What it is
Onboarding-and-golden-paths is the third leg of Archally, collaborative intelligence for SDLC: a knowledge-driven SDLC platform with an archive of architecture and business decision records; onboarding curricula and golden-path templates (Backstage / Cortex / Port.io targets) are proposed outputs. New engineers ramp against the same model the team uses to architect, govern, and ship - not against stale wiki pages or whiteboard reconstructions.
Built on
- Curriculum projected from the blueprint (proposed) - guided model tours per new-hire role
- Golden-path template emit (proposed) - Backstage / Cortex / Port.io bundles
- archive of architecture and business decision records (delivered) - time-travelable + evidence-linked
- Multi-view living docs (delivered) - provides the docs substrate
- MCP-backed agent-led onboarding (delivered) - new hires query the model through agents
How it works
The blueprint already holds the bounded contexts, decisions, and evidence. The curriculum generator (proposed, defining) would walk the model and emit a guided tour per new-hire role: which contexts to read first, which decisions explain the architecture, which evidence backs which trade-off. Golden-path templates (proposed) would emit as Backstage / Cortex / Port.io bundles. The decision archive (delivered) surfaces architecture and business decision records as time-travelable, evidence-linked entries. New hires can query recorded decisions and evidence through an MCP-backed agent; review answers against source evidence and agree supported access for the engagement.
The future of software is no longer written - it is architected, governed, and continuously learned.
- 44% of organisations report developer onboarding takes >2 months Cortex - Developer onboarding guide
- 62% faster time-to-productivity with structured onboarding (Stack Overflow 2024) Cortex - Developer onboarding guide, citing Stack Overflow 2024
- -55% Spotify's time-to-tenth-pull-request after Backstage rollout (external benchmark) PlatformEngineering.com - Backstage: a mid-year snapshot
- ~9% typical Backstage adoption outside Spotify - DIY portals fail at scale Port - Top Backstage alternatives
Decision archive
Architecture decisions stop being Markdown drift. Capture every architecture decision (ADR D###) and business decision record as first-class typed entities - time-travelable, cross-referenceable to evidence (EVD###), value-streams, bounded contexts, and migrations. MADR-compatible Markdown emit for ecosystem-tool interoperability is on the roadmap. Aligns with the community ADR ecosystem (MADR, joelparkerhenderson/ADR, UK Government ADR Framework) but goes structurally further than plain Markdown.
Onboarding curriculum
Onboarding is a model-emission target, not a wiki-writing exercise. The proposed curriculum generator would walk the blueprint, rank contexts by entry-pathway importance for the new hire's role, and emit a guided tour: starting bounded contexts, foundational decisions, the evidence trail behind the architecture, and live MCP queries the new hire can run themselves. Spotify cut time-to-tenth-PR by 55% with Backstage; this is an external benchmark, not a measured Archally result.
Complements your existing knowledge surfaces
Archally is the upstream model - it complements, never replaces, the page-storage and Q&A surfaces your teams already use.
Confluence
complement - page storage + Rovo AI
Archally can publish declared blueprint-derived pages to Confluence after preview and review. Review the model against source evidence; Confluence stays the page surface your stakeholders already read.
Notion
complement - work OS + Notion AI
Notion database rows are a proposed output for model entities: one row per offering, one row per decision, one row per bounded context. Notion stays the work-OS surface; any exported rows would need review against the source model.
Stack Overflow for Teams
complement - internal Q&A surface
Q&A pairs for Stack Overflow for Teams are a proposed output from the decision archive: “why did we choose X over Y” answered with the captured rationale and evidence_refs. Stack Overflow for Teams stays the Q&A surface; any exported answers would need review against the source model.