Tax Labs runs three revenue levers at once — usage covers the cost of doing the work, licensing makes revenue predictable, and equity captures the upside of every vertical it helps build. Blended and anchored on a co-build Studio tier, this turns an infrastructure business into a cash-flow-positive venture studio: transaction fees pay the engineers; equity is the return.
Adopt a three-tier blended model anchored on the Studio tier: 15% equity + cost-coverage usage fees for new co-built verticals; a pure Platform licence (usage + subscription, no equity) for operators who already have a team; and an Enterprise white-label for incumbents. Usage keeps the lights on, licensing floors the revenue, and equity is where the money is made.
Each lever does a different job. Usage recovers the marginal cost of every document, filing and bank read. Licensing turns lumpy usage into a predictable floor. Equity converts Tax Labs from a vendor into a co-investor aligned with each vertical's long-run success. The art is the blend — and which lever leads depends on the tier.
Per-document, per-filing, per-bank-read. Scales with work done; passes variable cloud + API cost through at a margin. Defensible even to skeptics: "you pay only when we do work." The model of Stripe, Twilio, AWS.
covers fixed + variable cost
A monthly platform fee covers access to auth, RAG-over-legislation, confidence scoring, the HITL queue, dashboards and white-labelling. It floors the economics even in low-volume months and gives revenue visibility.
revenue predictability
Tax Labs is the technical co-founder. For early verticals without a tech team it takes 15–20% in exchange for building and running the stack — the in-kind value of a 2–4 person team costing €300–600k/yr.
upside + alignment
No single lever is right for every customer. Usage alone under-captures a vertical that scales to a €100M exit; equity alone scares off operators who already have a team. The blend — tuned per tier — is what lets Tax Labs serve the whole market without leaving money, or alignment, on the table.
One platform, three commercial wrappers. Studio is the recommended default for new verticals; Platform for existing operators; Enterprise for incumbents.
Who: operators with a team switching infrastructure or adding verticals.
€1,500/mo base (€1,200 annual)
€0.80 / invoice · €2.00 / filing
€0.30 / bank-account/mo · €1.50 / HITL doc
RAG ≤10k queries included · 48hr SLA
Who: entrepreneurs launching a new vertical with no tech team. Tax Labs is the CTO.
€0/mo licence (waived 24 mo)
Usage at cost + 20% (cost-coverage only)
Full technical build + weekly advisory
15–20% equity · 3-yr vest, 1-yr cliff
Who: fuel-card, payroll & HR platforms wanting the stack under their own brand.
€60k–€250k/yr licence (custom)
Per-unit rate 40–60% below Tier 1
Dedicated infra · 99.9% SLA · 4hr response
White-label rights · no equity
| Element | Tier 1 · Platform | Tier 2 · Studio | Tier 3 · Enterprise |
|---|---|---|---|
| Price mechanic | Licence + usage | Equity + cost-coverage usage | Annual licence + volume usage |
| Platform licence | €1,200–1,500 / mo | €0 (waived 24 mo) | €60k–250k / yr |
| Per-invoice processing | €0.80 | cost + 20% | 40–60% below T1 |
| Per-filing | €2.00 | cost + 20% | negotiated |
| Technical build & advisory | ✗ | ✓ included | ~ integration team |
| Equity taken | ✗ none | 15–20% | ✗ none |
| Lead lever | Usage + Licence | Equity | Licence |
Studio migrates to Tier 1 at a 20% "portfolio company" discount after 24 months. Tier benchmarks anchored to Mindee €49–€649/mo 13 and Veryfi's $500/mo minimum 14. Md
The same infrastructure is monetised three ways so no customer is turned away. New verticals default to Studio (equity-led); established operators buy Platform (usage-led); incumbents license Enterprise. One build, three markets.
Reference: the EU truck fuel-VAT vertical (Studio tier), at 500 trucks. Each truck recovers ~€6,000/yr; the operator charges a 15% success fee 7 → €900 gross revenue/truck. Tax Labs sits underneath at <8% of that.
| Cost line (per truck / yr) | Unit cost | Qty | € / truck | Source · conf |
|---|---|---|---|---|
| LLM inference — invoice extraction | ~$0.003/inv | 120 | ~€0.36 | Claude Haiku ($1/$5 per M tok) 11 Hi |
| OCR / document preprocessing | €0.005/pg | 240 pg | €1.20 | Mindee Pro rate 13 Hi |
| Open-banking reads | €0.50/mo | 12 | €6.00 | Tink standard 10 Md |
| Tax-authority API / filing | ~€0.50/filing | 2 | €1.00 | AEAT free; wrapper compute Lo |
| Compute (serverless) + storage | — | — | ~€0.12 | AWS Lambda + S3 15 Hi |
| RAG / legal queries | €0.002/q | 50 | €0.10 | Embedded token cost Md |
| HITL review (10% escalation) | €1.50/doc | 12 | €18.00 | Freelance reviewer, 5 min/doc Md |
| Allocated platform fixed cost | shared | — | €8.00 | §6.6 allocation Md |
| Total cost-to-serve / truck / yr | ~€34.80 | at 500 trucks |
Key insight: HITL (human review) is ~52% of cost-to-serve — the single biggest line, and the one that automation compresses most as training data accumulates.
| Stream | € / truck / yr |
|---|---|
| Usage fees (cost + 20% margin) | ~€41.76 |
| Licence allocated (post Y2) | ~€29 |
| Total cash / truck / yr | ~€70–100 |
| + Equity upside | 15–20% × exit |
Cash gross margin ~20–40% early (HITL-heavy), scaling to 60–70% as escalation drops. Equity is the real return.
