Answer sales questions from scattered build docs and catalog records
Sales opens Sanity, scans Drive folders, checks Linear statuses, and scrolls Slack threads to recall which build used Apollo or which PRD covered dispatch routing, losing five minutes per question and missing adjacent examples.
Sales asks "show me go-to-market agents for HVAC" and receives three ranked matches with tool tags, workflow patterns, variant notes, and source PRD links so they cite the exact build while the prospect is on the line.
Your team has built dozens of agents, but when a prospect asks for a similar example, you search Sanity records, Drive PRD folders, and Linear tickets by hand.
You've delivered roofing quote agents, HVAC dispatch routing, finance reconciliation, and go-to-market lead intake, but when a prospect asks "Do you have an example for our industry?
Sales opens Sanity, scans Drive folders, checks Linear statuses, and scrolls Slack threads to recall which build used Apollo or which PRD covered dispatch routing.
You can't surface variant notes or workflow patterns that would strengthen the pitch or help the prospect see the fit.
Prior builds live across Sanity records, Google Drive PRDs, Linear tickets, Slack notes, and project docs with no unified retrieval layer.
Semantic search and taxonomy filters return three matching builds with tool tags, workflow patterns, and source PRD links so you cite the exact example while the prospect is on the line.
Each match shows why it fits your query, which tools and workflow patterns align, similar-build links, and variant notes.
Sales cites the exact build on the call with links to the source PRD in Google Drive and Linear ticket status.
Unanswered queries log as gap notes for the catalog pipeline so the team knows which examples to prioritize next.
This agent embeds Sanity catalog records, Google Drive PRDs, Linear ticket metadata, and taxonomy tags, then runs semantic search filtered by industry, tool, function, and workflow pattern. You ask "show me go-to-market agents for HVAC" and receive three ranked matches showing why each fits your query, which tools and workflow patterns align, similar-build links, variant notes, and direct links to the source PRD. Sales cites the exact build on the call, delivery teams find reusable examples for scoping, and member-success teams recommend adjacent builds without searching scattered docs. Unanswered queries capture taxonomy gaps so the catalog pipeline knows which examples to prioritize next.
How the agent retrieves prior builds
Workflow: from scattered docs to ranked build matches
The agent reads published and draft Sanity AI agent demo records with hidden internal metadata, links each record to its source PRD in Google Drive, and pulls Linear ticket statuses, tool tags, industry tags, function tags, audience, and workflow pattern.
Public-safe summaries cover problem, inputs, transformations, outputs, and tools. Internal-only summaries include member names, unpublished build details, variant notes, and exact source PRD excerpts. Both are embedded separately for semantic search.
Internal users type queries such as "show me go-to-market agents for HVAC" or "what builds use Apollo and HubSpot" into Slack or the internal dashboard chat.
The agent applies semantic search across embedded summaries, then filters by industry, function, tool, member tag, publish status, and workflow pattern to return the three best matches.
Each match shows why it fits the query, which tools and workflow patterns align, similar-build links, variant notes, and direct links to the source PRD so the team can cite the exact build or scope a new variant.
When no good example exists, the agent logs the query as a gap note for the catalog pipeline so the team knows which examples to prioritize next.
Sanity AI agent demo records +
Published and draft catalog records with hidden internal metadata, tool tags, industry tags, function tags, audience, and workflow pattern.
Google Drive PRDs and Completed Projects folders +
Source PRDs with problem, inputs, transformations, outputs, tools, and variant guidance for each build.
Linear build tickets and statuses +
Ticket metadata showing build status, assigned team, and completion date.
Member tagging taxonomy +
Industry tags, tool tags, function tags, go-to-market tags, and member benefit labels.
Slack or dashboard chat query +
Internal user query such as "show me go-to-market agents for HVAC" or "what builds use Apollo and HubSpot".
Link each catalog record to source PRD and taxonomy +
The agent maps Sanity records to Google Drive PRDs, Linear tickets, and taxonomy tags so every build has a complete metadata graph.
Embed public-safe and internal-only summaries separately +
Public-safe summaries cover problem, inputs, transformations, outputs, and tools. Internal-only summaries include member names, unpublished build details, variant notes, and exact source PRD excerpts.
Run semantic search filtered by industry, tool, function, and workflow pattern +
The agent applies semantic search across embedded summaries, then filters by taxonomy tags and publish status to return the three best matches.
Rank matches by query fit and add rationale +
Each match shows why it fits the query, which tools and workflow patterns align, similar-build links, and variant notes.
Capture unanswered queries as taxonomy gaps +
When no good example exists, the agent logs the query as a gap note for the catalog pipeline.
Three ranked build matches +
Each match shows build title, problem summary, tool tags, industry tags, workflow pattern, and why it fits the query.
Filters applied and suggested adjacent filters +
The agent shows which taxonomy filters were applied and suggests adjacent filters such as related industries or tools.
Similar builds and variant links +
Each match includes links to similar builds and variant notes so the team can scope a new variant or cite an adjacent example.
Internal source links +
Direct links to the source PRD in Google Drive, Linear ticket, and Sanity record so the team can cite the exact build or review implementation details.
Gap note when no good example exists +
The agent logs unanswered queries as taxonomy or content gaps for the catalog pipeline so the team knows which examples to prioritize next.
