authentik
The first patch on 2026.8 is twenty-odd fixes and no new surface.
A side-by-side editorial comparison of Stacker and Windmill — release velocity, themes, recent moves, and the top alternatives to consider.
Stacker's AI agent has absorbed every part of app building except the data itself.
Stacker builds internal tools and customer portals on top of existing data. From late October 2025 the entire changelog is one project: an AI Builder Agent that generates and edits apps by conversation. In under seven weeks it went from generating initial pages and layouts to editing tables and fields, creating lookups, rollups and formulas, reshaping create forms, reordering navigation, building white-labelled external portals with custom domains and permissions, and finally taking a spreadsheet upload and configuring a schema and matching layouts from it.
Windmill gives its AI agents a test harness, the piece that lets them into production flows
Windmill is a developer platform for scripts, flows and internal apps, shipping weekly across three lanes. The AI lane is the busiest: sessions that build scripts and flows as reviewable drafts, a read-only plan mode that refuses writes until you approve, version history on session artifacts, and now evaluation for reusable agents. The runtime lane keeps absorbing data tooling — dbt projects run natively, BigQuery and Snowflake came out from behind the Enterprise licence. The workspace lane is consolidating deployment semantics around fork lineage.
Stacker builds internal tools and customer portals on top of existing data. From late October 2025 the entire changelog is one project: an AI Builder Agent that generates and edits apps by conversation. In under seven weeks it went from generating initial pages and layouts to editing tables and fields, creating lookups, rollups and formulas, reshaping create forms, reordering navigation, building white-labelled external portals with custom domains and permissions, and finally taking a spreadsheet upload and configuring a schema and matching layouts from it.
The sequencing is deliberate: each release hands the agent one more thing that previously required the no-code editor, while the company repeatedly reassures users that the full editor still works on AI-built apps. That dual framing — build by conversation, refine by hand — is the position Stacker is defending, and it is a reasonable answer to pure prompt-to-app tools that leave nothing to adjust afterwards. The portal release is the notable widening, since it points the agent at the external-user product rather than internal tools.
Every surface except data connection has now been handed to the agent; connecting and mapping an external source — the step still done manually before any of this works — is the remaining gap, and the spreadsheet uploader is the first move toward closing it.
Windmill is a developer platform for scripts, flows and internal apps, shipping weekly across three lanes. The AI lane is the busiest: sessions that build scripts and flows as reviewable drafts, a read-only plan mode that refuses writes until you approve, version history on session artifacts, and now evaluation for reusable agents. The runtime lane keeps absorbing data tooling — dbt projects run natively, BigQuery and Snowflake came out from behind the Enterprise licence. The workspace lane is consolidating deployment semantics around fork lineage.
The AI work has moved through a recognisable sequence: first let the assistant build things, then constrain what it may do without approval, then make its output reviewable and revertible, and now make agents measurable against a fixed set of cases. That is the arc of taking AI from a demo surface to something you are willing to put in a scheduled flow. In parallel, Windmill keeps pulling data-stack runtimes inside its own execution model rather than integrating with them, and keeps giving away runtimes that were previously licensed.
Evals are the obvious hook for regression gating, so expect them to be callable from a flow or CI rather than only from the agent card, and expect the eval run history to feed the same version-comparison surface the session artifacts already use.
Other Infra & APIs products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Stacker or Windmill.
The first patch on 2026.8 is twenty-odd fixes and no new surface.
Alert rate limits stop being per-source and start being per-tenant, by query.
The essay series keeps running ahead of the products it describes.
Both lines patch on the same day, and the reverted JSONata upgrade finally sticks.
The 3.x line finally ships function: authenticated secret writes and render patches.
Folder-level RBAC lands, and approvals finally get a webhook to talk to.
See all Stacker alternatives → · See all Windmill alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
Both compete on the same themes — ai-agents — within Infra & APIs. Windmill is currently shipping more aggressively (velocity 6.3 vs 0.0), with 1 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Windmill is currently shipping more aggressively (velocity 6.3 vs 0.0), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Infra & APIs products to evaluate alongside.
Top Stacker alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Stacker alternatives" section above for the current picks, or visit /alternatives/stacker for the full list with editorial commentary on each.
Top Windmill alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Windmill alternatives" section above for the current picks, or visit /alternatives/windmill for the full list with editorial commentary on each.