Skip to main content

How Devplan Works

Devplan learns about your product from the tools and documents you already use, then surfaces what matters through daily digests, insights, and Ask Devplan. It also supports structured feature planning — PRDs, user stories, technical designs, and export to PM tools.

Two layers:

  1. Product intelligence — ingest context and integrations, build the Weaver knowledge graph and feature catalog, extract signals, synthesize insights, deliver Today.
  2. Feature planning — turn ideas into implementation-ready specs and coding prompts grounded in that context.

See Overview, Getting Started, Core Workflow, and Platform Overview.


Product intelligence pipeline

LayerWhat it isWhere you see it
Workspace contextGoals, customers, competitors — what you tell DevplanWorkspace, onboarding
IntegrationsLive data from GitHub, Slack, Jira, and other toolsIntegrations
Document uploadsFiles in the workspace knowledge graphKnowledge > Integrations uploads
Weaver (knowledge graph)Connected model of people, features, and researchPowers Ask Devplan and cross-source reasoning
Feature catalogStructured map of product featuresProduct
SignalsIndividual evidence itemsSignals
InsightsSynthesized, role-aware takeawaysInsights
Daily DigestSummary of what changed, with evidenceToday

Workspace context frames how Devplan interprets signals. Integrations feed the pipeline on a schedule — repo bootstrap often takes 30+ minutes; signals and insights grow over hours and days. Uploads are parsed into the knowledge graph alongside integration data.


Weaver: the knowledge graph

Weaver is the continuous knowledge graph that powers everything else. It does three things on an ongoing basis:

  • Ingest signals from code, conversations, tickets, and docs
  • Maintain memory of product context, decisions, and history over time
  • Route insight to people and AI agents — in the interface, Slack, or MCP-enabled tools

By linking workspace context, uploads, catalog features, tickets, and code into one persistent graph, Weaver enables cross-source answers in Ask Devplan and the evidence citations on Today and Insights. This shared, durable memory is what keeps responses grounded in your organization rather than hallucinated.

The feature catalog maps features to code and docs. Catalog bootstrap runs after code repos are connected; catalog update jobs keep Product and Updates current. Jira context from allowed projects enriches bootstrap.


Signals, insights, and planning

Signals are raw evidence — PRs, Slack threads, tickets, uploads. Insights synthesize related signals into risks, opportunities, and themes, personalized by Preferences roles. Both feed Today.

For feature planning, Devplan maps requirements to components and APIs for impact analysis, story sequencing, and context-rich coding prompts. Planning agents run discovery, PRD review, and attachment processing. See Spec Driven Development for the planning workflow.


Integrations

Connect sources under Knowledge > Integrations. For most providers, Connect is only the first step — you also enable repos, attach pages or folders, invite the Slack bot, or select Jira projects before data flows in.

IntegrationWhat Devplan readsWhat it powers
GitHubRepos, PRs, commits, code structureCatalog bootstrap, code-change signals, planning context
BitbucketSame as GitHub (requires Forge app)Catalog bootstrap, code-change signals
JiraIssues and project activity from Allowed JIRA Projects (requires Forge app)Catalog enrichment, Signals, Insights
LinearIssues and team activity from connected teamsSignals, Insights
SlackMessages in channels where the Devplan bot is a memberDiscussion signals, alignment gaps
NotionAttached pages and databases (on demand)Signals, knowledge graph, Ask Devplan
Google DriveAttached folders via a workspace service account (on demand)Signals, knowledge graph, Ask Devplan
ConfluenceAttached spaces and pages (on demand; same Forge app as Jira)Signals, knowledge graph, Ask Devplan
Upload filesPDFs, docs, spreadsheets, images you upload directlyKnowledge graph, signals, insights
GranolaCall notes and meeting summaries (paid Granola accounts only)Interview and stakeholder signals

Code repos drive catalog bootstrap — first analysis often takes 30+ minutes. Other sources populate Signals and Insights over hours and days as background jobs run.

Setup guides: Integrations overview.

CLI

The CLI writes a context file for your AI IDE — project context, current task, code patterns, file references, and repo guidance. See CLI Cheat Sheet.


Access control

RoleInternal namePermissions
AdminOWNERFull access — members, settings, integrations
UserEDITORProjects, documents, integrations — no member/settings management

See Workspace Members.


Data privacy

Devplan analyzes code structure and metadata — not source code, business logic, credentials, or customer data. Hybrid deployment is available for on-premises codebase analysis.