Is this for you?
- + Sales teams who need to cite prior build examples on prospect calls - You're on a call with an HVAC contractor who asks "Do you have an example for our industry?" and you need to cite a specific build with tool tags, workflow patterns, and source PRD links without searching Sanity, Drive, and Linear by hand.
- + Delivery teams who need reusable examples for scoping new builds - You're scoping a new dispatch routing agent and need to find prior builds that used similar inputs, transformations, and tools so you can reuse workflow patterns and variant notes.
- + Member-success teams who recommend adjacent builds to members - A member asks "What should I request next?" and you need to surface adjacent builds filtered by their industry, tools, and function tags without relying on memory or scattered Slack threads.
- + Operators who need to identify taxonomy gaps in the catalog - You want to know which queries return no good examples so you can prioritize new builds that fill those gaps.
- - Public-facing catalog search for prospects who haven't talked to sales - This agent returns internal-only records, unpublished build details, member names, and exact source PRD excerpts. Public-facing catalog search needs a separate implementation that filters to published records only.
- - Teams who need full-text search across PRD paragraphs or Slack threads - This agent embeds catalog metadata and source doc summaries, not full-text PRD paragraphs or Slack message history. If you need to search exact PRD wording or Slack threads, you need a different retrieval pattern.
- - Teams who need real-time Linear ticket updates or build status alerts - This agent reads Linear ticket metadata at ingestion time. If you need real-time alerts when a build status changes, you need a separate Linear webhook integration.
Pricing
This is an internal Sagan agent. Pricing covers the scoped build to ingest catalog records, link source PRDs, embed summaries, and deploy the semantic search and taxonomy filter logic, plus usage-based runs for each query from Slack or dashboard chat.
- Scoped build includes Sanity API integration, Google Drive API integration, Linear API integration, vector embedding pipeline, semantic search logic, taxonomy filter implementation, and Slack or dashboard chat interface.
- Usage-based runs cover each query from internal users, including semantic search, taxonomy filtering, ranking, and gap-note capture.
- Vector store and embedding provider are configurable based on your infrastructure: Supabase/Postgres, Pinecone, or another vector database.
- Access control rules for internal-only records and unpublished builds are included in the scoped build.
- Public page demo scope and sample queries are configured separately if you want to show the retrieval pattern to prospects.
How does the agent decide which three builds to return?
The agent embeds catalog metadata and source PRD summaries, then runs semantic search across those embeddings filtered by industry, tool, function, and workflow pattern. It ranks matches by query fit and returns the three best, showing why each fits your query, which tools and workflow patterns align, similar-build links, and variant notes. Unanswered queries log as gap notes for the catalog pipeline.
Can I search for builds that use specific tools like Apollo or HubSpot?
Yes. The agent filters by tool tags, so you can ask "what builds use Apollo and HubSpot" and receive matches that include those tools in their catalog metadata. Each match shows which tools and workflow patterns align with your query.
What happens when no good example exists for my query?
The agent logs your query as a gap note for the catalog pipeline so the team knows which examples to prioritize next. This helps Sagan identify taxonomy or content gaps and build the examples that internal teams and members need most.
Does this agent search full PRD text or just catalog summaries?
The agent embeds catalog metadata and source doc summaries, not full-text PRD paragraphs or Slack message history. If you need to search exact PRD wording or Slack threads, you need a different retrieval pattern. This agent is optimized for fast retrieval of build examples by industry, tool, function, and workflow pattern.
Can I use this agent to find builds for a specific industry like HVAC or roofing?
Yes. The agent filters by industry tags, so you can ask "show me go-to-market agents for HVAC" or "what builds exist for roofing" and receive matches filtered by those industry tags. Each match shows build title, problem summary, tool tags, industry tags, workflow pattern, and why it fits your query.
What source systems does the agent read to build the catalog?
The agent reads Sanity AI agent demo records with hidden internal metadata, Google Drive PRDs and Completed Projects folders, Linear build tickets and statuses, and the member tagging taxonomy including industry tags, tool tags, function tags, and go-to-market tags. It links each catalog record to its source PRD, Linear ticket, and taxonomy so every build has a complete metadata graph.
How is this different from public-facing catalog search for prospects?
This agent returns internal-only records, unpublished build details, member names, and exact source PRD excerpts. Public-facing catalog search needs a separate implementation that filters to published records only. This agent is built for internal teams who need fast retrieval of prior builds and reusable examples, including unpublished and draft builds.
What does the scoped build include?
The scoped build includes Sanity API integration, Google Drive API integration, Linear API integration, vector embedding pipeline, semantic search logic, taxonomy filter implementation, and Slack or dashboard chat interface. Usage-based runs cover each query from internal users, including semantic search, taxonomy filtering, ranking, and gap-note capture. Vector store and embedding provider are configurable based on your infrastructure.
Stop searching Slack threads and Drive folders for the build example you need.
Your team has built dozens of agents across roofing, HVAC, finance, and go-to-market workflows. This agent embeds catalog metadata and source docs, runs semantic search filtered by industry, tool, function, and workflow pattern, and returns three matching builds with tool tags, workflow patterns, variant notes, and source PRD links so you cite the exact build while the prospect is on the line. Talk to Sagan to scope your internal knowledge retrieval agent.