Warehouse supervisors
On the floor, between tasks: ask what is late, what is stuck, who is falling behind — and get the number before the shift meeting starts. The answer arrives while the shift is still running.
The Docklify AI assistant: read-only lookups and reports across OMS, WMS, YMS, TMS, Purchase, and Drayage — just ask in words.


AI Rollie is the Docklify AI assistant: ask a question in plain words, and it answers from the live data of the platform — read-only, across 30+ apps. No report training, no query language, no waiting for someone to build a dashboard: ask how many orders are late, which warehouse is short on stock, or what a customer bought last month, and the answer comes back with the numbers and the records behind them. It reads every app's data and writes nothing, so asking for numbers never touches operations — and every answer, every scheduled report and every Credit it spends is on the record. Long jobs run in the background and report back when they finish, so a big report never holds the conversation hostage.
On the floor, between tasks: ask what is late, what is stuck, who is falling behind — and get the number before the shift meeting starts. The answer arrives while the shift is still running.
One question reaches across orders, warehouse, yard and transport: the morning check is a conversation, not a tour of five dashboards. The numbers are live, not yesterday's export.
Ad-hoc questions without an ad-hoc ticket: pull the numbers for this month, this client, this lane — then ask for the report to prove it. Every figure can be traced to the record behind it.
No training course stands between you and the answer: if you can say the question, you can get the number — in the language you asked it in. The question is the interface; there is nothing else to learn.


Type the question the way you would say it to a colleague — "how many orders are late today?", "which warehouse is short on stock?" — and Rollie does the rest. It works out which apps hold the answer, runs the lookups, and replies in the same language you asked in, with the numbers and the records behind them. There is no query language to learn and no report to build first: the question is the interface. If the question is unclear, it asks what you meant instead of guessing — a wrong number delivered confidently is worse than a question asked twice.


A real question rarely lives in one app: "what did this customer buy last month, is it in stock, and when does the next replenishment arrive?" reaches into CRM, OMS, WMS and Purchase in a single turn. Rollie orchestrates the lookups across the registered tools — one question, several systems, one answer — and shows its steps as it goes, so you can see which system it checked and which one it is still waiting on. If one system times out, the answer says so plainly and tells you which part is missing, instead of quietly pretending the whole picture arrived. The tools are registered per system, so a new app joins the conversation by registering its lookups, not by rebuilding Rollie.


Ask for last month's sales report and Rollie does not just describe it — it builds it. The numbers come from the report engine, so the figures in the chat match the figures in the report center, and the file arrives as a download card in the conversation: CSV for a quick look, PDF and Excel for the meeting. Big reports run in the background instead of blocking the chat, and the finished file is filed in the document center too, so the report is still there when the chat has scrolled away. Every download is logged with the question that asked for it, so a report can always be traced to the person who wanted it.


A conversation is not one question and one answer: ask "and for the Shanghai warehouse?" and Rollie knows you mean the same report, the same month, the same measure — just a different warehouse. Chat history keeps every past question, searchable, and a conversation started in the floating assistant inside one app continues in the main app without starting over. Save a good round of questions as a template, and next month the same check is one click instead of one conversation. Context is carried, not retyped: the follow-up inherits the filters of the question before it.


Say it once in the chat — "send me yesterday's sales summary every morning at 8" — and the task runs itself from then on. Report tasks bind to a real report, so the numbers are always the report center's numbers, and threshold alerts watch a condition and speak up when it breaks, instead of waiting for someone to ask. Before a task is switched on, the wizard shows its cost in plain Credits — one Credit per run, about that many a month at the chosen frequency — and the task expires on a date you set, so no automation runs forever on a forgotten schedule. Tasks nobody opens, tasks that keep failing, and tasks whose monthly pool runs dry all pause themselves and tell their owner why.


The checks worth repeating should not have to be retyped every time. Templates bundle a task into a reusable shape — a report with its parameters, or a prompt skeleton with slots for the warehouse and the time window — and come from three places: platform presets maintained for everyone, shared templates your organization builds together, and personal ones only you see. Start a task from a template and the wizard walks the same path: fill the slots, see the cost, set the expiry, switch it on. A satisfying round of chat can be saved as a template too, so a good question never has to be asked twice. Templates never consume Credits themselves; only the runs they start do.


Rollie reads and never writes: the engine is forbidden from changing business data, so a question can never accidentally cancel an order, move stock or refund a customer. Your data stays inside your tenant — every lookup is filtered by organization on the server, and a tool that returned another tenant's rows would fail outright. Spending is bounded the same way: a complete question costs one Credit, an automated run costs one Credit, and each person has a monthly Credit budget set by the organization admin; when a budget is reached, new AI work stops with a clear message instead of a surprise bill. Costs are shown before a task is switched on, and usage is always visible, by person and by task.
Start asking, get answers.
Past questions, kept and searchable.
One question can pull from 30+ apps.
Daily reports and threshold alerts, running on their own.
Platform, shared and personal templates for repeat checks.
Your Credits by question, task and month — always visible.
Rollie in the corner of every app, with that app's context.
Task results, pauses and budget warnings, in-app and by email.