Ask your PIM catalog. In plain English.
AI Copilot translates natural-language questions about your product catalog into safe database queries, executes them, and answers in plain text. Accessible to your whole team — no SQL, no filters, no training.
How it works
Natural language → SQL → answer
Type a question in plain English. The Copilot interprets intent, generates a safe SQL query against your product catalog, executes it, and returns the answer in natural language. No training required.
SQL injection protection
Generated queries are parsed and validated against an allowlist before execution. Read-only by default — Copilot can never DROP, DELETE or UPDATE. Even if a user asks "delete all products", it will refuse.
Suggestion chips
Pre-built question prompts adapted to your catalog: completeness audits, missing images, compliance issues, supplier-side imports. Click and refine — no blank-page anxiety.
Aggregations & comparisons
"Top 20 most expensive products without a main image", "Suppliers with import errors this month", "Completeness progress per category". Complex queries that would take 10 minutes in SQL — 5 seconds in Copilot.
BYOK on any LLM
Assign the AI Copilot context to any of the 14 configurable AI providers (OpenAI GPT-4o, Anthropic Claude, Gemini, Mistral, local Ollama…). Use your own API keys — transparent costs, no markup.
Accessible to your whole team
Buyers, category managers, content writers, support — anyone with a question about the catalog can ask. No SQL knowledge, no training, no waiting for a developer. The catalog speaks human.
Real questions you can ask
« How many products are missing a description in the Electronics category? »
« Which products have a completeness score below 60%? »
« Show me the 20 most expensive products without a main image. »
« Which suppliers delivered imports with errors this month? »
« List REACH non-compliant products currently in stock. »
« What is my completeness progression per category? »
Why natural language beats filters for ad-hoc questions
Traditional PIMs answer queries through filter interfaces: choose a category, tick three checkboxes, hope the report you need is in the dropdown. It works for the 5 questions you ask weekly. It collapses for the 50 questions you ask once a quarter — the audit that an exec drops on your desk, the compliance request from legal, the "what would happen if…" exploration before a category review.
For these ad-hoc questions, building a filter UI is the wrong abstraction. The right abstraction is the language you already think in. AI Copilot reads "Show me products with high margin and low stock that we discontinued last year" and turns it into the SQL JOIN you would have spent 15 minutes writing — if you knew SQL, which most catalog managers don't.
The point isn't to replace dashboards or reports. It's to remove the friction between an unexpected question and its answer — so the question gets asked instead of postponed.
AI Copilot is included in every paid plan
No add-on, no per-seat fee. Your team asks, the Copilot answers — from Starter (€490/month) to Enterprise.