System OneENGINE

Find the right item in a small catalog

Exact filters run in code first. Jev picks from the shortlist, checks each item's fit on its own and can report no match.

  • Agent check after deterministic filters
  • Tool sysone_select
  • Recorded examples here. Live runs through your account or engine.

Recorded example: Three-day work trip. The interactive version loads with JavaScript.

Evidence and options

A bag for a three-day work trip by train. I have client meetings, so it should look smart, and I carry a 16-inch laptop.

Exact filters in code: price at most $200, in stock, fits a 16-inch laptop.

  • Metro Sling, sling, 8 L, $79. Filtered out: No laptop sleeve.
  • Commuter Pack, backpack, 20 L, $129. On the shortlist sent to Jev.
  • Convertible Tote, tote, 24 L, $119. On the shortlist sent to Jev.
  • Weekender Duffel, duffel, 40 L, $149. Filtered out: Fits 15 in., needs 16 in..
  • Trail Pack, backpack, 32 L, $159. Filtered out: Fits 15 in., needs 16 in..
  • Carry-on Roller, wheeled case, 38 L, $229. Filtered out: $229 is over $200.
  • Office Brief, briefcase, 12 L, $139. Filtered out: Out of stock.
  • Camera Pack, backpack, 26 L, $189. Filtered out: Fits 14 in., needs 16 in..

Recorded Jev answer, September 24, 2026

  • Choice choice: convertible_tote, 99%
  • Yes/no fit_1: P(yes) 26%
  • Yes/no fit_2: P(yes) 86%

Jev picked Convertible Tote, and its fit check agrees.

Show Convertible Tote with its details

The pick meets your fit rule. Link the product page so the person can confirm size, materials and returns. Nothing was bought.

Measured when recorded: 483 ms in the engine, 1 model call, 873 input tokens. A recorded answer, not a live run.

Exact typed request for this example
{
  "state": {
    "task": "Need: A bag for a three-day work trip by train. I have client meetings, so it should look smart, and I carry a 16-inch laptop.\nCode already applied these exact filters: price at most $200, in stock, fits a 16-inch laptop.",
    "options": [
      {
        "key": "o1",
        "id": "commuter_pack",
        "description": "Backpack, 20 liters. Slim everyday backpack with a water-resistant shell and a sporty look. Laptop sleeve fits up to 16 inches."
      },
      {
        "key": "o2",
        "id": "convertible_tote",
        "description": "Tote, 24 liters. Tote that converts to a backpack. A smart finish that suits client meetings, with a cable organizer. Laptop sleeve fits up to 16 inches."
      }
    ]
  },
  "questions": {
    "choice": {
      "type": "choice",
      "instructions": "Which option best fits the task and its stated constraints? Choose none if no option fits or the facts needed to choose are missing.",
      "criteria": {
        "o1": "Backpack, 20 liters. Slim everyday backpack with a water-resistant shell and a sporty look. Laptop sleeve fits up to 16 inches.",
        "o2": "Tote, 24 liters. Tote that converts to a backpack. A smart finish that suits client meetings, with a cable organizer. Laptop sleeve fits up to 16 inches.",
        "none": "No option fits, or the facts needed to choose are missing"
      }
    },
    "fit_1": {
      "type": "boolean",
      "instructions": "Does option o1 (commuter_pack) directly fit the task and its stated constraints? It is described as: Backpack, 20 liters. Slim everyday backpack with a water-resistant shell and a sporty look. Laptop sleeve fits up to 16 inches. Judge this option on its own and do not assume capabilities that are not described.",
      "criteria": {
        "true": "This option directly fits the task",
        "false": "This option does not fit, or its fit is not established"
      }
    },
    "fit_2": {
      "type": "boolean",
      "instructions": "Does option o2 (convertible_tote) directly fit the task and its stated constraints? It is described as: Tote, 24 liters. Tote that converts to a backpack. A smart finish that suits client meetings, with a cable organizer. Laptop sleeve fits up to 16 inches. Judge this option on its own and do not assume capabilities that are not described.",
      "criteria": {
        "true": "This option directly fits the task",
        "false": "This option does not fit, or its fit is not established"
      }
    }
  }
}

One Choice over the shortlisted items with a none option, plus one independent yes/no fit question per item.

How this experiment works

  1. Code Filter by price, stock, laptop size and capacity with exact comparisons.
  2. Code Send only the shortlist, described in words, with a note naming the filters.
  3. Code Map the returned option key back to the catalog ID before showing it.
  4. Jev One Choice over the shortlisted items with a none option, plus one independent yes/no fit question per item.
  5. You Decide which constraints are exact rules and which are preferences.
  6. You Set the fit probability below which a person compares the shortlist.
  7. You Open the product details before any purchase. The check never buys anything.

Limits

  • Price, stock and size limits are exact, so code applies them.
  • Jev can only choose from the shortlist code sends. It cannot find an item code removed.
  • The choice is relative: some item always ranks first. The fit questions are what can show that every item fits poorly.
  • Nothing is purchased, reserved or added to a cart.
  • Recorded examples are real Jev answers replayed only for an identical request. An edit that changes the request needs a live run.

All experiments and community demos