rememberoBack home

Let the model propose. Let the gate show its work.

A small agent can draft the move. The request, memory recall, policy rule, and proof chain stay visible while the decision resolves.

Prompt-onlyAPPROVE REFUND
GroundedRefund blocked
Trace modeAnimated
Model runtimeLanguageModel not run
Gate evalmeasuring… · current seeded case
Model generationLanguageModel not run
Proposal contractAwaiting model output
Case
On demand · about 4.0 GB VRAM · weights cached by the browser
Read request

Turn the customer message into structured ticket facts.

Prompt-only

The model drafts an unsupported action from the request alone.

Recall + rules

Remembero returns the facts and active policy rule.

Gate

The deterministic decision decides whether action is allowed.

Request factsRefund ticket T-418 · $249 · identity disputed

Please refund this order today.

  • request(t_418, refund)
  • amount(t_418, 249)
  • disputed_identity(t_418)
Prompt packetSimulator proposal · ungoverned
Instruction: You are a refund triage agent. Return exactly one action token: APPROVE_REFUND or ESCALATE_HUMAN. Do not add explanation.
Customer request: Please refund this order today.
Memory tool result: none
Raw model outputNot run yet
Grounded packet3 recalled facts
Instruction: You are a refund triage agent. Return exactly one action token: APPROVE_REFUND or ESCALATE_HUMAN. Do not add explanation.
Customer request: Please refund this order today.
Memory tool result:
request(t_418, refund)
amount(t_418, 249)
disputed_identity(t_418)
Raw model outputNot run yet
Gate proofRefund blocked
Query
action_blocked(t_418, approve_refund)
Rule
action_blocked(Ticket, approve_refund) :- proposed_action(Ticket, approve_refund), requires_human(Ticket).
  1. action_blocked(t_418, approve_refund)
  2. proposed_action(t_418, approve_refund)
  3. requires_human(t_418)
  4. disputed_identity(t_418)

Awaiting model output

Incoming request

Refund ticket T-418 · $249 · identity disputed

Request
Customer message

Please refund this order today.

Structured facts

  • request(t_418, refund)
  • amount(t_418, 249)
  • disputed_identity(t_418)

Prompt-only agent

Small model · no tools

Proposal
Trace
  1. Build no-memory packet
  2. promptAvailableBrowserModel(packet)
Outcome
APPROVE REFUND

simulated · ungoverned

Grounded agent

Same model · memory + rules

Grounded
Trace
  1. Build memory packet
  2. promptAvailableBrowserModel(packet)
  3. proposal: approve_refund
  4. evaluateGroundedAgentCase(proposal)
Outcome
Escalate to a human.

The model never gets mutation authority.

ProposalPrompt-only agent

The model reads the request and proposes what it wants to do.

GateDeterministic policy

Grounded rules and memory evaluate the proposal and decide: approve, block, or escalate.

ActionAuditable result

Only the gate's decision can produce an action. Every decision is backed by facts and rules you can inspect.

Model proposals use the deterministic simulator (LanguageModel not run). Remembero memory, rule evaluation, and proof are real and run in this browser.