PortalJS: state of the nation and next 6 months

An assessment and diagnosis 2026-09-28

Site: https://portaljs.com/ · Code: https://github.com/datopian/portaljs

Situation

• There are now three PortalJSes:
  • PortalJS (framework) — open-source, AI-native toolkit for building data portals.
  • PortalJS Arc (maybe to be renamed Studio) — AI-driven builder and deployer: scaffold, set up, manage and host a portal. Lightweight hosting: datasets stored in Arc by default, site statically built, no backend service. Free today, no pricing model.
  • PortalJS Cloud — hosted, multi-tenant, CKAN-backed PortalJS (managed CKAN SaaS).
    • Plus an enterprise option: dedicated cluster / servers per client (e.g. CDT). Valuable to many clients but not clearly presented as part of Cloud.
• Most client hosting is still bespoke / enterprise-dedicated rather than self-serve sign-ups. TODO: numbers.
• Queryless is out. Hard to make reliable, and people already have their own AI: they want to talk to a portal via MCP or API, not a built-in chat. (Aside: an open-source skill for talking to data portals may still be worth doing.)
• Market: competitors are OpenDataSoft, Socrata (Tyler), Esri ArcGIS Hub, and new entrant EntryScape.
  • We have lost several deals to OpenDataSoft. We have no feature comparison with them. TODO: create one.
  • Socrata market looks to be eroding: customers moving to open source or cutting costs (e.g. NASA moved Socrata → CKAN).
  • Socrata-instance consultants approached us; exchanged emails, no ongoing relationship. Our position: partner if they bring clients.
• Marketing / community:
  • Community activity has gone quiet: Discord had active involvement (Joao) but nothing for months, likely as client work took over.
  • No visible changelog on the site (homepage links blog, docs, pricing, Discord only).
  • Sign-up / onboarding flow after someone signs up is unclear.
  • We have usage data in PostHog but no recent funnel analysis. A colleague is doing one now.

Complication

• Little business coming in on the self-serve end. The issue is now less product and more marketing — including using the product itself (e.g. PoCs) as marketing.
• Arc and Cloud both offer hosting, so they partly compete; no clear reason for a buyer to pick one. Not seeing confusion from clients yet, possibly because few find Arc at all.
• Arc is free with no pricing, so the most self-serve piece earns nothing and doesn't clearly feed Cloud.
• Enterprise Cloud (dedicated cluster) is what clients actually buy, yet it isn't visible in how we present Cloud.
• Small, steady marketing / community upkeep isn't protected, so it dies when people get busy (e.g. a ruthlessly protected half-day a week would make a big difference).
• Socrata erosion is an opening, but we have no deliberate play (positioning, migration offer, partner channel) to capture it.

Question

• What should PortalJS focus on over the next 6 months? Candidate framing: how do we turn PortalJS into a working marketing and self-serve funnel — a clear, paid path from framework → Arc/Studio → Cloud (incl. enterprise) — and use it to capture the Socrata / proprietary-portal migration market?