Skip to content

Application Studio

How review-first auto-submit works

Review-first auto-submit is the most restricted feature in the product. It is gated by a three-layer check before any submission attempt is allowed. The full gate lives in src/lib/applyworth/submission-gate.ts and the logic is short enough to audit directly (lines 14-48).

The gate evaluates four conditions in order. Any failure short-circuits and returns a typed error response:

1. Feature enabled. If the autonomous submission feature flag is off for the environment, the gate returns 503 not_configured. There is no override. 2. Plan = Max. Any plan other than Max returns 403 max_required. Sprint, Pro, and Free cannot reach the submission path. 3. Explicit user confirmation. The request must carry an explicit user confirmation flag for this specific submission. Missing confirmation returns 403 confirmation_required. There is no 'submit on my behalf' default and there is no remembered confirmation across runs. 4. Kit status = approved. The Application Kit attached to the submission must be in approved status. Anything else - draft, generating, awaiting review, regenerating - returns 409 kit_not_approved.

Only when all four conditions pass does the gate return allowed: true. The submission attempt then proceeds against the host adapter.

Supported ATS hosts: Greenhouse, Lever, and Ashby. These are the only hosts the product will attempt review-first auto-submit against, and only because they have stable form schemas and have been fixture-proven.

Manual-required hosts: Workday, LinkedIn, iCIMS, SmartRecruiters, and any custom or non-standard application form. These return a manual_required state. The product will not pretend it submitted; it will stop, preserve the kit, and hand the application back to you to complete by hand. This is intentional: silently failing on an unsupported host would leave you thinking an application was sent when it was not.

What this is not: this is not 'apply to everything overnight'. The gate exists because unattended autonomous final submission is a category of action with real downside (wrong applications sent under your name, host TOS issues, ATS rate-limiting against your account, no audit trail). The review-first design ensures every submission is one you specifically approved, for a kit you specifically approved, on a host the product has proven it can submit to.

NextGenerate application kits

Product proof

See the product surface behind the claim.

Each page carries the matching RoleWorth surface in a glass-framed proof card: the radar, extension overlay, ATS matrix, review queue, dashboard, or package flow behind the promise.

Core promise
RoleWorth command center dashboard showing today's radar, decision queue, and audit feed

The public promise, visible above the fold: score first, package second, approve before anything leaves.

Command center
RoleWorth command center dashboard showing radar metrics, pipeline health, and audit feed

The internal cockpit: today's radar, active runs, best opportunities, pipeline health, and audit history.