npbuilds Value Architect — Designing How Value Flows
Pricing an intelligence product is unlike pricing software. Software delivers a fixed capability at a fixed price. Intelligence delivers a growing capability that improves with use. The value increases over time — for both sides.
This creates a design challenge traditional pricing models don't address: the system gets more valuable to the user the longer they use it, AND the user gets more valuable to the system the more they contribute.
The Value Flow Model
Before pricing, map how value moves:
User → System: Usage data, feedback, behavioral signals, direct input
System → User: Intelligence, capability, personalization, time saved
System → System: Learning, capability compounding, pattern recognition
User → User: (Network effects, if applicable)
The value architecture must respect all four flows, not just the system→user transaction.
Pricing Models for Intelligence
Free Surface / Paid Depth
The surface layer is free. Deeper capability costs money.
| When | Risk | Mitigation |
|---|---|---|
| Core capability is enough to demonstrate value but depth is where real power lives | Free tier is too good — no conversion motivation | Make the depth genuinely different, not just "more of the same" |
AI-native variant: The free surface learns from use and becomes more valuable — but the depth of personalization and capability only unlocks with payment.
Usage-Based
Pay for what you consume.
| When | Risk | Mitigation |
|---|---|---|
| Usage directly correlates with value received. Heavy users get more value. | Users ration usage to save money, reducing the system's ability to learn from them | Generous free tier for the learning loop; charge for output, not input |
AI-native variant: Charge for outputs (deliverables, analyses, artifacts) not inputs (questions, conversations). This aligns incentives — the user is paying for value delivered.
Capability-Tiered
Different capability levels at different prices.
| When | Risk | Mitigation |
|---|---|---|
| Clear capability tiers exist (basic analysis vs. deep synthesis vs. cross-domain orchestration) | Users feel artificially limited. The intelligence COULD do more but WON'T. | Tiers should reflect genuine complexity, not artificial restriction |
AI-native variant: Tiers based on domain breadth — single-domain intelligence is one tier, cross-domain synthesis (The Loom's unique value) is a higher tier.
Outcome-Based
Pay based on results achieved.
| When | Risk | Mitigation |
|---|---|---|
| Outcomes are measurable and attributable to the system | Attribution is hard. What if the user would have succeeded anyway? | Hybrid: base fee + outcome bonus. Or money-back guarantee if outcome not achieved. |
AI-native variant: The system tracks the outcomes its recommendations produce. Price adjusts based on demonstrated value. Radical alignment of incentives.
Subscription
Fixed recurring payment.
| When | Risk | Mitigation |
|---|---|---|
| Ongoing relationship, continuous value, the default for most SaaS | Users forget they're paying. System has no incentive to improve for retained users. | Regular value demonstrations. "Here's what you got this month." |
AI-native variant: Subscription with value reporting — the system actively shows the user what it did for them each period. Not just features used, but value delivered.
The Value Architecture Document
# Value Architecture: {surface name}
## Value Flows
User → System: {what the system receives from use}
System → User: {what the user receives}
Compounding: {how value increases over time for both sides}
## Pricing Model
Primary: {model chosen}
Rationale: {why this model fits this intelligence surface}
## Tiers / Levels (if applicable)
{tier}: {what's included} — {price}
## Free Component
What's free: {what and why}
Why free: {learning loop value / demonstration / network effects}
## The Reflexivity Question
How does pricing affect behavior?
{Does the price signal quality? Does free usage reduce perceived value?
Does the pricing model incentivize the usage patterns that make the system better?}
## Anti-Models
{Pricing models explicitly rejected and why}
Cross-Domain Synthesis
Value architecture is the most cross-domain product decision:
- game-theory/mechanism-design — Pricing IS mechanism design. Is the pricing incentive-compatible? Does it produce honest behavior?
- archon (investing) — Market positioning through price. What does the price signal about the product's nature?
- prose-orchestrator — Value framing. The words used to present pricing shape perception as much as the numbers.
- investing/reflexivity — Price affects perception affects value affects price. The reflexive loop in pricing intelligence.