Basedash Skills

Product Introduction

1. Definition: Basedash Skills is an AI context management and semantic layer feature within the Basedash business intelligence (BI) platform. It is a technical system for defining, storing, and distributing reusable business logic and analytical playbooks to AI agents.
2. Core Value Proposition: It exists to eliminate repetitive manual instruction and ensure AI-powered data analysis is consistent, accurate, and aligned with company-specific definitions. Its primary value is "write once, apply everywhere," enabling teams to scale AI governance and embed institutional knowledge directly into their analytics workflow.

Main Features

1. Modular Skill Catalogue: Admins create discrete "skills" in a centralized repository (Settings → AI context). Each skill is a plain-text instruction set up to 50,000 characters, focused on a single concept like a metric definition, chart preference, or operational playbook. This modularity allows for granular management and on-demand retrieval, unlike a monolithic context block.
2. On-Demand, Auditable Retrieval: The system does not blindly inject all skills into every AI prompt. Instead, AI agents receive a lightweight catalogue of skill names. When a user's request is semantically relevant, the agent performs a tool call to fetch the full skill instructions. This retrieval is visible in the AI's "thinking trace," providing full auditability and transparency into which rules influenced the final answer.
3. Cross-Surface Context Propagation: The singular skills catalogue feeds every AI surface within the Basedash ecosystem. This includes the AI Chat for Q&A, the Chart Builder for visualizations, Dashboards, Automations for scheduled reports, Insights for trend detection, and background Tasks. This ensures a unified semantic layer across all interaction points.

Problems Solved

1. Pain Point: Inconsistent AI outputs and "metric spaghetti" where the same business concept (e.g., "Monthly Recurring Revenue" or "activation") is calculated differently across chats, charts, and reports due to vague or repeated prompt instructions.
2. Target Audience: Data & Analytics Leaders (Heads of Data, Analytics Engineers) who need to enforce a single source of truth. Product & Growth Managers who require consistent metric tracking. Operations & Support Managers who need to standardize workflow analysis. Company Admins in SaaS companies tasked with scaling AI tool governance.
3. Use Cases: Enforcing a semantic layer for key SaaS metrics (MRR, churn, retention). Standardizing analytical playbooks for A/B test reporting or support ticket triage. Encoding team charting conventions (e.g., "always use a line chart for trends"). Onboarding new team members' AI interactions with institutional knowledge automatically.

Unique Advantages

1. Differentiation: Unlike traditional BI semantic layers that only work for pre-built reports or require complex SQL/Model definitions, Basedash Skills operates in natural language and is dynamically integrated into generative AI interactions. Compared to manually pasting context into each ChatGPT or Claude prompt, Skills provide a managed, scalable, and auditable system.
2. Key Innovation: The hybrid retrieval mechanism. By giving AI agents a dynamic "read" capability instead of a static, context-length-limited prompt, it enables a much larger, more organized body of knowledge to be available. The audit trail (seeing which skills were "read") directly addresses the "black box" concern of AI-assisted analysis, building trust and enabling debugging.

Frequently Asked Questions (FAQ)

1. How does Basedash Skills ensure AI uses the correct metric definition? Basedash Skills acts as a centralized semantic layer; when a user asks about "activation rate," the AI agent retrieves the specific "Activation Rate" skill containing the exact definition (e.g., "within 7 days, excludes trials"). This enforced definition is then applied whether generating a SQL query, a chart, or a text summary.

2. What is the difference between Global Context and Skills in Basedash AI? Global Context is a static block of instructions (e.g., "We are a SaaS company") loaded into every AI prompt, best for universal, always-relevant facts. Skills are modular, on-demand instruction sets for specific concepts (e.g., "Revenue Analysis Playbook"), loaded only when relevant, making them ideal for detailed, situational playbooks without bloating the core prompt.

3. Can Basedash Skills replace a traditional data warehouse semantic layer? Basedash Skills is a complementary, AI-native semantic layer focused on governing the interpretation and presentation of data through natural language. It works alongside your warehouse semantic layer (like dbt or LookML), which governs the underlying data transformation. Skills instruct the AI how to use the data that is already modeled.

4. Who has permission to edit or create Skills in an organization? Administration of Basedash Skills is restricted to organization admins via the Settings → AI context menu. This permission model ensures control and consistency over the business logic that all AI agents use, while all other workspace members automatically benefit from the defined skills in their AI interactions.