MCP server · ChatGPT · Claude · any assistant
The memory and math your AI assistant is missing
Weighted, persistent product comparisons for considered purchases. Deterministic scores, the formula in plain sight, and the same answer next week — in any chat, on any device.
Live demo — drag the weights
Noise-cancelling headphones
saved · cmp_x7f3aSony WH-1000XM6 leads with 8.81
Ranked for youSnd 8.8 · Cmf 8.2 · ANC 9.4 · Bat 9.0
Snd 8.5 · Cmf 9.3 · ANC 9.0 · Bat 7.5
Snd 9.0 · Cmf 7.4 · ANC 8.8 · Bat 5.5
winner’s math (9×8.8 + 8×8.2 + 7×9.4 + 5×9.0) ÷ 29 = 8.81
01 / How it works
Three steps, zero spreadsheets
Describe your purchase
“I’m choosing between three robot vacuums. What matters: suction, noise, price, and app quality.”
Your assistant builds it
It creates the comparison with your weighted attributes, researches each candidate, and scores them against absolute anchors.
Get a ranked answer — anytime
See the weighted ranking, refine scores as you learn more, and reopen the same comparison in any future conversation.
02 / Why Komparable
Assistants research. Komparable remembers.
AI assistants are great researchers and terrible bookkeepers. Komparable keeps the state and does the math, so the model can do what it does best.
Your attributes, your weights
Define exactly what matters — battery life, comfort, warranty, resale value. Weight each from 0 to 10. The ranking reflects your priorities, not an average of everyone's.
Deterministic, transparent math
Scores are stored and the weighted ranking is computed server-side — the same answer every time you ask, with the formula in plain sight. No regenerated, vibes-based verdicts.
Persistent across conversations
Considered purchases take weeks and many sessions. Your comparison lives under a permanent ID — reopen it next week, on your phone, in a brand-new chat.
Built for AI assistants
An MCP server any assistant can drive. ChatGPT or Claude researches products, proposes criteria, and fills in scores — you just describe what you're shopping for.
03 / Get started
Connect it in two minutes
Komparable is an MCP server — point your assistant at the endpoint, then ask it to set up your first comparison.
MCP endpoint
https://komparable.com/mcp
REST API / OpenAPI spec
https://komparable.com/openapi.yaml
In ChatGPT
Settings → Apps & Connectors → Developer mode → add the MCP endpoint above.
In Claude
Settings → Connectors → Add custom connector → paste the MCP endpoint.