Skip to main content

Using Ask Stash

Ask Stash is the AI assistant for inventory questions in plain language. Works for stock levels, sales, reorders, and supplier data. BETA.

Written by Jake
Updated today

Using Ask Stash

Ask Stash is the AI assistant built into Stash. Ask it questions about your inventory in plain language and it returns answers backed by your real data — no SQL, no exports, no spreadsheet wrangling.

Ask Stash is currently in BETA. Some queries may return imperfect results. Feedback shapes future improvements.

Where to find it

Look for the Ask Stash button or chat icon, usually pinned in the bottom-right corner or accessible from the sidebar. Click it to open a chat panel.

Who can use it

Only Admins can use Ask Stash. It has access to inventory, sales, suppliers, POs, transfers, and consumption data — sensitive enough to keep gated. See What Admins vs Members can do.

What kinds of questions work well

Stock and inventory questions

  • "How much milk do I have left at Le Koffee Marais?"

  • "Which items are below their low-stock threshold?"

  • "What's the total inventory value at the warehouse?"

Sales and consumption

  • "What were my top-selling products last week?"

  • "How many cappuccinos did I sell yesterday?"

  • "Which items are consuming the most coffee beans?"

Forecasting and reordering

  • "What should I reorder this week?"

  • "When will I run out of oat milk if consumption stays the same?"

  • "Which suppliers have the longest lead times?"

Supplier and PO questions

  • "Show me all open purchase orders."

  • "Which supplier did I last buy beans from?"

  • "How much did I spend on milk last month?"

What kinds of questions don't work yet

  • Cross-org comparisons ("compare my shops to other Stash users") — not possible, your data only

  • Forward-looking promotions ("predict next week's sales if I run a discount") — Ask Stash doesn't do speculative simulation

  • External data lookups ("what's the market price of oat milk?") — Ask Stash only knows your data, not the wider world

  • Actions ("create a PO for milk") — Ask Stash currently answers questions but doesn't perform actions in BETA

How to phrase questions for best results

  • Be specific. "How much beans?" is vague. "How much espresso beans at Marais?" is better.

  • Use timeframes. "Last 30 days," "this week," "yesterday" all work.

  • Reference real names. Use the exact item, supplier, or shop name as it appears in Stash.

  • Ask one thing at a time. Compound questions confuse the model.

How accurate is it?

Ask Stash translates your question into queries against your live database. For straightforward questions, it's very accurate. For ambiguous ones, it sometimes guesses wrong — always sanity-check the answer against the underlying data when the stakes are high.

If an answer looks wrong, click the thumbs-down to flag it. We use that feedback to tune the system.

Privacy

Ask Stash queries run on your organization's data only. No other Stash customer's data is visible. Conversation history is stored to improve the assistant — let us know if you'd like that disabled for your account.

Related articles

Did this answer your question?