Ask the question no report predicted.
For Medical Affairs, insights and competitive intelligence teams. The agents you already use (ChatGPT, Claude, Grok or your own) search what clinicians say in public about clinical trials, in podcasts, video and X, with no new app to log into. Every answer comes back with the passage, the speaker, the date and a link.
Where a report stops, your agent keeps going
A report answers the questions it was scoped for. The work that follows asks new ones every week.
The question no report predicted
The clinician you meet next week
The trials you watch every week
What teams ask their agents
Each use case shows the prompts, the tool calls they trigger, and the job done on a published report.
- Meeting prepMCP
“I'm meeting five oncologists next week. For each, summarize what they have said publicly about NEONIPIGA and neoadjuvant immunotherapy in the last 12 months, with sources.”
Walk into a scientific exchange knowing what the clinician has already said in public.
- Readout reactionReports + MCP
“What are clinicians questioning after this week's data? Separate interpretation from what the primary results establish.”
See how clinicians read a trial result in the days and weeks after it lands.
- Congress briefingReports + MCP
“Brief me on the HCP reaction to the late-breaking abstracts in our therapeutic area at this congress, with sources.”
A sourced briefing on how a congress presentation was received.
- Launch and access signalsReports + MCP
“Ahead of approval, what practical barriers do clinicians raise about using this regimen, separate from their view of the efficacy data?”
Hear the practical barriers clinicians raise before and after an approval.
- Trial watchlistReports + MCP
“For the trials on my watchlist, tell me which reports changed this week and summarize only what is new.”
Follow your own and competitor trials, and see only what changed.
- All 9 use cases, including the ones a report answers on its own →
59 example prompts for your agent, each with the tool calls that answer it.
What you have today, and what Erudio adds
Most teams already pay for some of these. Erudio fills the gap between them: the clinician's own reasoning about a trial, in public, with the source.
| You have | It gives you | Erudio adds |
|---|---|---|
| KOL and HCP databases | Tell you who a clinician is: affiliations, publications, trials, influence scores. | Shows what they said about a trial, where and when, in their own words. Use both. |
| Social listening | Counts mentions and sentiment, mostly from short posts. | Reads long-form podcasts and video as well as X, and separates a clinical view from a reaction. |
| Field medical insight platforms | Hold what your MSLs heard in private conversations. | Covers what clinicians say in public. It complements your field insights and never ingests them. |
| Analysts and agencies | Answer one question per project, in weeks. | Answers the next question in the same session, with every passage linked for review. |
Ten tools, three tiers
Discovery tools describe what exists. Licensed tools read the reports you hold. Enterprise tools query the full commentary corpus.
| Tool | Tier | What it does |
|---|---|---|
| search_reports | Discovery | Search report catalog records by trial, NCT ID, drug, indication, sponsor, phase or update time. |
| get_report_card | Discovery | One report's catalog record: scope, counts, version, limits and preview. |
| list_reports | Licensed | List the reports your organization holds, with trial, edition, aspect coverage and counts. |
| read_report | Licensed | Read a report in full or by section: findings, aspects, verified claims, sources and people, with stable IDs. |
| get_report_changes | Licensed | What changed in a report since the edition you hold, with a recommendation: read the changelog, the affected sections, or the full report. |
| search | Enterprise | Hybrid semantic and keyword search across HCP podcasts, YouTube, X videos and X threads, with date, source and author filters. |
| fetch | Enterprise | Retrieve the matching passages for up to 20 results at once, with speaker, date and a link to the original. |
| fetch_x_thread_sources | Enterprise | The exact posts in an X thread, with stable author IDs, replies and quotes, and links to each post. |
| list_sources | Enterprise | Discover the podcasts, channels and X authors in the corpus. |
| lookup_profile | Enterprise | Find clinicians by NPI, X account, Erudio ID or name, up to 100 at once, including podcast and video speakers not yet linked to an NPI. Each profile says whether it is NPI-linked, X-linked or media-only and how it matched, and returns NPI credential, specialties, practice locations and practice and facility affiliations, profile facets, verified ORCID, OpenAlex, Google Scholar and NIH IDs, research institutions, podcast and video appearances, and the self-described X bio. Every block names its source. |
Public HCP commentary, indexed continuously
Podcasts
YouTube
X threads
X videos
Built for medical, legal and compliance review
Written for the people who review a vendor before your team can use it.
Public sources only
Passages, never transcripts
Identity you can check
Scientific use
Counts are not consensus
Gaps stated
Built for vendor review
Security-only SOC 2 Type II
Encryption
Audit logging
What an Enterprise license includes
Quoted per organization, by therapeutic area and team size.
An organization license
Your agents, your team
Every report you license
The full commentary corpus
Usage you can plan
Administration
One line in your client config
{
"mcpServers": {
"erudio": { "url": "https://mcp.erudio.com/mcp" }
}
}The server speaks the Model Context Protocol over streamable HTTP and signs in with OAuth 2.1. It works with ChatGPT, Claude, Grok and any MCP-compatible client.
Access is set per organization. Your team signs in with its own accounts; we never ask for shared credentials.
Agents without MCP can use the JSON and Markdown endpoints.
Start with your own questions
Pick one therapeutic area and bring three questions your current reports and vendors could not answer. Your team runs them through its own agent, and you judge the answers against the sources.