Demo · structured lookup, not chat

Pick the right partner before the second customer meeting.

Start with the opportunity, not a blank search box. The seller picks or arrives from an MSX opp, the engine ranks partner fit, and every score explains itself. First choice is to surface this inside MSX as an interactive card. If that is too hard for the pilot, launch a standalone partner-led surface from MSX with the opportunity context passed in.

Eight explainable signals: workload fit, project experience, industry, capability, priority tier, prior relationship, segment, and region. Each one shows why it scored.
Weights are dynamic: the signal weighting adapts to the account, the use case, and the person asking. Drag the sliders and watch the ranking move.
Seller action layer: ranked shortlist, evidence, counterfactuals, PDM contact, and MSX write-back actions.

1Pick an opportunity

Click one of the opportunity cards below to run the recommender. In production this would be the current MSX opportunity; in the demo, these five curated opportunities let you test different deal shapes.

2Ranked partner shortlist

After you pick an opportunity, this area fills with ranked partner recommendations, evidence, and actions.

Waiting for opportunity No opportunity selected yet. Click an opportunity card above to pass the deal context into the engine.

Partner recommendations will appear here

Show partners filtered out by hard gates

Pick an opportunity above. The engine will rank partners, show why each score was given, and expose the source fields behind the recommendation.

Why this matters

A seller asking "who do I bring to this deal?" today chases the Priority List, PMX, Partner Finder, Offering cards, and their PDM's inbox. The answer takes hours and is rarely complete. This assembles it in two seconds and shows why.

How the engine scores

Not a black box. Hard gates first remove partners that fail on activity, geography, or solution-area fit. Then eight signals rank the rest:

Workload fit - does the partner do the workloads on this deal? (Offering Cards + Marketplace offers)
Project Experience fit - have they delivered this kind of work? (PMX delivery, recency-weighted)
Industry fit - delivered for similar customers? (PMX projects by industry, not self-asserted)
Capability fit - Microsoft-validated specializations and designations
Solution Area Priority tier - position on the curated agentic priority list
Prior relationship - already in motion at this customer? (PMX + PI ACR by TPID)
Segment fit - partner size vs customer need. A counter-lever against big-SI bias.
Regional fit - actual in-country delivery, not a country on a profile

The weights are dynamic

The signals do not carry fixed weights. The weighting adapts to the account (enterprise vs SMB), the use case (an agentic deal turns on the priority tier; a security deal drops it and re-normalises), and the person asking (a PDM tuning for relationship, a seller for delivery proof). The sliders in the demo are the live proof, not a mock-up.

The path to production

This demo runs against a frozen snapshot under Johan's identity. It is not maintainable as is. The next version needs:

A decision on surface: MSX card first, or standalone surface launched from MSX.
Live data refresh under each seller's own token.
Outcome telemetry to prove whether recommendations lead to better partner motion.
MSX won-deal history feeding the workload signals, at the workload-family grain it can support.

See the roadmap →

Known objections

We did the pre-mortem already. Read the FAQ if you'd rather poke holes than be sold to.

Read the FAQ →

How this could sit inside MSX

The target is not another portal. The cleanest production shape is an interactive card on the MSX Opportunity Partners tab, next to the partner engagement and partner activity surfaces sellers already use.

Summary Revenue Partners Activities Files

Partner engagements

No ranked partner recommendation selected yet.

Primary partnerNot set

Partner motionImplementation partner

If MSX native integration is too hard for a pilot, this card can deep-link to the standalone recommender with the opportunity context pre-filled.

The product is the recommendation layer.

The UI can be MSX, a card, a command, or a temporary standalone surface. The durable value is the engine that turns fragmented GTM and delivery signals into a ranked, explainable partner recommendation.

1 · Surface

Interactive MSX card

Best outcome: a card or command on the Opportunity Partners tab. The seller does not leave the deal. The card shows the shortlist, evidence, and action buttons.

2 · Fallback

Standalone, launched from MSX

If native MSX integration is too heavy, use a standalone surface for the pilot. It must receive opp context from MSX and keep an explicit path back to write-back.

3 · Engine

Where the IP sits

Hard gates remove bad fits. Eight explainable signals rank the rest. Source foldouts explain each score. Counterfactuals show what would change the result. Telemetry tunes the weights later.

4 · Adapts

Weights tuned to context

The same eight signals, reweighted per account, use case, and person. Signals that do not apply drop out and re-normalise. The source behind every signal stays visible, so the rank is always traceable.

EMEA partners in the demo snapshot

Anchor partners with curated Offering cards

Explainable signals, dynamically weighted

Preset scenarios across CAIP, AIBS, Security

What we're asking for

A surface decision

Surface this as an interactive card inside MSX if possible. If that is too heavy for a pilot, approve a standalone partner-led surface launched from MSX with an explicit migration path.

A pilot

30 to 50 EMEA sellers, scoped pilot window. Capture outcome telemetry so the scoring layer can be tuned against real seller decisions.

A pre-wire

Introduction to the MSX Partner experience owner, SI Finder Agent owner, and AI GTM PMMs for CAIP and AIBS. Pre-empt the duplication-versus-augmentation question.