi is live and matching real delegates in production today at Davos Catalyst. We are onboarding a small number of design-partner events now, working directly with each organizer to fit their programme. The organizer console and sponsor-ROI reporting are rolling out this quarter.