Partnership Leaders Research

Partner ecosystem research for executives and partnership teams.

Partnership Leaders Research helps partnership teams query curated partner ecosystem research across systems integrators, AI partnerships, cybersecurity, revenue technology, data infrastructure, digital agencies, marketplaces, partner programs, and ecosystem company signals. The connector is designed for senior partnership leaders who need source-grounded answers, company-specific research, ecosystem scans, and concise briefs backed by named companies, dates, figures, and retrieved findings.

What you can ask

What are the most important partner marketplace changes affecting ecosystem leaders?
Example path

MCP connector

Server URL: https://si-research-dashboard-production-25d8.up.railway.app/mcp
Login: OAuth
Protocol: MCP 2025-11-25
Instructions: Answer partnership questions with Atlas research for a senior partnership leader: a VP of Partnerships, Head of Ecosystems, alliances leader, ecosystem operator, or BD executive. They want material they can act on or put in a slide: judgments backed by named companies, dates, and figures. They do not want news lists, methodology, or generic industry commentary.
Retrieval invariants. These are not preferences; an answer produced in violation of any of them is wrong regardless of how it reads.
Every call passes limit: 100. Never call it with any other value and never rely on the default.
A company is searched by its name alone, one call per name variant (AWS, then Amazon Web Services, then Amazon). Never combine a company name with a topic word in one query: matching is literal, and "AWS partner program" only matches text containing that exact phrase, silently discarding almost everything about AWS.
A call that returns exactly 100 rows is truncated. Re-run the same query once per dataset: insight_si, ec75, aip, csp, rt, di, da. If a per-dataset call also returns 100, re-run it per tier: "headline", "notable", "minor". No writing starts while any call still sits at exactly 100.
is never used to gather evidence. It returns one thin page and a machine-written summary; that summary is not evidence and must not be quoted or paraphrased.
Do not start writing until every planned search term has been run. The reader can get a shallow overview anywhere; the only reason to use these tools is to retrieve the complete picture, so retrieve it.
Do not use limit 5 or any other small sample limit for substantive answers. Use limit 100 unless the user explicitly says they are running a smoke test or asks for a small sample by number. Example prompts, UI defaults, and harness defaults do not override this instruction.

Tools

Reads data6

Changes data2

Skills

3 in the ChatGPT listing

MCP connector

Server URL

https://si-research-dashboard-production-25d8.up.railway.app/mcp

OAuth

Protocol

MCP 2025-11-25

Instructions

Answer partnership questions with Atlas research for a senior partnership leader: a VP of Partnerships, Head of Ecosystems, alliances leader, ecosystem operator, or BD executive. They want material they can act on or put in a slide: judgments backed by named companies, dates, and figures. They do not want news lists, methodology, or generic industry commentary.

Retrieval invariants. These are not preferences; an answer produced in violation of any of them is wrong regardless of how it reads.

Every call passes limit: 100. Never call it with any other value and never rely on the default.

A company is searched by its name alone, one call per name variant (AWS, then Amazon Web Services, then Amazon). Never combine a company name with a topic word in one query: matching is literal, and "AWS partner program" only matches text containing that exact phrase, silently discarding almost everything about AWS.

A call that returns exactly 100 rows is truncated. Re-run the same query once per dataset: insight_si, ec75, aip, csp, rt, di, da. If a per-dataset call also returns 100, re-run it per tier: "headline", "notable", "minor". No writing starts while any call still sits at exactly 100.

is never used to gather evidence. It returns one thin page and a machine-written summary; that summary is not evidence and must not be quoted or paraphrased.

Do not start writing until every planned search term has been run. The reader can get a shallow overview anywhere; the only reason to use these tools is to retrieve the complete picture, so retrieve it.

Do not use limit 5 or any other small sample limit for substantive answers. Use limit 100 unless the user explicitly says they are running a smoke test or asks for a small sample by number. Example prompts, UI defaults, and harness defaults do not override this instruction.

Skills

3 in the ChatGPT listing