Artifact · v0 (real content)

Pricing Model Spec

Four pricing model candidates stress-tested against three scenarios. Why the volume-floor subscription model is the recommended path, and how the SR&ED yield calculation changes the math.

1.2 · Revenue Architecture · artifact id: pricing-model-spec-v0.html · 2026-05-28 · v0 draft

The revenue mechanics problem

"we are not a SaaS model. So even if the customer says, I'm going to commit to a million dollar to you in twenty twenty-six, every month that that million dollar, he has no obligation to commit to that number, whatever happens, because it's really a spend per month." Discovery call, May 15

The structural problem was diagnosed precisely on the call. SmartOne has real revenue but no committed revenue. Every month's spending is re-decided. The burn math at $100K per month is manageable in theory; it is existential in practice because planning is impossible when revenue is weekly-variable.

The pricing model work is not about raising prices or creating new tiers. It is about converting existing spend-per-month relationships into minimum-floor commitments that let the company plan quarters instead of months.

Context: SR&ED yield and the pricing floor math

The B4 research on Canada's SR&ED program is directly relevant to pricing design. A Quebec-incorporated entity with 12 Montreal-based R&D employees and $1.5M in eligible annual R&D spend can generate approximately $801,250 in refundable cash credits per year (federal SR&ED at 35% plus Quebec CRIC at 30/20%). Effective recovery rate on eligible spend: 53%.

This matters for pricing because it changes SmartOne's effective cost base. If $1.5M in qualifying annotation methodology work generates $800K in cash back, SmartOne can offer a more aggressive annual commitment pricing floor than a competitor without SR&ED access. The prerequisite is a Canadian CCPC entity, which SmartOne's Montreal HQ does not yet confirm as incorporated.

SR&ED yield scenario (post-entity establishment). Annual eligible R&D spend: $1.5M. Federal SR&ED at 35% refundable: $446K. Quebec CRIC at 30/20%: $355K. Total annual cash return: ~$801K. This is non-dilutive cash that reduces effective service delivery cost and creates pricing room for volume discounts without margin compression.

Four model candidates

Model A: Pure Spend-Per-Month (current state)

Status quo

Client pays for actual volume consumed each month. No floor, no commitment. Rate is typically per-task or per-hour. SmartOne re-quotes or confirms continuation each month. Client has full optionality to pause, reduce, or exit at any time.

Scenario	Monthly revenue	Planning horizon	Risk
Upside (robotics pilot expands)	$600-700K	None	Client pauses in month 4
Base (current run rate)	$500K	None	Renegotiation each cycle
Downside (Amazon reduces spend)	$200-300K	None	60% revenue risk in one call

Assessment: This is the Sama pattern. Sama also had no committed revenue from Meta. When the contract ended, 1,108 workers lost their jobs in one week. The spend-per-month model is a structural risk that SmartOne has not yet resolved.

Model B: Annual Commitment with Volume Floor

Recommended

Client commits to a minimum annual spend (e.g., $600K/year, payable quarterly or in two tranches). Actual usage can exceed the floor; overage is billed at agreed rates. In exchange for committing, the client receives a rate advantage (5-10% discount versus month-to-month), a guaranteed capacity reservation (no queue during ramp events), and a dedicated account management contact. SmartOne gets planning horizon and revenue predictability.

Scenario	Committed baseline	Planning horizon	Risk
Upside (robotics pilot + 2 new accounts)	$2.4M/yr	12 months	Over-delivery capacity constraint
Base (current 4 warm accounts convert)	$1.2-1.8M/yr	12 months	One account non-renews at 12 months
Downside (only robotics converts)	$600-900K/yr	12 months	Below burn if no additional growth

Assessment: The volume-floor model is standard in managed services at this tier. The rate advantage (5-10% off) is the client's incentive to commit; the capacity reservation is the operational signal that SmartOne takes the commitment seriously. Most importantly: the floor converts survival-mode cash planning into manageable quarters.

Model C: Outcome-Based Pricing

Phase 2 consideration

Pricing tied to measurable output quality or downstream model performance. Example: pricing per validated annotation (QA-approved, not just delivered), or pricing per model accuracy improvement metric after training on SmartOne data. Requires clear measurement protocols and shared access to model performance data.

Scenario	Upside	Downside	Prerequisite
Physical AI buyer with mature evaluation pipeline	Premium pricing, 20-40% over volume rate	Margin risk if QA reject rate rises	Joint measurement protocol signed
Startup buyer without evaluation infra	Relationship anchor, preferred vendor status	Revenue variability tied to their model maturity	SmartOne builds basic eval tooling
Enterprise with DORA/AI Act compliance requirement	Compliance premium justified by documented provenance	Buyer unwilling to share model performance data	EU AI Act documentation framework built

Assessment: Outcome-based pricing requires measurement infrastructure that SmartOne does not have today. The prerequisite is the platform work described in Phase 3. Not recommended for Phase 1; worth building toward. The SR&ED credit infrastructure (annotation methodology R&D) is directly funded by the kind of measurement framework this model requires.

Model D: Platform Access + Services Hybrid

Phase 3 / after platform spec

A recurring platform access fee (e.g., $15-25K/month) covers the annotation tooling layer, QA dashboard, and data provenance audit trail, with usage-based billing for actual annotation volume. The platform fee stabilizes revenue independent of annotation volume; volume billing upside follows the work.

Scenario	Platform fee	Volume floor	Year 1 total
Enterprise Physical AI buyer (3 accounts)	$180-300K/yr	$600K/yr	$780K-$900K
Mid-market (5 accounts)	$100-150K/yr	$400K/yr	$500K-$550K
One anchor enterprise (scaled)	$300K/yr	$1.2M/yr	$1.5M

Assessment: This is the end-state model if Platform Architect (Role 3.1) succeeds. Not feasible in Phase 1 because the platform layer does not exist yet. The open-source tool stack described on the call (10-20 tools "stringed together") needs a UI layer and a named product before platform pricing is defensible.

Recommendation

Phase 1 recommendation: Model B (annual commitment with volume floor). Convert the robotics pilot and the 3-4 warm accounts to 12-month commitments with a floor of $50-100K per quarter per account. The rate advantage (5-10% discount) and capacity reservation are the mechanics. The SR&ED prerequisite (Canadian entity establishment) should be addressed in parallel because the refundable credits change the effective cost structure and create room for the commitment discount.

Model	Feasible Phase 1?	Revenue commitment?	Prerequisite	Recommended?
A: Spend-per-month	Yes (current state)	No	None	No
B: Annual floor	Yes	Yes	Pricing template + commitment language	Yes
C: Outcome-based	No	Partially	Measurement infrastructure	Phase 2+
D: Platform hybrid	No	Yes (fee layer)	Platform spec + product layer	Phase 3

Next artifact in this sequence

The annual commitment template (see linked artifact from Role 1.2) is the operational output of this pricing model spec. It contains the actual commercial language, floor structures, and overage mechanics for Phase 1 account conversations.