Boutique houses
Skip the data team — ready-made consignor reports and bidder invitations generated per sale, with zero setup.
AI-assisted analytics
Auction-trained predictive analytics
Every customer, sale, lot, and bid in one connected data layer — analysed in your console or any AI assistant your team already uses
Auction Rabbit — Sale insights
Modern Asian Art · Hong Kong
Sale 042 · Insights
Sell-through · vs. high estimate
0%
closed today, 14:30 HKT
09:0010:0011:0012:0013:0014:00
Hammer
$0M
+0.6M
Bidders
0
+212
Est. accuracy
0%
+3%
Suggestion·Lots 042, 047, and 051 show strong early interest. Consider raising reserves by 8–12%
Data domains
Eight auction data domains, one connected schema
The data layer behind every insight — exposed through a Model Context Protocol server so analysts and AI assistants speak the same auction vocabulary
In numbers
Analytics that measurably change sale outcomes
92%
Estimate accuracy within one increment on comparable-rich categories.
+18%
Average sell-through lift on sales that use bidder-targeted invitations.
8 domains
Unified auction entities exposed through one MCP server — no warehouse project required.
How it works
From your raw data to live decisions
- Step 01
Secure ingest
We pull your existing sales history and live signals into your dedicated tenant — encrypted in transit and at rest.
- Step 02
Unified data layer
Every entity is mapped into the eight-domain auction schema and exposed via a Model Context Protocol server.
- Step 03
Insights everywhere
Query from the auctioneer console, your BI tool, or any MCP-compatible AI assistant — no exports, no data leaves your tenant.
Built for
Tuned to how different houses use data
Analysts and heads of strategy self-serve without an IT ticket — the same engine surfaces the numbers each house cares about most
Mid-market analysts
Forecast hammer ranges, identify drop-out risk, and query bidder history directly from the MCP server in your tool of choice.
Enterprise groups
One schema across regions and departments — pipe into your warehouse or expose to your house's own AI agents.
Pairs well with
The insights get sharper when the whole stack feeds them
FAQ
How the analytics layer works
Your own sale history first; we augment with anonymised auction signals across the platform only with your written consent.
Model Context Protocol is an open standard that lets AI assistants and BI tools talk to your data in a structured way. We expose your eight auction data domains through it so any compatible tool can answer questions without exports.
Inside your dedicated tenant in the region you choose (EU, UK, US, or APAC). Nothing is shared cross-customer without your written consent, and audit logs record every query against the MCP server.
Access is role-scoped. Analysts and heads of strategy can query directly; bidder-identifying fields are masked unless your compliance team grants the role.
Models are calibrated per category and we publish confidence scores alongside every prediction, so your specialists know when to lean on them and when to override.
“For the first time we have a number next to every lot that we trust. Our specialists spend less time guessing and more time selling.”
Let's talk
Ready to modernize your auction house?
Book a personalised demo and see Auction Rabbit tailored to your sale calendar