docs: build GPT-Live guides around conversation quality

[dev-seth-vapi]

Description

Rebuild GPT-Live as seven guides centered on conversation quality: useful waits, concurrent speaker/reasoner work, corrections, personality, and caller outcomes. Introduce the audio model and the two roles, then cover a first call, tools, migration from a single assistant, testing, and settings. Preserve links from the existing configuration and limitations pages.

Build with Tools explains the decisions with small inline examples and no downloadable project or local-server exercise. Idle-message examples use say.prompt; generated speech is never described as verbatim.

Reasoner skills and the squad migration recipe that depends on them are excluded from this release because the feature is not deployed. Their complete draft is preserved separately. The migration page states the current squad limitations.

Written with local Claude Opus 5.5 and independently reviewed by Codex. No runtime changes or migrations.

Validation

• Fern local validation: zero errors and 11 unchanged API schema warnings.
• Internal GPT-Live links and anchors, remaining JSON examples, and absence of skills/download references checked.
• Earlier desktop and mobile rendering checked; Preview Docs builds the final revision.
• Earlier isolated live verification covered quickstart/endCall, delayed lookup with a policy detour and confirmed booking, and a date correction during lookup. The evidence retains an initial failed confirmation run and its corrected repeat. These calls support the guidance; they are not a reliability benchmark.

Verification report and call audio.

Verification used fictional data. Phone transfers, every voice/language, Composer edits, Voice Simulations, and production billing were not live-tested.

[github-actions]

🌿 Preview your docs: https://vapi-preview-01a0f467-dd1c-76bf-81ba-e40f64735479.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://vapi-preview-01a0f496-88b2-742c-9f8c-fd62c2fa8f11.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://vapi-preview-01a0f4ab-23f0-703e-9ad1-37ade46ea5c0.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://vapi-preview-01a0f4bb-4285-750c-8e3f-65f7ed6ea809.docs.buildwithfern.com

[lightsage-app]

Lightsage docs evals

Lightsage could not queue docs evals for this PR.

Docs URL: https://vapi-preview-01a0f4d2-4e71-7189-ad50-255c1cd5b22b.docs.buildwithfern.com
Commit: 8d8f270
Reason: No custom evals are selected for GitHub PR evals. Open Lightsage > Custom Evals and enable the GitHub checkbox for at least one custom eval.

[github-actions]

🌿 Preview your docs: https://vapi-preview-01a0f4d2-4e71-7189-ad50-255c1cd5b22b.docs.buildwithfern.com