Jexi

Mentioned, not specified.

Partially answerable. Pricing basis: Jexi explicitly prices per resolution, not per seat — the site states 'Pay per resolution. Not per seat.' and 'Jexi charges for outcomes — resolved tickets, recovered renewals, completed actions. Not for software licences.' The 'What you don't pay for' list explicitly rules out 'Per seat licences,' 'Unanswered tickets,' and 'Platform fees for zero output,' while 'What you pay for' lists 'Tickets autonomously resolved,' 'Renewals recovered automatically,' and 'Actions completed in systems.' So the unit is AI resolutions/outcomes — not seats, and not raw conversation volume. Monthly cost: no dollar amounts are published anywhere in the retrieved site content — no per-resolution rate, no monthly tiers, no plan pricing. The only pricing-adjacent detail is 'Free to start — no credit card.' Overage behavior: the site says nothing about included volume, caps, throttling, or overage billing. Because billing is per resolved outcome rather than a capped plan, an 'included volume' concept may not apply, but Jexi never states this explicitly or publishes overage rates — you would need to email hello@jexi.com for actual numbers.

Mentioned, not specified.

No — Jexi does not currently hold SOC 2 Type II (or Type I). Its own site states 'SOC 2 in progress' in two places on the multilingual-support page: the 'GDPR-ready, EU-hosted option' feature bullet and the data-privacy FAQ ('EU-hosted option available. GDPR-compliant... SOC 2 in progress.'). Because certification is not yet complete, there is no Type II report to review, and none of the retrieved pages — including the Security Overview, Privacy Policy, and Data Processing pages — describe any process for sharing an audit report under NDA. The Security Overview does invite security questions via hello@jexi.com, which would be the channel to ask about the audit timeline and what documentation is available before committing.

Mentioned, not specified.

The site describes handoff only at a high level and does not answer either specific sub-question. On escalation generally: the homepage FAQ says Jexi 'handles the repetitive, high-volume workflows and flags the accounts that need human attention' and 'drafts, flags confidence levels, and waits' (https://jexi.com/, Questions). The AI Use Policy instructs operators to 'escalate low-confidence, unusual, legal, security, refund, complaint, cancellation, abuse, or high-value account matters to a qualified person' (https://jexi.com/ai-policy, Human oversight). The Terms say you are responsible for 'configuring escalation rules' (https://jexi.com/terms, AI output and human review), which implies escalation behavior is configurable — but no page documents a specific confidence-threshold setting, its granularity, or its default. On transcript passing: the homepage says 'Your agents see every conversation, step in whenever needed, and have full account context right there' (https://jexi.com/, An AI coworker), which implies agents have full conversation visibility inside the Jexi helpdesk, but the site never explicitly states that the full conversation transcript is attached or passed during an AI-to-human handoff, nor how the handoff is triggered mechanically (notification, assignment, etc.). A careful buyer can infer that escalation to humans exists and that agents work with full context, but cannot confirm transcript passing or threshold configurability from the site.

Mentioned, not specified.

Partially. On timeline: Jexi does not publish an onboarding duration for migrating an existing help center. The homepage FAQ says only 'We handle the migration with you during onboarding' (in the context of Jexi replacing Zendesk/Freshdesk rather than layering on top), and that setup is otherwise self-serve: 'You can create an account and set up your workspace yourself. Connect an inbox and invite your team during onboarding,' with hands-on help available if you contact them. The only time figure anywhere on the site — 'Most agents are up and running within an hour' — refers to connecting a Gmail inbox on the multilingual-support page, not to help center migration, so it can't be taken as a migration estimate. On article import: the site never mentions importing articles from Zendesk Guide or Confluence. Zendesk appears only in the claim that Jexi 'replaces' Zendesk/Freshdesk; Confluence is not mentioned at all. There is no documented direct article import from either source — a buyer would need to ask Jexi directly (hello@jexi.com) whether help center content can be imported.

Mentioned, not specified.

Partially. On prompt injection: Jexi's abuse playbook page states that 'Reports can hide instructions aimed at AI. The playbook treats every report and web page as untrusted, so the agent reads them but never takes orders from them' — so injection-resistant handling is built into that workflow, and 'Enforcement waits for authorized review by default.' The AI Use Policy also prohibits using Jexi to 'bypass safety controls, approval workflows, rate limits, consent requirements, authentication, authorization, or audit logging,' which implies such controls exist — but the site never describes what actually happens to a customer who prompt-injects or abuses the support bot (no detection mechanics, blocking, or suspension process is documented). On escalation: the Terms of Service state you are responsible for 'configuring escalation rules,' and the AI Use Policy instructs operators to 'escalate low-confidence, unusual, legal, security, refund, complaint, cancellation, abuse, or high-value account matters to a qualified person'; the abuse playbook escalates 'conflicting signals' to a person and notes 'Failed checks are retried or escalated.' However, the site does not document an automatic escalation trigger tied to detected prompt-injection or bot abuse — the escalation described is human-in-the-loop by default, and no admin-configurable auto-escalation on abuse detection is mentioned.

