01Call log

EVERY CALL, accounted for.

when	call	use case	turns	v2v p50	worst	llm	judge	errors	ended

02Call detail

Live transcript · redacted

Latency waterfall (per turn, ms)

endpoint llm tts

Judge verdict

Provider errors

Config

03Fleet stats

DID THE LAST DEPLOY move p95?

group	calls	turns	endpoint p50/p95	llm p50/p95	tts p50/p95	total p50/p95	llm>2s	llm>5s	errors	judge

05Agent launcher

THE ROSTER, live-editable.

＋ New agent

Create = a valid starter config to edit by hand. Create with Bonham = an interview that writes the config with you (only the slug is required — Bonham asks the rest).

06Bonham create-chat

TALK IT into existence.

Bonham · onboarding interview

📎 Corpus

Draft config valid

Single source of truth — hand-edit freely; Bonham sees your edits next turn. Schema-validated on every Bonham turn and on Finalize.

04Place a test call

RING THE agent.

Number (E.164)

Who answers (first name)

Use case

Stage (optional — which caller journey)

Quick test cases (same number)

Duplicate-call suppression is bypassed with a unique key per click. The record appears in Calls when the call ends.

BONHAMmission control.