Tools and vendor stack
2026 pricing snapshot — verify at build time, these shift often
2 min read
| Category | Vendor examples | 2026 pricing |
|---|---|---|
| LLM API | Anthropic Claude and comparable frontier APIs | Entry-tier models run roughly $1/$5 per million tokens in/out; mid-tier roughly $2–3/$10–15; top-tier roughly $5/$25; prompt caching typically reads at a fraction of base input cost, and batch processing typically halves both rates [Established, verify live — pricing moves] |
| Orchestration | n8n | Cloud starter tier roughly $24/mo; self-hosted is free on your own low-cost VPS |
| Orchestration (alternative) | Make.com | Comparable tiered pricing, similar range |
| Vector database | A managed Postgres/pgvector offering | Free to roughly $25/mo at typical SMB scale |
| Vector database (scale) | A dedicated managed vector database | $70–300+/mo depending on volume |
| Voice AI | Several specialist voice-agent platforms | Roughly $0.05–0.10/min platform fee plus pass-through costs; blended realistic cost $0.10–0.30/min |
| Telephony | A standard VOIP provider | Roughly $1–2/mo per US local number plus usage |
| Hosting | A modern app-hosting platform | Free to roughly $20/mo entry tier, scale as needed |
| Entity (local) | A freezone-style or equivalent light-touch registrar | Varies widely by jurisdiction — verify current fees directly with the registrar |
| Entity (US) | A remote-incorporation service plus a remote-friendly bank | Roughly $300–500 formation, plus a modest annual registered-agent fee; an international transfer service moves funds at a low percentage fee |
| Legal / contracts | A template MSA, or a lawyer-drafted one | A few hundred dollars for a template; a few thousand for a lawyer-drafted agreement for regulated verticals |
| Social/content scheduling and AI generation | See The social and marketing angle | Layered on top of the orchestration/API base above — see that lesson for the dedicated stack |
The orchestration layer and the LLM API are the shared spine of every offer in this course — a backend intake agent and a client's automated content calendar run on the same underlying infrastructure. What changes by offer is the layer on top: a vector database and voice-AI stack for backend implementation work, versus a scheduling and AI-content-generation stack for the social/marketing angle.
AI Agency · progress saved in this browser · sign in to sync across devices
Up next
KPIs and kill switches
The specific numbers at which you stop or pivot, not scale through
2 min