| 1 | AI Coding / Dev Tool Platforms Replit, Cursor - Provide an AI-assisted coding / dev environment (agent-driven code generation, compute-heavy).
- Goal: keep the entry seat price low and predictable, while converting heavy agent/compute usage into overage revenue instead of forcing an abrupt plan upgrade.
| - Multiple seat tiers, each with a different usage allowance
- Monthly-resetting usage-credit pool sized per tier
- Real-time metering against that pool with depletion alerts
- Overage billed only after the tier's pool is exhausted, at a rate/logic tied to that specific tier (not a flat global rate)
- Pro-rated re-scoping of the usage pool on mid-cycle tier upgrade/downgrade
| - Hybrid seat + pooled usage credit with plan-scoped overage
- A tiered subscription where each tier carries its own usage allowance, plus metered overage billing keyed to the active tier
|
| 2 | Consumer AI Companion / Chat Apps Replika, Character.AI - Provide a consumer AI companion / chat experience at consumer-app price points.
- Goal: maximize subscriber retention (minimize involuntary churn from failed payments) while opening a second, incremental revenue lever through virtual-currency micro-transactions.
| - High-volume, low-ACV recurring billing at consumer scale
- Strong dunning / involuntary-churn recovery (retry logic, smart card-decline handling, account updater)
- Support for one-off virtual-currency micro-transactions alongside the base subscription
- Simple flat-rate plan structure — metering precision is not the priority
- Localized pricing/payment methods for global consumer reach
| - Flat consumer subscription + virtual-currency micro-transactions
- Dunning-optimized recurring billing rather than usage metering
|
| 3 | Foundation Model / API Providers OpenAI, Mistral AI - Provide raw compute and intelligence (model inference at API scale).
- Goal: maximize GPU utilization while ensuring every single token processed is billed accurately.
| - Per-token metering at extreme volume/granularity (billions of events/day, sub-cent unit economics)
- Prepaid credit balance with real-time balance enforcement (block usage at zero)
- Minimum top-up thresholds and credit-expiration handling
- High-throughput rating/aggregation pipeline that keeps billing near-real-time at that event volume
- Custom-quoted enterprise layer on top (committed spend, volume discounts, true-up against prepaid)
| - Prepaid metered/token balance with real-time enforcement
- Enterprise committed-spend contracts for large accounts
|
| 4 | Outcome-Based AI Platforms Intercom Fin - Provide an AI agent that resolves customer-support outcomes, not just usage.
- Goal: price and bill strictly on delivered business value (a resolved ticket), so customers pay for outcomes, not effort or tokens.
| - Ability to ingest an externally-asserted "outcome" event via API (not a platform-observed usage event)
- Per-outcome pricing with event-specific rates
- Base plan bundling an included outcome allotment, with per-unit overage beyond it
- Outcome validation/reconciliation & audit trail, since the platform isn't the source of truth for what counts as billable
- Flexible, customer-configurable rate cards per outcome type
| - Per-outcome/event billing driven by externally ingested outcome signals
- Bundled-allotment base plan plus per-unit overage
|
| 5 | Vertical Enterprise AI-Native (High-ACV, Quote-Based) Harvey (legal AI) - Provide a high-trust, vertical AI product (e.g. legal) to enterprise buyers.
- Goal: capture large, multi-year contract value through negotiated seat licensing, not self-serve volume.
| - No self-serve pricing — fully sales-quoted, negotiated contracts
- Seat-based licensing with minimum seat commitments
- Multi-year/12-month term support with custom renewal cycles
- Custom invoicing reflecting negotiated terms, milestone billing, and bespoke payment schedules
- Contract lifecycle management (amendments, renewals, seat-count true-ups mid-term)
- Support for per-deal discounting/bundling
| - Quote-based enterprise licensing
- Negotiated seat-minimum contracts with custom invoicing and contract-lifecycle management, not self-serve plans
|
| 6 | Traditional Seat-Based SaaS Layering In AI Notion AI, Salesforce Einstein / Agentforce - Provide AI capability as an enhancement to an already-adopted seat-based product.
- Goal: monetize AI as incremental revenue on the existing subscription base, without disrupting the base contract or renewal cycle.
| - Ability to layer a metered/credit-based add-on onto an existing seat contract without disrupting its renewal cycle or terms
- Support for gating AI into a specific existing tier as an upsell alternative to an add-on
- Mid-term add-on attach with proration aligned to the base contract's billing cycle (not a separate cycle)
- Credit-pool purchase/top-up mechanism for the add-on independent of base renewal timing
- Consolidated invoicing so add-on and base seat charges appear on one bill/relationship
| - Seat subscription + metered/credit add-on layered on the existing contract
- Cycle-aligned to the base subscription with consolidated invoicing
|
| 7 | AI Infrastructure / Inference & Compute Providers CoreWeave, Together AI, Fireworks AI - Provide GPU compute and managed inference as a service, to AI builders who need dependable capacity.
- Goal: lock in committed revenue through reservations while monetizing burst demand, and keep expensive GPU capacity highly utilized.
| - Reserved / provisioned capacity commitments with terms and discounts
- On-demand and spot rates alongside reserved capacity
- Multiple units in one account (GPU-hours, tokens, throughput units)
- Commit drawdown first, then overage at the on-demand rate
- Metering per resource (instance, endpoint, cluster) rolled up to the customer
- Utilization and spend visibility, with limits as balances run down
- Enterprise contract terms for large reservations
| - Commit + burst: reserved / provisioned capacity commitment with on-demand or per-token overage
- Multi-unit metering rolled up per customer
|