Type a question the way you'd ask a coworker. Ask writes SQL against the connection you have access to, runs it, and shows the answer as a chart, grid, KPI, map, gauge, or plain text. Every question and answer stays in a conversation you can revisit, export, and share.
Follow a real example — a Top 10 Products YTD question — from typing it in to sending the answer to a coworker.
Open Ask and you land on a chat interface. Click New Chat, pick a connection if you have access to more than one, then type. "Top 10 products by sales YTD". "Which customers haven't ordered in 60 days?". Anything you'd normally ask by walking over to someone's desk.
The sidebar keeps every conversation you've had so you can pick up where you left off. Rename, archive, or delete from a small menu on each row.
Ask sends your question together with the AI Schema you set up in Hook-Tech One — table descriptions, column meanings, business rules, worked examples. That's what keeps the generated SQL grounded in your real data instead of a plausible-sounding hallucination.
The AI comes back with a SQL statement and a suggested component type (chart, grid, KPI, map, gauge, or text). Ask runs the SQL, renders the result, and shows the full query in a collapsible Show SQL panel under the answer so you can inspect exactly what it did.
Ask picks the component type that best matches the shape of the result. Trend over time gets a line or bar chart. A list of records gets a sortable, filterable grid. A single number gets a KPI tile. Rows with latitude / longitude become a map. Progress-toward-target becomes a gauge. Simple yes/no answers come back as text.
Don't like the shape? Ask a follow-up — "show that as a chart", "give me the underlying rows" — and Ask regenerates the answer.
When the answer comes back as a grid you get the full interactive treatment — sortable columns, per-column quick filters, row grouping, and inline formatting. It's the same grid component Hook-Tech Grid uses, so the interaction model is familiar the moment you see it.
Ask uses your Hook-Tech One permissions to decide what data you can even query. Someone without access to a customer's records can't accidentally see them through a cleverly phrased question — Ask never queries what your account can't see directly.
The sidebar lists your recent conversations, newest first. Click one to reopen it — Ask remembers the whole thread including the SQL, the answers, and the AI's reasoning, so a follow-up like "and by branch?" makes sense in context.
A retention setting keeps conversations for the number of days you choose (90 by default). Set it to zero if you'd rather keep every conversation forever. A daily background job handles the cleanup automatically.
Under every answer, an action bar gives you three export buttons. PNG saves a picture of the component. PDF saves a printable version — auto-landscape for wide charts and grids, portrait for KPI and text. Excel saves the underlying rows as a proper .xlsx with typed columns.
Everything runs off the same query Ask already made — no second round-trip to the AI, no risk of the numbers drifting between what you saw and what you exported.
Two more buttons in the action bar hand the answer off to Messenger for delivery. Email opens a small form for recipient, subject, and message; a PDF of the answer rides along as the attachment. MMS takes a phone number and message; a picture of the answer sends as the media.
Delivery uses whichever SMTP and ClickSend accounts your Messenger install already has configured — nothing to set up in Ask itself. The message drops straight into Messenger's queue and typically ships within a few seconds.
Ask was designed for a phone in one hand from day one. The sidebar collapses into a hamburger overlay. The input pins to the bottom of the viewport respecting iOS home-indicator safe area. Text inputs never trigger the iOS zoom-on-focus quirk. Every answer type — chart, grid, KPI, map, gauge — renders sensibly at a narrow width.
And Ask appears on the Hook-Tech One mobile chooser at /m alongside Vision-AI, Grid, and Labeler, so it's one tap from the mobile home screen.
Every question goes to the AI with your AI Schema from Hook-Tech One — table descriptions, column meanings, business rules, worked examples. That's what stops the AI from writing a plausible-sounding query against columns and joins that don't exist in your database. The SQL is grounded in your real schema every single time.
Email or MMS an answer straight from the action bar. Ask hands the artefact off to Messenger, which uses whichever SMTP and ClickSend accounts you've already configured — no separate delivery settings, no copy-and-paste into a mail client. The answer lands in someone's inbox or phone within seconds of you clicking Send.