Portfolio Intelligence

Strategic portfolio knowledge graph and natural language query interface. Surfaces OKR alignment, project lineage, and data domain connections across all registered Wasteology projects.

Page Overview

Portfolio Intelligence

Ask natural language questions about the portfolio, explore the project graph, and understand OKR and data domain relationships

How to Use

1Type a natural language question in the NL Query interface and review the structured answer

2Navigate the Project Graph to explore relationships between owned and support projects

3Review the OKR Alignment view to understand how projects map to organizational objectives

4Open the Data Domains view to see which projects share data contracts

Features

💬

NL Query Interface

Ask plain-English questions about the portfolio and get structured answers drawn from the knowledge graph and wg_digests

🕸️

Project Graph

Interactive node graph of all owned and support projects with labeled relationship edges

🎯

OKR Alignment

Visual mapping of each project to its contributing organizational objectives and key results

🔗

Data Domains

Cross-project data contract visualization — see which projects share schemas, tables, or pipelines

NL Query Interface

The query interface accepts plain-English questions and traverses the portfolio graph combined with a live query against wg_digests to produce a structured answer.

Example Queries

Question	What It Returns
"Which projects write to the enevo_normalized schema?"	Projects + flows that have a write relationship to that schema
"What is the health of the CieTrade pipelines this week?"	Pipeline health summary from the latest digest
"Which projects are owned vs support?"	Categorized project list from the manifest registry
"What OKR does wdp-import-pipeline support?"	OKR alignment entry for the project
"What open tasks are there across the portfolio?"	Open digest tasks from `wg_digests.digest_tasks`

How It Works

The query is sent to the wdp-palantiri sidecar
The sidecar traverses the in-memory portfolio graph built from TOML manifests
Relevant wg_digests tables are queried for live data (pipeline status, tasks, digest summaries)
A structured response is assembled and returned to the page

Sidecar Dependency

The NL query interface requires the wdp-palantiri sidecar to be running with WG_ORCHESTRATION_PATH set to the wg-orchestration repo on the server. If the sidecar is unavailable, the page falls back gracefully to an empty state with a "Sidecar offline" message.

Project Graph

The Project Graph renders all registered projects as nodes with labeled directional edges describing relationships.

Node Types

Node Color	Category
Blue	Owned project (primary Wasteology systems)
Gray	Support project (third-party integrations or vendor systems)

Edge Types

Edge Label	Meaning
`writes_to`	This project produces data consumed by the target
`reads_from`	This project depends on data produced by the target
`deploys_via`	This project uses the target for CI/CD
`monitors`	This project observes or alerts on the target

Interaction

Click a node to highlight its direct relationships and collapse unrelated nodes
Hover an edge to see the specific schemas or tables involved in the data relationship
Use the search bar to locate a project by name or alias

OKR Alignment

The OKR Alignment view shows a matrix of projects vs. organizational objectives.

Column	Description
Project	Registered project name from the manifest
Objective	Top-level organizational objective
Key Result	Specific measurable outcome the project contributes to
Contribution Type	`primary` (project directly enables the KR) or `supporting` (indirect contribution)

Data Domains

The Data Domains view visualizes data contracts between projects — which projects share or depend on the same schemas, tables, or pipeline outputs.

Use this view to:

Understand blast radius before making schema changes
Identify which teams to notify when a shared pipeline changes
Find undocumented dependencies between projects

Data Sources and Configuration

Project Registry

Projects are loaded from TOML manifests at wg-orchestration/manifests/builtin/*.toml. Each manifest declares:

[project]
id = "wdp-import-pipeline"
name = "WDP Import Pipeline"
type = "pipeline"
path = "~/projects/wdp-import-pipeline"
category = "owned"
aliases = ["wdp-import", "import"]

No code changes are needed when adding a new project — create a manifest file and the graph updates on the next sidecar restart.

Live Data

Data	Source Table
Pipeline health	`wg_digests.project_statuses`
Open tasks	`wg_digests.digest_tasks`
Digest summaries	`wg_digests.daily_digests`

Environment Variable

Variable	Description
`WG_ORCHESTRATION_PATH`	Path to the wg-orchestration repo on the server — required for the sidecar to locate TOML manifests

Empty State

If WG_ORCHESTRATION_PATH is not set or the path does not exist, all four views fall back to an empty state. The NL query interface displays "Portfolio graph unavailable — sidecar offline." Check the server environment configuration if this occurs in production.

Page Overview​