:: Value API for AI Agent Output Evaluation
: What Is Value?
We use the word constantly. We talk about creating value, adding value, value for money, valuable work, value alignment,
value creation in business, technology and society etc. Yet when asked precisely what value is, the answer is rarely clear.
- Is it profit?
- Efficiency?
- More freedom?
- Greater understanding?
- Better outcomes, such as being cheaper; more for less ≠ less for more?
In practice, value can be advertised in advance, yet recognised after a transaction if a preferred change or relationship shift occurred and something improved compared to how it was before:
- A decision made something clearer,
- A tool reduced friction,
- Better outcomes, such being lower cost?
- A process became fairer,
- A solution solved the real problem.
In other words, 'value' appears as a measurable improvement between two states.
But is that definition enough?
For instance, is value not a weighting behind meaning?
Maenen Labs reflects deeply to produce all core primitives and the Value API is built around a deeper insight than a somewhat superficial transaction imbalance or extraction measurement definition.
AI outputs are easy to produce. Improvement is harder to prove.
: The Problem
Modern systems produce enormous amounts of output. But it's often unclear:
- Whether the output actually improved anything?
- Whether a seemingly good automated action aligned with the original intent?
- Whether an agent made the situation better or worse?
Tool reports can present usage, speed, or cost — but not the outcome value.
Without an effective way to measure improvement, systems simply cannot learn which actions are truly beneficial.
: The Value Solution
Maenen Value API introduces a deterministic way to evaluate improvement via a deeper framework.
Every evaluation produces:
- A delta across key dimensions.
- An overall value score.
- A signed receipt.
- Optional ledger anchoring.
And because the evaluation is deterministic, the same input always produces the same result. This makes value itself auditable, portable, and verifiable.
: Why Value Matters
For developers and autonomous systems alike, this creates a new capability — systems can now:
- Measure whether an action truly improved a situation.
- Compare different solutions objectively.
- Verify claims of improvement.
- Track value over time.
- Anchor results to a ledger for later review.
➤ In other words, systems gain the ability to learn from outcomes, not just outputs.
: Compare Value API
 |
| Technique
| Smaller |
Reusable |
Deterministic |
Auditable |
| |
|
| Raw Prompting |
✅ |
❌ |
❌ |
❌ |
| Summarise |
✅ |
❌ |
❌ |
❌ |
| Heuristic Structuring |
⚠️ |
⚠️ |
⚠️ |
⚠️ |
| WaveInform™ |
✅ |
✅ |
✅ |
✅ |
|
: Example Use Cases
AI agent workflows
|
Evaluate whether an agent's actions genuinely improved a task outcome before accepting its results.
|
Decision systems
|
Compare multiple candidate solutions and identify which one produced the best outcome.
|
Process optimisation
|
Track whether changes in a workflow actually increased clarity, fairness, or efficiency.
|
Swarm systems
|
Allow groups of agents to compete or collaborate while measuring which path produced the best outcome.
|
|
|
Endpoints (click to copy )
Measure coherence, policy alignment, and whether outputs tangibly improve or degrade meaning:
|
|
|
|
FREE Gated Proof-layer |
|
|
|
FREE SERVICES
|
GET /v1/value/baseline/latest |
GET /v1/value/baseline/changelog |
GET /v1/value/baseline/{version} |
|
Value API
Measure whether the result improved the situation
|
Integrity API
Verify that meaning survived the transformations |
This≠That API
Compare meaning between documents |
|
: Support
For technical issues or questions, use the contact options below.
Full support and service-level agreements are not offered during the beta trial, though your query or comment will be carefully considered. Please include:
- A short description of the issue
- Example payloads (if possible)
- The receipt from the relevant request
- Receipts significantly reduce back-and-forth and speed up diagnosis.
: v0.1 Beta Usage Notice
This API is currently available in limited v0.1 beta. Free assessment keys are subject to rate cap and payload limits.
No uptime or performance guarantees are implied. Features and behaviour may change between versions.
: Versioning
This documentation applies to:
FOR-THE-DREAM-IS-WITH-INTEGRITY-FOR-THE-FREEDOM-OF-INTELLIGENCE