The Stack

Complete service catalog with pricing — every SaaS, cloud platform, and operational tool the Foundation Architecture depends on.

The Stack

A complete inventory of every paid service behind the Foundation Architecture — what each one does, what it costs, and how the bill scales from a 3-person startup to a 100-person organization.

Service catalog — by vendor

Every service the architecture depends on, grouped by the vendor that provides it, with the logical function each one fills (sourced from the Recombinate IT Functions Map). Pricing detail and per-team totals follow below.

* Shared / usage-based services — the per-user figure is the total monthly bill divided by team size; actual unit cost drops as the team grows. Usage-based estimates (Telnyx ≈ 5 numbers + moderate volume; PostHog ~$50/mo light → ~$100/mo heavier; Clay / ReachInbox flat-rate shown as a 3-person share) scale with traffic, not headcount.

Open-source stack — $0 license

The product itself runs on an open-source build/runtime stack — no license cost. Runtime for the self-hosted pieces (Better Auth, FastAPI, plus Cerbos / Coder above) folds into the GCP compute already counted.

Undecided / WIP

Functions still in bake-off or active build — tracked in the IT Functions Map; no line-item cost committed yet.

HR & Payroll — pick one

Remote Development Environments

The architecture assumes Coder (self-hosted, Community edition) as the primary development environment. GitHub Codespaces is a viable managed alternative — pricing for both is shown below.

Coder (Community — self-hosted)

Coder is free and open-source; you pay only for the underlying infrastructure. The estimates below assume GCP, no HA, and no paid Coder support.

GitHub Codespaces (managed alternative)

Total Cost Estimates by Team Size

Assumptions

The estimates below apply these conventions:

• Neon (Scale) is the database. Pricing is usage-based with no monthly minimum (~$0.26 / CU-hr compute + storage). HIPAA is included on Scale — set hipaa:true and accept the self-serve BAA at no separate add-on cost. Because HIPAA no longer changes the bill, the totals below are a single figure — there is no separate "+ HIPAA" column, as there was under the old Supabase-based estimates.
• All web runs on GCP Cloud Run — the applications UI (Vite + TanStack) and the marketing site (Next.js). Vercel is not used.
• Observability is GCP-native (Cloud Logging / Trace / Monitoring via OpenTelemetry) — effectively free at launch volumes and folded into GCP usage. No Elastic / ELK.
• Durable execution (background jobs + the agentic human-in-the-loop agent loop) runs on Restate Cloud (managed, serverless) driving Cloud Run scale-to-zero handlers — no always-on node to operate; its free tier (50k actions/mo) covers launch volumes, moving to usage-based pricing as traffic grows.
• Microsoft 365 Business Premium is the primary productivity + security suite for office / clinical / eng / sales staff; Google Workspace Essentials (free) covers field / ops collaboration.
• Microsoft 365 is added for a subset of employees who need Office desktop apps (Excel, etc.).
• Rippling is the HR / payroll provider (US-centric team).
• Coder (Community, self-hosted on GCP) is the dev environment — not Codespaces.
• GPU add-on is excluded from the base estimates (add ~$55 / mo per AI-focused developer).
• Pulumi Cloud is tiered by the number of Pulumi users, not total employee count: Individual (free, 1 user), Team ($40 / mo flat, up to 10 users), Enterprise ($400 / mo, unlimited). In the estimates below, even larger companies can remain on Team if only a small platform / infra group needs Pulumi access.
• Shared / usage-based services (Neon, GCP, Cloudflare, Telnyx, Clay, ReachInbox, PostHog) appear once in the total — they are not duplicated per user. PostHog is billed on event / session-replay volume, not seats.
• Usage-based cloud services scale sub-linearly with headcount.

Team of 3

At three people everyone wears every hat — founder, engineer, salesperson, support. All three need access to every service. Pulumi starts free on the Individual tier — one person can manage deployments for the whole team.

Team of 20

At twenty people, light specialization begins. Not everyone needs every tool, but overlap is still high.

Assumed department split:

Team of 100

At a hundred people the company has distinct departments. Not every role needs an engineering seat or a sales tool.

Assumed department split:

Programmatic pricing — role → cost

The tables above are reference snapshots. The live model is a typed function that prices any team from a request object:

import { calculateStackCost } from '@/lib/pricing/stack-calculator';
 
calculateStackCost({
  composition: { cto: 1, 'software-engineer': 6, scheduler: 2, caregiver: 25 },
  mau: 15000,        // REQUIRED — drives usage-based services (Neon, GCP, PostHog, Telnyx)
  hipaa: true,       // default true — toggles per-service HIPAA add-ons
  overrides: {       // optional, per service id
    github: { billTier: 'office' },  // switch which tier a per-seat tool bills against
    linear: { seats: 5 },            // hard-code the scale factor
    clay:   { disabled: true },      // turn a service off
    gcp:    { usageMultiplier: 2 },  // scale a usage service
  },
});
// → { headcount, lines[], perCategory[], grandTotalMonthly, perPersonMonthly, hipaaPremiumMonthly, notModeled, warnings }

• Roles + tiers in src/lib/pricing/team.ts (Builder / Office / Field); the service cost model in stack-model.ts (per-seat / flat / MAU-usage, each with optional HIPAA add-ons). The "everyone builds" rule: all Office + Builder get Linear + a Claude coding-agent seat + Tailscale; only Builders add GitHub + Copilot + Coder.
• MAU is required — usage services (Neon, GCP, PostHog, Telnyx) scale with monthly active users, not headcount.
• HIPAA (default true) adds per-service premiums (M365 Business Premium, Neon Scale, PostHog Boost); the total is returned as hipaaPremiumMonthly so the HIPAA delta is visible directly.
• Overrides per service id: billTier (switch group), seats (hard-code scale), usageMultiplier (scale usage), disabled (off).
• CLI: pnpm pricing '{"composition":{"software-engineer":2},"mau":5000}'; pnpm pricing:gen regenerates the static JSON.
• OpenAPI spec (for agents): served at /api/pricing/openapi.json. The docs UI and this API run in one Cloud Run service (fronted by Cloudflare); the API is mounted under /api, so POST /api/pricing, POST /api/pricing/export.xlsx, GET /api/pricing/roles.json / scenarios.json, and MCP-over-HTTP at POST /api/mcp are all live. The spec's servers URL is relative (/api), so it resolves against whatever origin serves it — http://localhost:8080/api locally (pnpm api), https://foundation.h11h.io/api in prod.
• Role-tied services: WellSky (caregivers + clinical), Sage/Ramp (finance + C-suite), Clay/ReachInbox (billed only if a sales role is present), Figma (design/PM), Sanity (marketing), Peony data room (admin seats: CEO + CPO/CTO + the Corporate Development / Strategy role, which defaults to 0). Still not modeled (in notModeled): Dropbox Sign, AWS SES.

As You Grow

The services above cover a team from founding through Series A. As the organization scales beyond ~50 people or traffic grows significantly, consider adding: