Market Positioning: Domain-Specific KB Creationdesign_doc

pluginspositioningplatformdomainsvertical
2026-02-23 · 4 min read · Edit on Pyrite

Market Positioning: Domain-Specific KB Creation

Priority: 3 — Build partnerships

Market Overview

Every specialized field needs structured knowledge management, and every field builds bespoke tools or forces their data into generic systems that don't fit:

  • Legal teams track cases, statutes, precedents, and relationships between them — in SharePoint folders
  • Medical researchers manage patient cohorts, trial data, literature connections — in Excel spreadsheets
  • Policy analysts track legislation, stakeholders, timelines, regulatory changes — in Google Docs
  • Patent analysts map technology landscapes, prior art, inventor networks — in proprietary databases
  • Compliance teams maintain regulatory requirements, audit findings, control mappings — in GRC tools that cost six figures
  • The common pattern: domain experts know what types of knowledge they need (cases, trials, regulations, patents) and how those types relate to each other — but they can't express this without hiring developers to build custom tools.

    Competitive Landscape

    | Competitor | Approach | Why It Falls Short | |-----------|----------|-------------------| | Notion databases | User-defined properties, relations, rollups | No temporal queries; no full-text search across relations; flat (not graph); vendor lock-in | | Airtable | Spreadsheet-database hybrid with linked records | Not a knowledge base; no markdown content; no version history; expensive at scale | | Obsidian + Dataview | User-defined frontmatter + query plugin | Powerful but requires learning Dataview query syntax; single-user; no typed validation | | Semantic MediaWiki | Wiki + structured semantic properties + SPARQL | Mature but aging; MediaWiki dependency; steep setup; limited visualization | | Custom Neo4j builds | Graph database + domain ontology | Requires developer; no out-of-box UI; no markdown content model; no collaboration | | Vertical SaaS (Relativity, Veeva, etc.) | Domain-specific platforms | Expensive ($50K-500K/yr); vendor lock-in; can't customize types; no AI agent access |

    Key gap: No platform lets domain experts define their own typed knowledge structures without writing code, while still getting full-text search, relationship graphs, temporal queries, and AI agent integration.

    Pyrite Differentiation

    YAML-defined types without code — A legal researcher defines their domain in `kb.yaml`:

    ```yaml types: case: description: "Legal case or proceeding" fields: jurisdiction: type: select options: [federal, state, international] status: type: select options: [active, decided, appealed, settled] filing_date: type: date parties: type: list items: type: text ```

    No Python. No developer. Auto-validation and UI generation from this schema.

    Plugin protocol for deeper customization — When YAML isn't enough, a developer can write a plugin that adds custom entry types, relationship semantics, workflows (e.g., peer review state machine), and domain-specific MCP tools — all while maintaining the same markdown-in-git data model.

    Multi-KB for multi-domain — A policy organization can maintain separate KBs for legislation, stakeholders, and news events — each with its own types — and query across all of them. The cross-KB link model is unique.

    Agent-powered domain workflows — An AI agent can be given read access to a legal KB and write access to a research-notes KB, with domain-specific MCP tools: "Find all cases citing Chevron since 2024" or "Create a timeline of regulatory actions affecting Company X."

    Content negotiation for diverse consumers — API responses in JSON, Markdown, CSV, or YAML via `Accept` header. Domain teams can export search results as CSV for spreadsheet workflows or get Markdown for documentation.

    What's Already Built

    | Capability | Status | |-----------|--------| | YAML-defined types with 10 field types (text, number, date, select, multi-select, etc.) | Shipped | | Field-level validation (range, format, required, options) | Shipped | | Plugin protocol with 12 extension points | Shipped | | 3 proof-of-concept extensions (zettelkasten, social, software-kb) | Shipped | | Type metadata with AI instructions per type | Shipped | | Templates system for entry scaffolding | Shipped | | Multi-KB with cross-KB links | Shipped | | Content negotiation (JSON, Markdown, CSV, YAML) | Shipped |

    Ideal Customer Profile

    1. Research organizations (think tanks, policy institutes, NGOs) with structured knowledge needs 2. Specialized consulting firms (legal research, patent analysis, compliance) who build internal tools 3. Academic research groups managing multi-year longitudinal studies 4. Government analysts tracking regulatory landscapes, stakeholder networks, policy timelines

    Go-to-Market

    Immediate:

  • Publish 2-3 domain extension examples beyond the existing four (legal-kb, policy-kb, medical-research-kb)
  • Write guide: "Build a custom knowledge base for your domain in 30 minutes"
  • Identify 2-3 partner organizations willing to co-develop domain extensions
  • Next quarter:

  • Extension marketplace / registry (even if just a GitHub topic tag initially)
  • "Domain starter kits" — pre-built kb.yaml + sample entries for common verticals
  • Community contribution workflow: fork an extension, customize, publish
  • Later:

  • No-code extension builder UI (backlog item already exists)
  • Hosted offering for non-technical domain teams
  • Certification/partnership program for consulting firms building on Pyrite
  • Feature Gaps

    | Gap | Effort | Impact | |-----|--------|--------| | Domain starter kit templates (legal, policy, medical) | S each | High — proves platform versatility | | Extension marketplace / registry | M | High — network effects | | No-code extension builder UI | L | Transformative — opens non-developer market | | Import/migration tools (CSV, JSON, existing databases) | M | High — adoption requires data migration | | Role-based access per KB (not just per tier) | M | Medium — enterprise requirement |

    Risks

  • "Just use Notion" — for simple domains, Notion databases are good enough and already adopted
  • Schema design is hard — domain experts may need guidance; bad schemas create bad KBs
  • Chicken-and-egg — platform value depends on extensions; extensions depend on platform adoption
  • Vertical SaaS incumbents have deep domain features Pyrite can't match without significant plugin development
  • Maintenance burden — each domain extension needs ongoing maintenance as the core evolves
  • Single-maintainer project — platform credibility requires visible project health; mitigated by 583 tests, 11 ADRs, comprehensive plugin guide, and clear contribution path