Status
What is live, what is pilot, what is planned
One dated list of every part of Wetel and where it stands today. Other pages on this site read their status chips from this list, so they cannot disagree with it.
What the chips mean
- Live
Live: running in production and usable today.
- Pilot
Pilot: works on staging or in demo workspaces, or the code is in production but switched off until enabled for a workspace. Not offered as a general production feature yet.
- Planned
Planned: designed or partly built, not usable yet.
Last checked: . This page describes product readiness, not uptime. We do not publish an uptime figure yet.
// Platform
-
Workflow engine
LiveVisual workflows with approvals, retries, per-node time limits, structured output and step-by-step traces. A guard against duplicate side effects is built but switched off.
-
Knowledge bases
LiveUpload documents and an AI employee answers from them. Memory that carries across conversations is not built.
-
Knowledge search step
LiveA workflow step that looks up passages in one knowledge base before an AI step answers, so the answer can use what was found. In production since 2026-11-05, for every workspace.
-
Text helpers in workflow expressions
LiveCondition, switch and set-variable steps can check text (for example whether it contains a word or starts with a prefix) without calling a model. In production since 2026-11-05; existing expressions are unchanged.
-
Workspaces and tenant isolation
LiveEach workspace's data is separated at row level. A dedicated-database option is designed for large tenants but not offered yet. Backup and recovery tooling for it is still in progress.
-
Long-term memory
PlannedRemembering a person across conversations is specified, not built.
// Channels
-
GraphQL API and web chat SDK
LiveBuild your own interface against the same API we use, or embed a chat widget.
-
Telegram
LiveConnect a Telegram bot to an AI employee.
-
Messaging through a partner platform
LiveChannels such as WhatsApp reach an AI employee through a partner messaging platform that connects over our SDK.
-
Smart group messaging
PilotIn Telegram groups an AI employee answers only when mentioned or replied to, with a cooldown and a daily cap. Runs on staging.
-
Voice calls
PilotYou can talk to an AI employee by voice from your browser in pilot setups. The option of a live speech engine and a transcript record of that call are in the production code, switched off until enabled for a workspace. Voice calls themselves are not available in production yet, and response delay is still above our target.
-
Outbound calls from an AI employee
PilotA workspace owner can have an AI employee ring their own phone as a test. Calls to anyone else are refused. The code is in production, switched off until enabled for a workspace; no workspace has it on.
-
Native WhatsApp and Slack connectors
PlannedNot built. Today these channels are reached only through a partner platform.
// Workspace and media
-
Role templates
LiveRole templates are published and listed in the catalogue. The hire gallery that presents them is a demo; after a hire it offers buttons to chat with the new employee or see it in the office.
-
AI Lead Qualifier
PilotAsks a few set questions on a connected chat and writes a short lead summary. Runs in the Sales demo workspace; its documents and rules are yours to set.
-
AI Follow-up Writer
PilotPrepares a follow-up message marked as a draft. Nothing is sent until an owner approves it. Runs in the Sales demo workspace.
-
AI Sales Coach
PilotAnswers a rep's questions from the playbook documents you upload. It reads only what you give it. Runs in the Sales demo workspace.
-
AI Office
PilotA board and an office view of your AI employees, their tasks and what needs a decision. Runs on staging and demo workspaces.
-
Generated files: voice-over, music, explainer video
PilotWorkflows can produce a voice-over, a short jingle or a video script. Video export is recorded in your browser; server-side video rendering is not built.
-
AI Hub demo workspace
PilotA demo workspace for a sample company with several AI employees. A hub on a customer's own domain is not built.
-
Team sign-in, password reset and invites
PilotWorkspace owners and members sign in on a shared host. Verified on staging.
-
Hub on your own domain
PlannedNot built.
-
Delegation with owner approval
LiveA General Manager AI employee plans the work, asks the owner to confirm, then hands it to a colleague that drafts and waits for the owner's approval. Drafts only; nothing is sent. The engine support is in production since 2026-11-05: a delegating workflow reports a colleague waiting for approval, and such a workflow can be published. The General Manager and its colleagues are demo employees on the staging-backed app and are not yet hired into production workspaces.
// Control and trust
-
Weekly owner digest
PilotA plain summary of the last seven days for workspace owners: conversations, runs, approvals, files and most active employees. The code is in production, switched off until enabled for a workspace.
-
Assisted drafts
LiveAn AI employee drafts a reply or a next step and a person reviews and sends it. New AI employees start in drafts-only mode.
-
Owner approvals and member drafts
LiveA step can wait for a workspace owner to approve or reject it. Members draft; owners decide.
-
Request changes on approvals
LiveAn owner can send a draft back with a note instead of approving or rejecting it. The AI employee revises and asks again, up to 3 times. In production since 2026-11-05 through the API; the request-changes button in the workspace app is still on the staging-backed app.
-
Run traces and audit log
LiveEvery run is traced step by step. Sensitive changes are written to an audit log. Spend caps and per-key model limits apply.
-
Auto-approve under a threshold
PlannedLet routine work pass without a click below a limit you set. Not built.
-
Runs alone with a trace
PlannedFull autonomy for chosen tasks, with the trace kept. Not built.
-
Independent penetration test
PlannedNot done yet.
// Commercial
-
Self-serve billing and pricing
PlannedPricing is in design. Usage caps and cost tracking exist; there is no billing yet.
Want to know if your use case is covered today?
Tell us what you want an AI employee to do. We will say plainly what works now and what does not.