AI Decision Matrix & Option Comparison

Give it a decision, your options, and your criteria and get a ranked comparison, a recommendation with confidence, and the evidence that would change it.

Best for: decision analysis, compare options, evidence gaps.
Input and output: The output includes comparison criteria, a recommendation, ranking, and evidence gaps.

What can Decision Framer do?

State the choice, known context, candidate options, and priorities. The Actor asks Liminality to frame the decision and only charges when the engine returns a certified result.

What you get	Features
⚖️ Criteria and ranked candidates	📦 Batch input with one result row per decision
🧾 A recommendation, confidence, and facts that could change it	🛑 Uncertified results fail closed and are not charged

Who this is for

• Compare build and buy options
• Frame a vendor or tool choice
• Record what evidence could reverse a recommendation

What you get back

Field	Type	What you get	Example
index	integer	Position of this item in the input list.	1
status	string	Whether this item completed its requested analysis.	success
decision	string	Decision that the Actor compared.	Should our SaaS company build or buy customer support automation?
criteria	array	Priorities used to compare options.	["Launch speed","Total cost","EU data handling","Maintenance burden"]
rankedCandidates	array	Options ordered with their supporting rationale.	[{"name":"Build in-house","basis":"In-house build offers full control and EU compliance but is slow and expensive. Requires 6-14 weeks and $30k-$150k, which...
recommendation	string	Suggested option and supporting reasoning.	Build in-house (score 0.690, ungrounded (no registry match)). Commit now; validate the speculative inputs before the larger bet. Risks: The primary risks of...
whatWouldFlipIt	array	Evidence or conditions that could change the recommendation.	["Which specific EU-compliant platform is currently under consideration?","Are there any existing contracts or commitments with a specific platform?","Build...
analysis	string	Full written comparison of the options.	Build in-house. Commit now; validate the speculative inputs before the larger bet. Risks: The primary risks of building in-house include potential delays due...
confidence	string	Confidence level for the qualification.	candidate
error	null	Problem details when this item does not complete.	null
processedAt	string	Time the Actor processed this item.	2026-09-11T04:23:59.590246+00:00
confidenceNote	string	Reason the confidence level was assigned.	Ranked and reasoned, but the engine could not verify every factual claim against evidence. Treat as a structured starting point, not a final verdict.

The run also links to its dataset and any files named in the Actor output.

What you need to provide

Field	Type	Required	What it does	Example
items	array	Yes	Process one or more items in a single run. Each item produces exactly one dataset row.	[{"decision":"Should our SaaS company build or buy customer support automation?","conte...
items[].decision	string	Yes	The choice or decision to frame.	Should our SaaS company build or buy customer support automation?
items[].context	string	No	Known facts, constraints, risks, and assumptions for this decision.	12-person company; $20k budget; launch in 8 weeks; two engineers; EU customers.
items[].candidates	array	No	Optional candidate options to compare.	["Build in-house","Buy a managed platform"]
items[].priorities	array	No	Optional criteria in preferred priority order.	["Launch speed","Total cost","EU data handling","Maintenance burden"]

Quick start

1. Open the Actor in Apify Console.
2. Click Try for free or Create a task.
3. Replace the sample values with your own input.
4. Click Start.
5. Open the dataset and the named output files when the run ends.

Pricing

• decision-framed: $0.05 per decision framed.
• Example: 10 decisions cost at most $0.50 (up to 50 per run); a decision the engine cannot ground returns an explanation row and costs nothing.
• You pay only for successful results. Failed or skipped items are not charged.
• Normal Apify compute and proxy costs may also apply.

Limits and honest notes

• Only certified engine results are returned as successful decision frames.
• Candidates and priorities are optional, but missing detail may leave the decision unresolved.
• A rejected result returns an error row and does not charge decision-framed.

Code and API

The examples below use the same values as the Apify Console sample.

Input JSON

{
  "items": [
    {
      "decision": "Should our SaaS company build or buy customer support automation?",
      "context": "12-person company; $20k budget; launch in 8 weeks; two engineers; EU customers.",
      "candidates": [
        "Build in-house",
        "Buy a managed platform"
      ],
      "priorities": [
        "Launch speed",
        "Total cost",
        "EU data handling",
        "Maintenance burden"
      ]
    }
  ]
}

Real output sample

This excerpt comes from the real run named in the current marketplace release report.

{
  "index": 1,
  "status": "error",
  "error": "Decision frame rejected by the engine: Liminality could not verify enough evidence for this answer. Add the requested facts or sources, then try again.",
  "processedAt": "2026-09-11T04:30:29.622669+00:00"
}

curl

curl -X POST "https://api.apify.com/v2/acts/physealabs~decision-framer/runs?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d @input.json

Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("physealabs/decision-framer").call(run_input={'items': [{'decision': 'Should our SaaS company build or buy customer support automation?', 'context': '12-person company; $20k budget; launch in 8 weeks; two engineers; EU customers.', 'candidates': ['Build in-house', 'Buy a managed platform'], 'priorities': ['Launch speed', 'Total cost', 'EU data handling', 'Maintenance burden']}]})
items = client.dataset(run["defaultDatasetId"]).list_items().items

Node.js

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('physealabs/decision-framer').call({"items": [{"decision": "Should our SaaS company build or buy customer support automation?", "context": "12-person company; $20k budget; launch in 8 weeks; two engineers; EU customers.", "candidates": ["Build in-house", "Buy a managed platform"], "priorities": ["Launch speed", "Total cost", "EU data handling", "Maintenance burden"]}]});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

You can call this Actor from an agent or LLM tool that can send HTTP requests to the Apify API. Keep the Apify token in a secret store.

FAQ

Are candidate options required?

No. You can omit them, but named candidates make the comparison more specific.

Can I set the criteria order?

Yes. Put the criteria in order in priorities.

What happens to an uncertified result?

The Actor returns an error row, omits the candidate recommendation, and does not charge the event.