Questions
The questions sales engineers ask
The cross-cutting questions a sales engineer or commercial director accumulates: data handling, missing specifications, scale and import. Each answer restates a fact published elsewhere on the site and links to the page that owns it.
Cite or abstain
Answered, with the page that proves it
Open a question to read the answer. Every figure here already appears on a product, security or pricing page.
- No. There is no training on customer data, and the AI provider is configured for zero retention. Commercial data is competitive data, so it stays yours. The posture is set out on the security page and in the data processing agreement.
- It abstains and asks. When a request does not state a value, Kabaido does not guess one. The line carries a clarification instead, and every value it does extract points back to the customer's own words. You can see both the cited and the abstaining state on the AI page.
- Your primary data and file storage reside in the United Kingdom region of our infrastructure providers. The sub-processors that touch your data are named in the data processing agreement, with the full posture on the security page.
- It still returns a structured specification table of exactly what your customer asked for, with units normalised, abbreviations expanded and ambiguities flagged. Matching and pricing switch on once you import your products. You can try this on the send an RFQ flow before importing a single product.
- Fifty products or a million, the experience is the same: schema-based data, fast faceted search and search engineered to a p95 under 300ms target at 100,000 rows. Catalogue limits run from 250 products on Free to 1,000,000 on Scale. The detail is on the Catalog page.
- Upload a CSV or spreadsheet, map the headers onto your schema, read a type report and run. The ingest is resumable: rows land in 1,000-row batches with the cursor saved per batch, so a stopped million-row import picks up where it left off rather than starting again. The full wizard walkthrough is on the Catalog page.
- A credit is one ordinary resolved request line. Heavier lines weigh more: a service or configured line is two, a configured line that consumed geometry is three. Browsing, manual quoting and your own configurator runs are never metered, because credits meter the AI, not your people. The full weights and worked examples are on what a credit is.
- Every row is organisation-scoped with database-enforced row level security, so the filter lives in the database itself rather than in application code that has to remember to add it. Storage is tenant-prefixed, data is encrypted with TLS in transit and AES-256 at rest, and you can take a full export or request deletion at any time. ISO 27001 and SOC 2 programmes are on the roadmap; the practices we run today are stated plainly on the security page.
- Connect an AI assistant over MCP, script the v1 REST API, receive signed webhooks and post requests in through tokenised endpoints. Each provider is labelled by what it is today: native, a template you finish with your own credentials, via automation through Zapier, Make or n8n, or roadmap. The labelled list is on the integrations page.
Still have a question?
Send the last RFQ you quoted and see Kabaido structure it, or ask us directly.