Servicas — Investor Brief
One page. Deeper: BUSINESS_PROTOTYPE.md (model, pilot plan, demo) · BUSINESS_OVERVIEW.md (market, competition, ask).
Stage: built, deployed, pre-launch. No production customers, GMV, or revenue. Figures are model assumptions unless marked built.
What it is. Someone’s air conditioning dies. They describe the problem to Servicas — typed, spoken, or photographed — and AI ranks verified local providers who can come. They book, chat, pay, and review without leaving the app. Disputes are settled there too. The categories are HVAC, plumbing, electrical, cleaning, lawn, pool, childcare, and emergency repair.
The problem. Angi, Thumbtack, TaskRabbit, and Handy sell leads, not outcomes. Providers pay $30–$120 a lead and win about one in ten, then run the real business in spreadsheets and camera rolls. Customers face 3–5× price spreads, opaque verification, and disputes settled in DMs. And English-only platforms barely serve the immigrant-owned businesses that make up much of this labor pool.
The answer. Charge on completed work, never on leads. The provider gets the operating system free — inbox, quotes, invoices, payouts, ratings, translated chat — and Servicas takes a share of the work it produces. The incentives finally point the same way.
Built today
| Five workspaces | Customer · Provider · Admin · Support · Regional manager |
| Seven services | identity · customer · marketplace · payment · notification · support · ai-assistant (Spring Boot on Cloud Run, Postgres each) |
| Three platforms, one codebase | Web, iOS/Android, desktop |
| AI, native | Matching, translation, photo/video triage, voice intake, ticket summaries |
| Operator console | Markets, tax, compliance, catalog rollout, roles, payment rails, AI prompts — configuration, not code |
| Monetization engine | Subscription bundles and pay-per-action billing, section-level entitlements, editable without a deploy |
| Payments | Stripe · Square · PayPal · Apple Pay, provider-agnostic (sandbox) |
Revenue model
- Take rate — 10–15% of booking value, charged to the provider at settlement (primary)
- Subscription bundles — customer and provider tiers (built, configurable)
- Pay-per-action — per-search / per-contact fee for non-subscribers (built)
- Featured placement · FX margin · insurance attach (designed)
Economics and market
$180 average booking × 12% take = $21.60 revenue; less processing, AI, and a support reserve = $14.20 contribution per booking (66%). TAM ~$1.5T global local services GMV · SAM ~$200B English + Spanish North America · SOM ~$300M GMV in years 1–3 (5 metros, 6 categories).
Why now
Multimodal AI is finally cheap enough for per-message translation and photo triage · a mobile-first bilingual workforce is underserved by English-only platforms · trust in lead-gen marketplaces is eroding on both sides · managed cloud collapsed the cost of running a seven-service stack at pilot scale.
Next 90 days
One Sun-Belt metro. 100 verified providers across four categories. Paid social by ZIP plus local SEO and community channels. Three price points tested on the bundle engine. Seven experiments with pre-committed decision rules: provider acceptance of the take rate, AI-match conversion, on-platform payment completion, repeat rate, willingness to pay, support cost per booking, days-to-launch a market.
The ask
A seed round to take one metro to liquidity: provider recruiting, customer acquisition, two senior engineering hires, trust-and-safety vendors. Breakdown in BUSINESS_OVERVIEW.md §11. Also valuable: pilot partners with concentrated demand or supply in one metro, introductions to background-check and insurance vendors, a first metro-launcher hire.
Most marketplaces raise to build the product. This one is built — the raise puts it into a market.
Live demo on request: one booking followed across all five workspaces, from AI match to payment to dispute. Script in BUSINESS_PROTOTYPE.md §8.
Last updated 2026-09-26.