Inbox on autopilot.
An IMAP/SMTP mailbox queue: the AI classifies, drafts and routes — nothing sends until a human approves it in the queue.
Features
Everything an AI needs to run real operations — safely.
The DNA is pull-based work — Pull Queues. Work items flow through workflows made of queues, routing, rules and forms. On top of that: auto-enrichment, multi-tenant orgs, plans, project logic, and a local agent for the risky parts — with every step tracked and a human approving anything that leaves the building.
What you can automate
Six shapes of office work, one platform underneath
Each one runs on the same Pull Queues, forms and Golden Rule described below — just pointed at a different job.
Ask your database in plain English.
SQL/ODBC query-enrichment queues run a saved query against your data and attach the result to the work item automatically.
Scrape it, structure it, done.
Point a workflow at pages or feeds you define; the AI gathers what's there and structures it into your form fields.
Data entry problems… solved.
The local agent drives the target application's UI directly — clicks and keystrokes, the way a person would.
Documents that fill their own forms.
Drop a PDF, scan or email attachment on a queue; AI extraction fills the workflow's form, ready for review.
Any data, any source, captured.
Define the fields that matter on your form; the AI populates them from whatever source you connect.
Pull Queues
Pull-based work. A person — or the AI — asks for the next item and gets exactly one, leased and locked. No cherry-picking, no dropped work, natural load-balancing across a team.
Workflows, forms & rules
Compose queues into a process with routing and business rules. WYSIWYG forms capture structured data; validation and conditional logic keep every work item clean and complete.
Query-enrichment queues
Auto data-gathering. A dedicated queue runs saved queries and data-source lookups against a work item, attaching the context the AI needs to decide — before a human ever sees it.
Organizations & teams
Multi-tenant from the ground up. Invite members, scope roles, and isolate each organization’s workflows, queues and data. Bring a team; keep everyone in their lane with RBAC.
Plans & tiers
Quota enforcement built into the seam. Free starts at 5 workflows and 100 work items; upgrade for headroom, teams, and enterprise controls without changing how anything works.
Project management Coming soon
A project layer over your queues: dependencies, milestones and critical-path scheduling so the AI can manage by project logic — not just work items one at a time.
The local edge agent
The risky part runs on your machine.
A downloadable, open-source desktop agent executes anything that touches an outside system — using credentials that never leave your device. The cloud orchestrates; the edge acts.
- Credentials stay in your OS keychain, encrypted at rest
- External calls originate from your network, not a shared cloud IP
- Open-source and auditable — inspect exactly what it does
- Same agent for free-SaaS and self-hosted deployments
Cloud orchestrates · edge acts
The control plane never holds a plaintext secret.
Flagship use case
AI document → structured form → routed work item
Drop in a document. The AI ingests it, fills the workflow’s form, enriches it from your data sources, routes it to the right queue — and waits for a human before anything leaves the building.
1
Ingest
The AI reads the document and extracts fields, entities and intent.
2
Enrich
Query-enrichment queues attach related records and context automatically.
3
Route
Rules place the work item in the correct queue with the right priority.
4
Approve
A human confirms; only then does the external action fire.
Build your first workflow free.
Five workflows and 100 work items, no credit card. Add the local agent when you’re ready to reach outside systems safely.