Tax Labs takes <8% of the operator's revenue in cash — well within the 10–15% infra-COGS benchmark — and leaves a strong net contribution to the operator.
At the level of one truck the model is unremarkable — a few tens of euros of cash. The point is that this repeats across thousands of trucks and dozens of verticals on shared fixed cost, and every one carries an equity option. Unit economics are the floor; leverage and portfolio are the return.
Shared fixed cost is amortised across every truck on the platform, and HITL escalation falls as cross-vertical data trains the models. Cost-per-truck collapses; margin approaches SaaS quality.
| Trucks on platform | Fixed cost / truck | HITL rate → cost | Total / truck / yr | Gross margin on €70 cash |
|---|---|---|---|---|
| 500 | €8.00 | 10% → €18.00 | ~€35 | ~50% |
| 2,000 | €3.00 | 7% → €12.60 | ~€25 | ~64% |
| 5,000 | €1.50 | 5% → €9.00 | ~€17 | ~76% |
| 10,000 | €0.80 | 3% → €5.40 | ~€11 | ~84% |
| 20,000+ | €0.40 | 2% → €3.60 | ~€8 | ~89% |
Illustrative operating-leverage model. Md — unit costs are sourced; escalation-rate improvement and fixed-cost allocation are internal estimates.
This is the whole argument in one curve. A single vertical at 500 trucks runs ~50% margin; a platform serving 10,000 trucks across five-plus verticals reaches 84–89% — SaaS-quality economics — because fixed cost becomes essentially free and HITL automation compounds with data. No single-vertical operator can reach the right-hand side of this chart.
Small equity across many verticals = recurring cash plus venture upside. Even with modest per-vertical odds, the portfolio expected value is material. Illustrative model — venture outcomes are estimates.
10 verticals over 5 years · Tax Labs holds ~15% (post-dilution) · outcomes below.
| Scenario | Exit | TL 15% | P | EV |
|---|---|---|---|---|
| Material exit | $30M | $4.5M | 30% | $1.35M |
| Scale | $100M | $15M | 20% | $3.00M |
| Modest | $2M | $0.3M | 40% | $0.12M |
| Failure | $0 | $0 | 10% | $0 |
| EV / vertical | $4.47M |
The infrastructure revenue funds the engineering team; the equity portfolio is the return on that investment. Tax Labs is, in effect, a cash-flow-positive venture studio — recurring revenue underneath, portfolio optionality on top.
Illustrative portfolio model, independent outcomes assumed. Equity value depends on venture outcomes years 4–8 and is the lowest-confidence figure in this dossier. Lo
Any one vertical may fail — 10% of the model assumes exactly that. But across ten independent bets, small stakes compound into a portfolio worth more than the entire recurring business, on top of that recurring business. That optionality is what a pure-SaaS pricing model can never capture.
| Cost category | Early ~1k trucks | Mature ~10k trucks |
|---|---|---|
| Platform engineering | €200k | €350k |
| AI / RAG infra | €30k | €50k |
| Tax-authority connectivity | €40k | €80k |
| Cloud base | €15k | €30k |
| Legal / compliance | €20k | €40k |
| Fixed / semi-fixed | ~€305k | ~€550k |
| Variable (LLM·OCR·OB·HITL) | ~€13k | ~€130k |
| Total | ~€318k | ~€680k |
| Cost / truck | €318 | €68 |
Fixed cost dominates early and is shared across all customers — the source of the operating leverage in §6.4. Md
What a vertical operator spends to build the same stack alone, Year 1:
| Build-it-yourself line | Year 1 |
|---|---|
| CTO hire | €120–180k |
| 2 engineers | €160–240k |
| OCR · LLM · cloud | €30–50k |
| Tax-authority integrations (one-time) | €100–200k |
| Total Year-1 tech cost | €430–770k |
Tax Labs charges a 500-truck vertical €35–50k/yr cash + equity for an equivalent, better, shared-data stack. Against €430–770k build cost, that is an 85–95% cost reduction 16 — the operator's core reason to buy.
A 10-vertical platform spends ~€680k/yr total vs €4.3M–7.7M if each of 10 verticals built independently. Tax Labs captures ~€1–2/truck in platform margin and passes ~€120–140/truck in savings to each operator. Md
"We don't want to give up equity." Three structured responses:
The message: equity is for co-builders, not for platform users. Want infrastructure? Pay for it. Want a technical co-founder? Share the upside.
Fallback for resisters who still want alignment: modified Tier 1 at €2,500/mo + a 5% success-fee kicker on recovered amounts — alignment without dilution.
Default to Studio (15% equity + cost-coverage usage) for new verticals; Platform for existing operators; Enterprise for incumbents. This captures upside asymmetrically, aligns Tax Labs with each vertical's success rather than its API volume, and builds a venture-studio cash flywheel. Biggest risk: downstream funding dilutes Tax Labs's stake — mitigate with pro-rata rights, clear vesting & anti-dilution terms, and a pre-agreed conversion to pure Tier 1 pricing if a large round makes the equity relationship complex.