Customer-support agent YundaDesk discloses per-reply audit trace and human approval gate; names no insurance or compliance certification

YundaDesk, an AI customer-support platform, publishes two governance disclosures that remain rare among agent vendors reviewed by AI Dispatch: a per-reply audit trace and a runtime rule requiring human sign-off on financial actions.

"AI answers first on every channel, humans back up, and every reply carries proof of delivery," the company''s site states. Each reply generates a retained record: "Every reply writes a trace: which skill, which knowledge, delivered or not," according to the same page.

On authority limits, the company states that "refunds, compensation and order changes always require human approval," and that "high-risk actions always wait for your confirmation." AI Dispatch''s index scores both disclosures at 2 out of 3 on its runtime-governance and audit-trail dimensions — a named, enforced limit on a specific class of actions, and a per-action retained record — short of the top score because neither disclosure states the trace is tamper-evident or hash-chained, and no spending cap or scope boundary beyond the named action types is published.

The platform operates as multi-tenant, white-label software for agencies managing multiple brands, with "data and config isolated" per client, the company says, and integrates with WhatsApp, Telegram, Messenger, Instagram, LINE, WeChat, WeCom, VK, Zalo, YouTube, a website widget, email and a custom API. Per-customer AI memory is scoped to that customer and, per the company, "does not become an AI skill or general learning." Pricing runs from a free tier through $20-a-month Starter and $200-a-month Pro plans to custom Enterprise pricing, with AI usage credits bundled into the subscription rather than billed per resolution.

The company''s site names no compliance certification or regulatory framework, and no insurance carrier, coverage limit or customer indemnity — AI Dispatch''s index scores both dimensions at zero. It is also silent on outcome-linked settlement: the subscription model bills per plan tier regardless of resolution outcome, the opposite of payment conditioned on a verified result.

YundaDesk''s disclosures on audit trail and human oversight put it ahead of most agents in AI Dispatch''s index on those two dimensions specifically. The remaining three trust-gap dimensions — compliance, insurance and outcome settlement — remain undisclosed, the same gap found across nearly every agent reviewed to date.