Help desk features, grouped by the job they do
Latch Workflow is a self-hosted ticketing platform. Every capability below runs inside your own infrastructure, and the queue gets sharper as your operators triage, approve, and correct. Start with the ticket — or the case, if you work in a regulated function — that is already slowing your team down.
Ticket Triage and Routing
Email, forms, and exceptions land in one queue that classifies, prioritises, and routes each ticket. Every correction an operator makes is training signal. Bring your own AI model or use the default.
Approval Workflows
Review steps for refunds, vendor bank-detail changes, account modifications, and other sensitive actions, with role checks and thresholds enforced by the platform rather than by an email thread.
Audit Trail
An append-only record on every ticket: notes, status changes, who approved what, denied attempts, and what the downstream system returned. Exportable for an auditor who asks how you know.
Maker-Checker
Two-person review (also called four-eyes control) on the actions that warrant it. The operator who prepared the request cannot approve it. Role separation is enforced in the platform.
Plugins and Integrations
Write plugins in TypeScript or Go to reach Stripe, your ERP, internal APIs, or your own models. Each plugin action passes through permissions and approvals and lands in the audit trail.
Self-Hosted Deployment
Run on-prem, in private cloud, or air-gapped. Ticket data, AI models, and identity stay inside your boundary, tied to your OIDC provider and your data-residency rules.
Email and Form Intake
Forward a shared inbox or connect a website form and every submission becomes a ticket with an owner, an SLA clock, and a record. No separate triage spreadsheet.
Ticketing API and SDK
REST API, webhooks, and TypeScript and Go SDKs for teams that want to script intake, drive workflow automation from their own services, or embed the queue in an internal tool.
Deployment questions and side-by-side comparisons
Most teams arrive here already running a help desk. These pages cover where the current one stops working and what a self-hosted queue changes.
Self-hosted ticketing
What runs where when the help desk sits on your own infrastructure: on-prem, private cloud, or air-gapped, under your data-residency rules.
Ticketing with approval workflows
How an approval step, a threshold, and a named reviewer attach to a ticket without moving the decision into email or Slack.
Zendesk alternative
Where Zendesk is the stronger choice, and where a cloud help desk stops working for teams that cannot send ticket data off-site.
Jira Service Management alternative
ITSM and issue tracking side by side, including what Jira Service Management does better and what changes when the queue runs in your environment.
ServiceNow alternative
Cost and implementation time for mid-market teams that need governed service desk workflows without a multi-year rollout.
By team
Operations teams
Queue design, escalation policy, and SLA management for support and service operations.
Finance and controls
Reversals, write-offs, KYC and AML review, where the unit of work is a case and two-person review is a requirement.
Platform teams
The build-versus-buy question when the internal tool is already three admin panels and a Retool app.
Pricing
What a self-hosted deployment costs, and why the bill does not scale with every agent you add.
How the pieces fit together
Common questions about how the pieces of Latch work together.
Is this a feature list or one product?
It is one ticketing platform. Triage, routing, approval workflows, maker-checker, the audit trail, and plugins all serve the same job: moving a ticket from intake to resolution without losing context or control. The pages above are entry points, not modules you buy separately.
Where should a team start?
Most teams start by bringing intake into one queue, then add one or two plugin actions behind an approval, such as a Stripe refund or an ERP update. That is a week of work, not a quarter.
Does every queue need approvals?
No. Some tickets need an explicit review before the action runs. Others need the action and the record and nothing more. Latch lets a team decide which actions carry a review step and leave the rest alone.
Can this run on our own infrastructure?
Yes. Latch runs on-prem, in private cloud, or air-gapped. Ticket data, AI models, and identity stay inside your boundary. Teams can start on a managed deployment and self-host later.
Walk the queue before you compare feature tables
The walkthrough follows one ticket end to end: intake, routing, the approval step, the plugin action that runs, and the audit trail it leaves behind.