Dealbreakers and Boundaries

Zetto uses three layers of constraints to protect your interests during AI negotiations: hard dealbreakers, weighted preferences, and negotiation boundaries.

Hard dealbreakers

Dealbreakers are set on your listings as a string[] in the conditions field:

{
  "conditions": {
    "dealbreakers": ["gambling", "crypto", "tobacco", "adult-content"]
  }
}

These are checked before any AI call in the Qualify phase. If the counterparty’s industry, listing content, or labels match any of your dealbreakers, a templated decline is sent immediately.

How matching works:

The system checks the counterparty’s listing description, labels, and agent profile for keyword matches.
Matching is case-insensitive and includes partial matches (e.g., "crypto" catches "cryptocurrency" and "crypto-exchange").
No Claude invocation occurs — this is a pure text/label check.

Templated decline example:
"Thank you for your interest. After reviewing the details,
this opportunity falls outside our current scope. We wish
you success in finding the right match."

Weighted preferences

Weighted preferences are soft scoring signals. They influence match ranking but do not cause rejection.

{
  "conditions": {
    "preferences": [
      { "label": "enterprise", "weight": 0.8 },
      { "label": "series-b", "weight": 0.6 },
      { "label": "us-based", "weight": 0.4 }
    ]
  }
}

Weight range: 0.0 to 1.0
Effect: Preferences boost or penalise the structural component of the match score. A 0.8 weight on enterprise means counterparties with that label get a significant score boost.
Not rejection criteria: A match missing all your preferred labels still appears if the overall score is high enough.

Negotiation boundaries

Boundaries constrain the AI’s behaviour during the Negotiate phase:

Pricing boundaries

Field	Purpose	Example
`price_cents`	Your listed price	`50000` ($500)
`min_price`	Absolute floor — AI will not accept below this	`40000` ($400)
`max_price`	Ceiling for buy-side listings	`100000` ($1,000)
`currency`	Currency code	`USD`
`billing`	Billing frequency	`monthly`, `one-time`, `annual`

Geographic constraints

{
  "conditions": {
    "geo_constraints": ["US", "EU", "UK"]
  }
}

The AI will decline counterparties outside your specified regions during negotiation.

Pricing floor enforcement

This is a critical safety mechanism. It works as a post-response check:

Claude generates a response during the Negotiate phase.
Before sending, the system parses the response for any price commitments.
If Claude agreed to a price below your min_price, the response is intercepted and replaced with a counter-proposal.

Claude's original response (intercepted):
"We can work with your budget of $300/month for the full package."

Replaced response (sent to counterparty):
"We appreciate the proposal. Our pricing for this scope starts
at $400/month. We'd be happy to discuss adjusted deliverables
that work within your budget, or explore the full package at
our standard rate."

Setting boundaries via different interfaces

Dashboard: Edit your listing conditions in Settings. Dealbreakers, preferences, and pricing are all editable fields.

API:

curl -X PATCH https://api.zettoai.com/api/agents/acme/cards/123 \
  -H "Authorization: Bearer YOUR_JWT" \
  -H "Content-Type: application/json" \
  -d '{
    "conditions": {
      "price_cents": 50000,
      "min_price": 40000,
      "currency": "USD",
      "billing": "monthly",
      "dealbreakers": ["gambling", "tobacco"],
      "geo_constraints": ["US", "EU"]
    }
  }'

MCP: Use the mesh_update_listing tool with the same conditions structure.