A four-step wizard: pick the work type, attach documents and libraries, set tags, schedule it with a priority. The form knows the rules and enforces them.

Scheduled work runs without a person driving it. One-click views for Urgent, Overdue, Today, Tomorrow and Last 24 Hours keep the AI’s desk visible.

Each work is a full session: streaming reasoning, the work’s files, citation previews inline, milestones and linked calendar tasks — all in one place.

Create assistant works through the same wizard, converse on the same surface — with the business’s files, libraries and citations in the room.

Personal AI conversations outside any work thread, strictly scoped to their owner, with their own file workspace and reusable skills.

Five personal preset slots pin each user’s favorite assistants directly into the navigation — one click from anywhere.

Create libraries, upload documents, set the priority order that controls what the assistant weighs first. Curated knowledge — not a junk drawer.

Every answer is checked against your documents before you see it. No source in your library, no claim — the AI takes a note instead of inventing.

PDF, Word, Excel, email, images, video and specialized formats — the ingestion pipeline turns what your business receives into usable knowledge.

Multiple accounts side by side, compose and inline reply, drafts and failed-send handling — mail lives where the business data lives.

Everything the business receives crosses one visual board: Queued, Processing, Completed, Failed. Routed to a client, supplier, library or work.

Every automation gets its own drawer of generated documents — paper-style thumbnails, full preview, and a clear status: awaiting your decision, or already sent.

Describe the job in the side panel; the system assembles the flow: one trigger, any mix of data sources, one AI step, up to five actions.

The flow is drawn as connected blocks you can read: no hidden logic, no consultant-only configuration. Change a block, not a project.

Automations run on their own; finished documents flow to Output. Success and failure states are visible directly on the canvas — with retry.

The AI writes from your libraries and past materials — product names right, claims grounded, tone consistent across every campaign and channel.

Lead nurture, reactivation, event follow-ups: multi-step sequences run on their own, personalized from the client record, stopping when the client answers.

Opens, clicks and replies land back in the system — the AI reports what worked and drafts the next round accordingly.

Categories, subcategories, up to fifteen tag slots and per-field naming — the CRM takes the shape of your business in settings, no code.

Client records are context for every AI work: histories, related contacts and notes ground what the AI drafts, schedules and answers.

Every vendor carries an approval status — Approved, Unapproved, Banned — plus commercial terms, contacts and documents.

Shipments pinned on a world map with animated routes, checkpoint history and exception flags — UPS, DHL, FedEx, USPS and more in one view.

What was ordered, what shipped, what remains — per supplier, per order, connected to the same records the AI reads.

SKU-level stock across warehouses with barcode-scanner check in/out, a shipping composer with live carrier rates, and one-click resupply POs.

Each task belongs to a client or a work, so the schedule stays connected to the actual business — click through from the slot to the substance.

Urgent, deadline-reached and behind-schedule tasks collect in visible buckets. Nothing disappears into next week by accident.

Permission-aware to the end: a user who cannot read a task’s client still sees the time slot — the details are redacted by the server.

See conversations as they happen: where visitors are, which channel (chat or voice), what they are doing — in real time.

Every conversation shows what it produced: lead captured, question answered, demo scheduled. The metric is results.

Inbound phone numbers route to configured AI agents, each with its own model and prompt setup, connected to the same work contexts.

A robot chains steps and branches on what it finds: read the document, check it against the record, escalate or proceed — decision logic included.

Apps and robots are configured from the same building blocks as everything else: your work types, your libraries, your rules. If you can describe it, you can assemble it.

Robots do not wait for office hours: intake processed at midnight, checks run before opening, exceptions waiting in the morning briefing.

CRM, accounting, calendar and mail suites connect in settings: map the fields, choose push, pull or two-way, preview with a dry run before anything writes.

Everything the system knows and does is reachable: send data in, trigger work, receive results — with API controls governing exactly who and what.

An MCP server exposes your business knowledge to the AI tools you authorize — the modern integration surface, included.

Tutorials live where the work lives: the AI guides new users through every feature on the real screens, not in a PDF nobody reads.

The bookkeeper, the operator and the manager each learn their own system: role-based onboarding shows each person exactly what their job touches.

When experienced staff leave, their procedures do not: the how-to lives in the libraries, and the assistant teaches it to whoever comes next.

Role-based access over every domain — and the server enforces it: users without rights see redacted details, even inside shared calendars and lists.

Location, department, division, group, team: a five-level tree managed as finder-style columns — permissions follow the structure.

A searchable log of every audited action, with configuration of exactly which categories get recorded. Compliance by design.

Priced per seat, answering mostly from internet training data. Useful for individuals — but grounding, records, scheduling and automations remain YOUR project.

Powerful toolkits — that someone must design, build, test and maintain. The demo is quick; the working system is a project with consultants.

Open models on your own GPUs: full control, and a full-time job — installation, tuning, debugging, updates, all yours. (If you want this, we sell the machine too.)

Your private tenant, hosted: all modules, your industry package, unlimited users. One subscription, live in days. Buy online.

Volume grows? Add the On-Prem Tokens Server: your own GPUs generate the intelligence, your tenant keeps the convenience. Priced by your volume — request a quote.

Private On-Premises AI: the air-tight server + your Zanus applications with permanent licenses. No recurring fees, no tokens, data never touches the internet. Request a quote.