Upsolve AI drops from 16 to 7/36
The pattern: 3 of the 6 website changes in this issue involve MCP (paymaster, KaroX and Phronesis).
What changed: Upsolve AI fell from 16 to 7/36 and is now classified as a Guided Assistant (L1). This is the day’s largest score move; Powerdrill AI also rose from 9 to 14/36.
Why it matters: A nine-point drop changes the comparative case for teams assessing how independently a tool can operate. BrowserAct, LocalAI, and Parall each gained two points, reaching 17/36, 14/36, and 15/36 respectively.
Do this: If Upsolve AI is in an active evaluation, rerun its core workflow against your autonomy and reliability requirements before relying on the earlier 16/36 assessment.
See what changed on Upsolve AI's listing →
Breaking
|
Moveworks updates website to reflect its acquisition by ServiceNow Moveworks updated its site to state that ServiceNow acquired it and describe their joint automation work.
|
What changed on the tools' own sites
|
Paymaster v0.6.0 adds agent-spend controls, MCP tooling, and provider adapters Paymaster expanded its reference implementation with spend controls, a command-line interface, MCP tools, reconciliations, and provider adapters.
Matters if you enforce budgets and reconcile spending by AI agents.
|
|
Wanna replaces its free/$20 offering with four paid plans from $10 to $150 Wanna replaced its previous free and paid offering with a set of paid subscription plans.
Matters if you allocate task capacity across multiple companies.
|
|
KaroX 5 beta advances with new quickstart, Git push, durable job, and MCP… KaroX made its latest beta line primary and added runtime capabilities, including quickstart, Git, job, and MCP updates.
Matters if you test beta runtimes with MCP integrations and Git-based workflows.
|
|
Zencoder repositions around Zenflow, an AI orchestration platform for code… Zencoder now presents Zenflow as an orchestration platform for code work rather than only a coding assistant.
Matters if you coordinate AI-assisted code work across development workflows.
|
|
Phronesis repositions as an agentic decision-assurance layer with MCP support Phronesis now focuses on auditable decision assurance before an agent acts and adds MCP support.
Matters if you oversee controls on automated agent decisions.
|
Changes we spotted on each tool's own website since the last issue.
Score moves
Source: agentic.ai's rubric re-evaluations.
|