A vendor-neutral, declarative format for AI-native analytical dashboards
Modern organizations face a fundamental problem with their analytical dashboards: they're locked into proprietary platforms, impossible to version control, and can't be generated by AI at scale. Each vendor uses its own format, dashboards are defined through point-and-click interfaces, and migrating between platforms requires complete rebuilds.
The Universal Dashboard Specification (UDS) solves this by providing a vendor-neutral, declarative format for describing analytical dashboards. Think of it as "the Kubernetes of dashboards" — a specification layer that sits above any rendering platform.
UDS enables:
- Semantic-first design: Dashboards reference metrics and dimensions from your existing semantic layer (Cube, dbt Semantic Layer, Looker, etc.) rather than hard-coding SQL
- Intent-driven declaration: Every dashboard explicitly declares its purpose, the decisions it enables, and who it serves
- AI-powered generation: Built from the ground up for AI assistants to generate dashboards with constraints, confidence scores, and human review workflows
- True portability: Write once, compile to Grafana, Metabase, or custom renderers
- Version control native: Dashboards are YAML files that live in git alongside your application code
- Persona awareness: Dashboards adapt their complexity, metrics, and visualizations to executives, managers, analysts, operators, or external users
UDS is not a rendering engine. It's a specification that owns the abstraction layer while letting you choose your rendering platform. This creates a sustainable ecosystem where tooling vendors can build parsers, renderers, and AI agents without platform lock-in.
- Semantic-first: References metrics/dimensions from semantic layer (Cube, dbt, Looker, etc.) instead of raw SQL
- Intent-driven: Every dashboard declares its purpose and the decisions it enables
- Persona-aware: Adapts to executive, manager, analyst, operator, and external user personas
- AI-native: Designed for generation with constraints, confidence scores, and human review
- Portable: Compile to Grafana, Metabase, Looker, or custom renderers
- Accessible: WCAG 2.1 AA compliant by design with screen reader support
- Extensible: Layered conformance (Core/Standard/Complete) allows gradual adoption
- Version control friendly: Plain YAML that works with git, CI/CD, and code review
Here's a minimal UDS dashboard that shows daily revenue:
uds: "0.1.0"
dashboard:
title: "Sales Overview"
intent: "Monitor daily sales performance and identify revenue trends"
persona:
type: manager
customization_level: moderate
semantic_sources:
- id: sales
type: cube
endpoint: "https://cube.example.com/cubejs-api/v1"
auth:
type: bearer
token_env: "CUBE_API_TOKEN"
layout:
type: grid
columns: 12
row_height: 60
panels:
- id: revenue
type: kpi
title: "Total Revenue"
position:
row: 0
column: 0
width: 4
height: 2
data:
source: sales
metrics:
- ref: "sales.total_revenue"
time_range:
type: relative
value: "today"
formatting:
number_format:
style: currency
currency: USDThis dashboard:
- Declares its intent (monitoring sales performance)
- Specifies it's for a manager persona
- References metrics from a Cube semantic layer
- Uses a responsive grid layout
- Includes a single KPI panel with currency formatting
- Specification: Complete UDS v0.1.0 specification
- Examples: See the examples/ directory for sample dashboards
- JSON Schema: Coming in Phase 3
- Tooling: Reference parser and renderer coming in Phases 4-5
Current Version: v0.1.0 (Draft)
- Phase 1: ✅ Specification complete (v0.1.0)
- Phase 2: Examples and sample dashboards (in progress)
- Phase 3: JSON Schema for validation
- Phase 4: Reference parser (TypeScript)
- Phase 5: Reference renderer (React)
- Phase 6: Conformance test suite
- Phase 7: Documentation site
- Phase 8: Public launch and community building
Future ecosystem projects (coming soon):
uds-parser: Reference TypeScript parser for UDS documentsuds-renderer: React-based reference rendereruds-validator: JSON Schema validator and conformance checkeruds-ai-agent: AI assistant for generating UDS dashboardsuds-compiler-grafana: Grafana JSON compileruds-compiler-metabase: Metabase API compiler
This specification accompanies From Policy to Practice: An Open Framework for AI-Ready Project Delivery (Newman, 2026). DOI: https://doi.org/10.5281/zenodo.18711384
UDS is designed for:
- AI-powered dashboard generation: LLMs generate dashboards from natural language with built-in constraints and review
- Platform migration: Move dashboards between Grafana, Metabase, Looker, or custom platforms
- Version control and CI/CD: Store dashboards in git, review changes, and deploy via pipelines
- Multi-tenant analytics: Generate personalized dashboards for different user personas
- Semantic layer adoption: Decouple dashboards from raw SQL and reference your semantic layer
- Compliance and governance: Enforce data access policies, accessibility standards, and documentation requirements
UDS is in early development. We welcome contributions from:
- Dashboard platform vendors
- Semantic layer providers
- AI/ML companies building generation tools
- Data practitioners and dashboard designers
- Open source enthusiasts
See CONTRIBUTORS.md for the list of contributors.
Note: A Contributor License Agreement (CLA) will be required for all contributions. Details will be published as the project matures.
If you use this specification in your research or work, please cite:
Newman, A. (2026) From Policy to Practice: An Open Framework for AI-Ready Project Delivery.
London: Tortoise AI. DOI: https://doi.org/10.5281/zenodo.18711384
Copyright 2026 Tortoise AI Ltd
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at:
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the LICENSE file for the specific language governing permissions and limitations under the License.