Hostaway
Hostaway builds out the direct-booking stack while tightening who sees what
A side-by-side editorial comparison of Camunda and Tracecat — release velocity, themes, recent moves, and the top alternatives to consider.
Camunda is untying the business key from process start — assign it late, propagate it down.
The 8.10 alpha series is built around one theme: business IDs. Across engine, protocol, REST, gRPC, the Java client and the exporters, Camunda added late assignment via CompleteJob, forward-propagation semantics, resolution into child instances at call activity, deploy-time validation of the businessId attribute on zeebe:calledElement, and exposure in FEEL expressions. Underneath it, agent-instance data now gets cleaned up at the end of a process instance lifecycle and loop counters were renamed across agent instance history.
Ten release candidates in, Tracecat is stabilising the sandbox its agents run inside
Tracecat is ten release candidates into beta.52, following the beta.51 release that made the Action Gateway mandatory. The last two candidates have shifted from adding agent surface to holding it up: registry cache capacity invariants and contention retries, actions retried after heartbeat timeouts, the gateway preserved during shutdown, stale loop device nodes recovered. The agent features still landing are refinements — inline version diffs for agent presets, tool search disabled for Bedrock models — rather than new capability.
The 8.10 alpha series is built around one theme: business IDs. Across engine, protocol, REST, gRPC, the Java client and the exporters, Camunda added late assignment via CompleteJob, forward-propagation semantics, resolution into child instances at call activity, deploy-time validation of the businessId attribute on zeebe:calledElement, and exposure in FEEL expressions. Underneath it, agent-instance data now gets cleaned up at the end of a process instance lifecycle and loop counters were renamed across agent instance history.
Business IDs are being promoted from a value you set once at start into a first-class, propagating, late-bindable identifier — which is what a process engine needs when the domain entity a process refers to is not known until partway through. In parallel, a canonical endpoint-to-permission mapping is being exposed as the single source of truth for RBAC, and the agent-instance surface implies agentic constructs are being maintained as ordinary engine state with their own lifecycle and cleanup.
Expect 8.10 to reach general availability with the business ID work complete across all three client surfaces, and the RBAC mapping to become the basis for generated authorization documentation or tooling. The agent-instance cleanup and history renames suggest that surface is still pre-stable and will keep changing shape before it settles.
Tracecat is ten release candidates into beta.52, following the beta.51 release that made the Action Gateway mandatory. The last two candidates have shifted from adding agent surface to holding it up: registry cache capacity invariants and contention retries, actions retried after heartbeat timeouts, the gateway preserved during shutdown, stale loop device nodes recovered. The agent features still landing are refinements — inline version diffs for agent presets, tool search disabled for Bedrock models — rather than new capability.
The train has crossed from building to hardening. Earlier candidates put agents into case comments as mentions and sessions; the recent ones repair what breaks when that runs under load, which is the registry cache, the socket budget, and the gateway lifecycle. Universal cursor pagination arriving this late reads the same way — an API consistency debt being settled before a final. A ten-candidate train on one beta says the surface changed more than the version number admits.
beta.52 should land as a final with agent-in-comments as its headline, and the fix-only shape of the last two candidates suggests that is close. Whether MCP reference correlation grows into workspace-level integration mapping is still not visible here.
Other PM 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 Camunda or Tracecat.
Hostaway builds out the direct-booking stack while tightening who sees what
ClickUp folds MCP into Super Agents, wiring outside tools into its AI coworker
Hive keeps tightening the same three seams: planned time, admin control, and AI review scope
Akiflow's newest release finally adds capability, but reliability work still fills the notes.
The changelog turned into a how-to blog — capability news now arrives disguised as tutorials.
A time tracker whose blog argues both sides of employee monitoring, and never mentions the product.
See all Camunda alternatives → · See all Tracecat alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Camunda and Tracecat are shipping at a similar cadence (velocity 6.3 vs 6.3, both within Sparkpulse's "active" band). 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. Camunda and Tracecat are shipping at a similar cadence (velocity 6.3 vs 6.3, both within Sparkpulse's "active" band). For your specific use case, the alternatives sections above list other PM products to evaluate alongside.
Top Camunda alternatives in PM are ranked by recent ship velocity. Browse the "Camunda alternatives" section above for the current picks, or visit /alternatives/camunda for the full list with editorial commentary on each.
Top Tracecat alternatives in PM are ranked by recent ship velocity. Browse the "Tracecat alternatives" section above for the current picks, or visit /alternatives/tracecat for the full list with editorial commentary on each.