BrainGraph

@BrainGraph search for all facts about our Q4 product roadmap decisions and who made them

BrainGraph returned 12 facts spanning three documents ingested in November. Key findings: the roadmap pivot to mobile-first was proposed by Sarah Chen on Nov 8; engineering capacity constraints were flagged by the platform team in a Nov 12 standup transcript; final sign-off came from the exec team on Nov 19. Each fact links back to its source document and includes confidence scores. Two entities were flagged as ambiguous: 'mobile strategy' appears in both the product brief and a separate marketing doc with different scopes.

This example shows BrainGraph's core strength: semantic search across ingested documents with entity resolution and provenance tracking. The quality of results depends on how much content you've fed into your brain. Search is read-only and does not consume quota, but ingesting the underlying documents does.

@BrainGraph ingest this YouTube video https://youtube.com/watch?v=example123 into our team brain so we can reference the keynote later

BrainGraph queued the video for ingestion. Processing ID: doc_8x3k9m. The system will extract metadata, generate a transcript, identify speakers, and analyze entities. Use 'Get Document Progress' to poll status—expect 5-15 minutes for a 45-minute video. Once transcript.status reaches 'analyzed', you can search facts from this keynote alongside your other documents. This ingestion counts against your monthly upload quota.

This demonstrates BrainGraph's ability to turn unstructured video into searchable knowledge. The MCP confirms the job was queued but doesn't wait for completion—you must poll separately. Be aware: this creates persistent data in your account and consumes quota. Failed ingestions (private videos, geo-restrictions) still count.

@BrainGraph find recurring relationship patterns in our brain where the same three entities appear together at least 5 times, then summarize what those clusters tell us about our workflow

BrainGraph identified 4 recurring patterns meeting the threshold. Most frequent: [Sarah Chen] → proposed → [Feature X] → blocked_by → [API Team] appears 8 times across sprint retros and planning docs. This suggests a persistent dependency bottleneck. The second pattern links [Budget Approval] → required_for → [Vendor Contract] → signed_by → [Finance] in 6 documents, indicating procurement follows a consistent approval chain. The AI notes these patterns reveal both a technical constraint (API team capacity) and a process strength (finance governance is well-documented).

This example pairs BrainGraph's graph-pattern detection with the AI's synthesis capability. The MCP returns raw relationship chains; the AI interprets what they mean for your team. Pattern detection is computationally expensive—expect slower responses on large brains. Minimum occurrence count filters noise but may miss emerging trends.

When BrainGraph wins for ongoing user-interview analysis

A 5-person product team runs 8-12 user interviews per sprint and needs to spot recurring pain points across transcripts without re-reading everything. BrainGraph's entity resolution and graph-pattern tools let you ask "what complaints appear in 3+ interviews this month" and get canonical answers—no manual tagging. The YouTube and web-page ingest tools mean you can drop in recorded calls or support-forum threads alongside interview notes. This works if your interview volume stays under 50 documents per month and you're comfortable waiting 2-5 minutes for transcript analysis to complete. If you're doing 100+ interviews quarterly or need real-time synthesis during the call, the processing lag becomes a blocker. For teams that review research async and want to query across sources, BrainGraph turns interview debt into a searchable knowledge graph.

BrainGraph as a living competitor-mention index

A 3-person go-to-market team monitors 20 competitor blogs, earnings transcripts, and product-launch videos to brief sales on positioning shifts. BrainGraph's web-page and YouTube ingest let you feed those sources into a single brain, then use Search Brain to ask "what new features did Competitor X announce this quarter" without opening 40 tabs. The graph-pattern tool surfaces which competitors are mentioned together in the same documents—useful for understanding who you're actually compared to in the wild. This setup works if you're ingesting 10-30 URLs per week and can tolerate a 24-hour lag between publish and query. If you need sub-hour freshness or are tracking 100+ sources, the manual-ingest workflow and quota limits become friction. For small GTM teams that review competitive intel weekly, BrainGraph replaces the sprawling Notion doc with a queryable brain.

When BrainGraph turns Loom libraries into searchable onboarding

A 6-person remote startup records 30-40 Loom walkthroughs covering deploy scripts, customer-success playbooks, and design-system conventions. New hires watch videos in random order and Slack the same questions because there's no index. BrainGraph's YouTube-ingest tool (Loom videos work if they're public or unlisted) builds a transcript corpus you can query with "how do we handle refund requests" and get the relevant 90-second clip reference. The entity-resolution feature means "Stripe" and "payment processor" collapse to the same concept, so search works even when terminology varies across videos. This pays off if you have 20+ onboarding videos and hire 2-3 people per quarter. If your video library is under 10 clips or you're hiring one person per year, a YouTube playlist with timestamps is simpler. For teams that record everything but can't find anything, BrainGraph makes the Loom library a living onboarding brain.