Changelog

What's new

Selected milestones from the ClevAgent early access build.

v0.05

Gemini CLI added to the onboarding wizard. Pick Claude Code, Codex, or Gemini at first run, and ClevAgent supervises whichever you choose.

v0.04

Pre-launch evidence updated from the current working-session replay: $1,308 estimated savings per 100 real sessions, 44,882 LLM calls analyzed, and 2,441 waste signals caught.
Our Logics expanded with rule-by-rule detection and savings methodology for duplicate reads, stale readbacks, file organization, long files, and memory hygiene.
Workstation page refreshed with current Native App screenshots showing live ClevAgent guidance and optimization logs.

v0.03

Early access begins for the ClevAgent Workstation on macOS. Join the waitlist to try it. Windows support is planned.
First Logic set in active validation: duplicate-read reminders, stale reread detection, file organization guidance, file length guidance, and memory hygiene reminders. See Our Logics for what each one catches.
ClevAgent runs locally in the Workstation and observes only supervised-terminal sessions.
Pricing finalized: Free for one agent and Pro at $29/mo for up to three agents.

v0.02

Rule set v2 specification finalized. Each Logic pairs a detection pattern with a specific corrective message for the agent session.
Test suite built per Logic and replayed across thousands of real agent calls.
Logic quality prioritized over count: live customer testing held until the full set is ready together.

v0.01

Workstation architecture recentered: ClevAgent is a native app where AI agents run, with detection and guidance built in.
Positioning framework finalized: agent efficiency today, with security and audit as future expansion areas.
Focused the product around the in-app Workstation experience instead of external setup flows.

Changelog

Selected milestones from the ClevAgent early access build.

v0.05

Gemini CLI added to the onboarding wizard. Pick Claude Code, Codex, or Gemini at first run, and ClevAgent supervises whichever you choose.

v0.04

Pre-launch evidence updated from the current working-session replay: $1,308 estimated savings per 100 real sessions, 44,882 LLM calls analyzed, and 2,441 waste signals caught.
Our Logics expanded with rule-by-rule detection and savings methodology for duplicate reads, stale readbacks, file organization, long files, and memory hygiene.
Workstation page refreshed with current Native App screenshots showing live ClevAgent guidance and optimization logs.

v0.03

Early access begins for the ClevAgent Workstation on macOS. Join the waitlist to try it. Windows support is planned.
First Logic set in active validation: duplicate-read reminders, stale reread detection, file organization guidance, file length guidance, and memory hygiene reminders. See Our Logics for what each one catches.
ClevAgent runs locally in the Workstation and observes only supervised-terminal sessions.
Pricing finalized: Free for one agent and Pro at $29/mo for up to three agents.

v0.02

Rule set v2 specification finalized. Each Logic pairs a detection pattern with a specific corrective message for the agent session.
Test suite built per Logic and replayed across thousands of real agent calls.
Logic quality prioritized over count: live customer testing held until the full set is ready together.

v0.01

Workstation architecture recentered: ClevAgent is a native app where AI agents run, with detection and guidance built in.
Positioning framework finalized: agent efficiency today, with security and audit as future expansion areas.
Focused the product around the in-app Workstation experience instead of external setup flows.