Mentioned, not specified.

Jexi's site never names specific hosting regions or data centers, so 'where is data stored' can only be answered indirectly. What the site does say: (1) For the Gmail/abuse workflow, 'Reports, threads and attachments stay in your Gmail and Google Workspace' and 'Jexi keeps only a short-lived cache while a case is processed. We don't store your data.' (2) The Privacy Policy's International transfers section states 'Jexi and its subprocessors may process information in countries other than where it was collected,' using standard contractual clauses or other lawful transfer bases where required — i.e., processing is not region-locked by default. (3) Encryption at rest applies to customer content 'where stored by Jexi,' implying Jexi-side storage, but with no location disclosed. On EU-only residency: a marketing bullet on the multilingual-support page advertises a 'GDPR-ready, EU-hosted option — Your customer data stays where it should,' suggesting EU hosting exists, but no page explains how to require it, whether it is plan-gated, or whether it covers subprocessors (who per the privacy policy may process data in other countries). The data-processing page notes that a signed DPA controls in conflicts, so EU-only residency would need to be locked into that agreement — but the site alone does not document enforceable EU-only data residency.

Mentioned, not specified.

Jexi advertises '50+ languages, day one' but never publishes an exact count or a complete list of supported languages — the multilingual page only names examples ('From Spanish and French to Japanese, Arabic, Turkish, and Vietnamese'), with German appearing in demos and pilot metrics. On the quality question, the site answers directly and explicitly: 'Same quality, every language' — no language is described as best-effort. The FAQ adds a quality mechanism: Jexi 'flags low-confidence translations and offers alternatives' and 'every reply is reviewed by your agent before sending.' So the uniform-quality claim is stated outright, but the exact number and full language list the buyer asked for appear nowhere on the site — only the '50+' figure, repeated in the page headline and meta description.

Mentioned, not specified.

Partially. (1) DPA before purchase: Jexi offers DPAs by email request — its legal pages state 'Need a signed packet? DPA, order form, security review, vendor packet, or privacy request: hello@jexi.com' — and the Data Processing page notes 'A signed data processing agreement or customer contract controls where it applies or conflicts.' However, the site never explicitly says a DPA can be signed before purchasing; the request channel is public and not stated to be gated on being an existing customer, but pre-purchase execution is not confirmed. (2) Published subprocessor list with advance notice: No. Jexi does not publish a subprocessor list. Its Subprocessors section states: 'Customers with signed agreements may request current subprocessor information and may receive notices or objection rights as stated in their agreement.' In other words, subprocessor details are available only on request to customers with signed agreements, and notice/objection rights are governed by the individual agreement rather than a published advance-notice commitment with a specified notice period.

Mentioned, not specified.

Partially. The site explains how a deletion request is initiated and what happens afterward, but not the mechanics of locating data or any timeline. Process: to stop Jexi's access to Google data and request deletion, a user can (1) disconnect the Google integration in Jexi workspace settings, (2) revoke Jexi's OAuth access from their Google account or Google Workspace admin settings, or (3) 'Contact Jexi at hello@jexi.com and request deletion of Google user data associated with your workspace or account.' What happens next: 'Upon request or termination, Jexi will delete or return Google user data according to the applicable customer agreement, product capabilities, and legal requirements,' and 'Residual copies may remain for a limited period in backups, security logs, audit records, or legally required records' — no specific duration is given for that 'limited period.' For end users specifically, the Privacy Policy says Jexi 'may need to verify the request before responding' and 'may refer the request to the customer because the customer controls that workspace data'; customers are responsible for end-user privacy requests where they act as controller. Not answered: the site never describes HOW Jexi finds the data across its systems (no technical deletion mechanics), and it states no timeframe, SLA, or 'within X days' commitment for completing a deletion request. It also doesn't describe a flow specific to requests submitted via Google's own data deletion process rather than directly to Jexi.

Mentioned, not specified.

Partly. Jexi's homepage states "Free to start — no credit card" and follows with "Create an account, connect an inbox, and watch Jexi resolve the first tickets in minutes" — so yes, you can start without paying, and no, you do not need to enter a credit card. However, the site never states how long the free period lasts, and it doesn't clarify whether "free to start" is a time-limited trial or a permanent free tier. No trial duration, trial-expiry terms, or trial limitations appear anywhere in the retrieved pages (the Terms of Service cover beta features but say nothing about trial length, and the pricing section only mentions pay-per-resolution).