88Intelligence Index
STEP 1 OF 2

Tell the Fabric what you need

Just describe a business outcome in your own words. You do not need to know which AI model to use, how to write a prompt, or any technical details. The platform handles all of that automatically based on your organisation's profile.

WHAT TO DO HERE
  1. Read the example request in the text box below, or delete it and type your own — e.g. "Summarise the key risks for the board meeting" or "Draft a client update on project status".
  2. Click EXECUTE OPTIMIZED → and wait a few seconds.
  3. Scroll down to see your result and the claim verification ledger.
Why this matters: You describe the outcome. The Fabric figures out the task type, selects the right AI strategy, enforces quality and evidence standards, executes, verifies claims, and records everything. That is the entire workflow.
Intent → Compile → Execute → Prove → Learn

Stop choosing models. Start commanding outcomes.

Describe the business outcome. The Fabric converts it into an Intent Contract, the right system/context/user stack, a model portfolio, verification controls, and an execution route optimized for business quality per dollar.

88/100
Enterprise Intelligence Index

Composite of prompt health, expected quality, truth defense, and cognitive efficiency for the current task.

01 / BUSINESS INTENT

What needs to happen?

AUTO COMPILER
INTENT INPUT26 words
83Intent
73Context
40Output
84Evidence
90Economy
75Health
02 / INTENT CONTRACT

What the enterprise understood

high risk
TaskVerification
Audienceexecutive
Riskhigh
Quality target96
Truth target98
Cost ceiling$0.350
Latency target< 8s
FunctionEnterprise PMO
AuthorityHuman approval before external or material action

Compiled intelligence stack

Constitution → situation → mission → execution

03 / PROMPT ARCHITECTURE

System × Context × User

RE-ENGINEERED
SYSTEMConstitution

Operate as an evidence-first enterprise analyst. Ground material claims in approved evidence, surface conflicts, distinguish fact from inference, and abstain when evidence is insufficient. Format your response in markdown: use ## for section headings, **bold** for key terms, bullet lists for action items, and > blockquotes for critical warnings or caveats.

CONTEXTSituation

Organization: Meridian Global. Industry: Financial Services. Function: Enterprise PMO. Data classification: confidential. Audience: executive. Use only context necessary to satisfy the request.

USERMission

Prepare tomorrow's Steering Committee update. Identify the decisions required, the material risks, and anything likely to move the delivery date. Use current approved project evidence only.. Return a decision-ready verification for a executive audience. Prioritize the information that changes action or judgment. Cite or identify support for every material factual claim; flag missing or conflicting evidence rather than guessing. Keep the response concise and do not repeat source material that does not affect the outcome.

04 / EXECUTION PLAN

Balanced

RECOMMENDED ROUTE
Extract
Economy model

Compress relevant facts before reasoning

Analyze
Balanced model

Decision-grade synthesis

Verify
Verifier

Check material claims and conflicts

96Quality
98Truth
$0.082Cost
4.4sLatency
81%Efficiency

Intelligence efficient frontier

Highest business value at the lowest adequate intelligence cost

Economy

90Quality
93Truth
$0.028Cost
1.9sLatency

Fast, low-cost execution for well-defined or low-materiality work.

Maximum Quality

98Quality
99Truth
$0.310Cost
12.6sLatency

Use when complexity, ambiguity, or materiality justifies premium reasoning and verification.

Compiler impact

Estimated against the unoptimized request

+16%Quality upliftBetter task architecture
58%AI costRight-size the intelligence
31%LatencyReduce avoidable work
+8%Evidence disciplineSource or silence