<div class="portada"><div class="meta">Menteia · SAVANT AI · October 2026 · Confidential</div>

# Business plan

**SAVANT AI** — AI contact center on voice, WhatsApp and email that resolves (not just answers) mobile carriers' customer requests.</div>


## 1. Summary
SAVANT AI (Cerebro-VAI engine) is an AI contact center on voice, WhatsApp and email, integrated with the carrier's real system. It checks line status and balance, assists with number porting, guides tech support, processes top-ups, suspends stolen lines and opens tickets; it hands off to a human with a summary when needed and measures real resolution (FCR at 48 h).

> Everyone sells chatbots. We sell resolution.

## 2. Problem
- Mobile virtual network operators (MVNOs) are growing fast (over 21 M lines in Mexico), but their customer service doesn't scale: every new line adds load to an expensive call center.
- Generic chatbots answer but don't resolve: they can't check balances, process number porting, or open tickets in the carrier's system.
- After hours and at peak times, customers drop off and port out to a competitor.

## 3. Solution and product
- AI agents per brand on voice (Vapi + ElevenLabs), WhatsApp and email, with each carrier's own knowledge base.
- Integration with the carrier's backend (Altán / operator systems) to execute real actions, not just reply.
- Human handoff in Twilio Flex with a mandatory summary; ticket numbers issued after hours.
- Multi-brand dashboard: voice, WhatsApp with human takeover, email, SMS, analytics, billing and tickets.
- Containment and first-contact resolution (FCR) measured and validated with a 48 h survey.

### Technology
- Node.js + Express; self-hosted n8n orchestration with per-brand flows.
- OpenAI GPT-4.1 / 4.1-mini with a vector store per brand; Vapi voice (Deepgram + ElevenLabs).
- Twilio (Conversations, Flex, DIDs), AWS SES, Stripe, Frappe Helpdesk.
- DigitalOcean + Cloudflare; superadmin / client admin / operator roles.

## 4. Market
| Level | Annual size | How it is calculated |
| --- | --- | --- |
| TAM | US$213.5 M | Conversational AI market in Mexico 2025 (IMARC), projected at US$2,495 M in 2034 |
| SAM | US$10 M | 21 M MVNO lines × US$0.50/year spend on AI customer service |
| SOM | US$350k | 12–15 carriers and telecom brands at an average ticket of US$24k/year |

**Supporting data:**
- MVNOs surpassed 21 M lines (14.2% of the market) in Q3-2024 (IFT, Mexico's telecom regulator).
- 141.7 M active mobile lines in Q2-2025 (CRT).
- Conversational AI in Mexico: US$213.5 M in 2025 → US$2,495 M in 2034 (IMARC).
- Contact center revenue in Mexico: >US$1,400 M (Frost & Sullivan, 2025 projection).

*Market figures are estimates based on public sources; validate them before sending to institutional funds.*

## 5. Target customer
- Mobile virtual network operators (MVNOs) and phone brands on the shared network.
- Regional internet and telecom providers.
- Future: utilities, fintech and retail with high support volume.

## 6. Business model and pricing
Monthly subscription per brand (platform + channels) plus a variable fee per resolved conversation or voice minute. One-time setup fee per integration.

| Item | Price |
| --- | --- |
| Setup and integration per brand | US$3,000–8,000 one-time |
| Platform per brand | US$800–1,500 / month |
| Resolved conversation (WhatsApp/email) | US$0.15–0.30 |
| AI voice minute | US$0.12–0.20 |

## 7. Go-to-market strategy
- Anchor case study (INXEL group: inXel, Jamachulel, XAP) as a reference for other MVNOs.
- Direct sales to the ~30 active MVNOs in Mexico; the shared network (Altán) makes integration repeatable.
- Partnerships with MVNO enablers (MVNE/MVNA) to sell as a bundle.
- Expansion to LatAm (Colombia, Peru, Chile), where the MVNO model is growing.

## 8. Competition and positioning
| Alternative | What it offers | Our advantage |
| --- | --- | --- |
| Traditional call center (BPO) | Human agents billed by the hour | 24/7 at a fraction of the cost, no headcount growth |
| Generic chatbots (Botmaker, Yalo) | WhatsApp replies | Resolve inside the carrier's system |
| CX suites (Zendesk, Genesys) | Expensive generic platforms | Telecom-specialized, live in days |
| In-house build | Months and a dedicated team | Proven product with 3 brands live |

## 9. Defensible advantages
- Measurable resolution: 48 h FCR and containment with an auditable calculation basis, not 'messages handled'.
- Deep integration with telecom operations (porting, balance, suspension): domain knowledge that's hard to copy.
- Critical logic enforced in code (brand control, duplicates, business hours, summary before transfer): the LLM doesn't decide sensitive actions.
- Per-brand isolation: independent flows, credentials and webhooks; a new brand onboards in days.

## 10. Traction
- In production with 3 telecom brands from the INXEL group (inXel, Jamachulel and XAP): voice, WhatsApp and email.
- Live since mid-2026; recurring revenue shared with KYC of ~US$3,200/month.
- Active invoicing (series MEN-2026) and daily operations with human handoff.

## 11. Team
| Partner | Role | Stake in Menteia |
| --- | --- | --- |
| Rafael Herrera | CEO / CFO | 33.3% |
| Vladimir Díaz | CTO | 33.3% |
| Francisco D'Angelo | Head of Legal and Compliance | 33.3% |

**Hires funded by the round:**
- Senior full-stack engineer — from month 1
- AI / ML engineer — from month 1
- Customer success / implementation — from month 1
- B2B account executive — from month 3
- Full-stack developer — from month 7
- Support / operations — from month 9

## 12. Financial projection (24 months)
| Quarter | Revenue | Expenses | Net cash flow | Cumulative cash flow |
| --- | --- | --- | --- | --- |
| Q1 | US$7,290 | US$73,090 | US$-65,800 | US$-65,800 |
| Q2 | US$10,440 | US$79,522 | US$-69,082 | US$-134,882 |
| Q3 | US$17,325 | US$92,886 | US$-75,561 | US$-210,443 |
| Q4 | US$24,345 | US$97,129 | US$-72,784 | US$-283,226 |
| Q5 | US$33,795 | US$97,723 | US$-63,928 | US$-347,154 |
| Q6 | US$43,110 | US$98,317 | US$-55,207 | US$-402,361 |
| Q7 | US$55,125 | US$98,911 | US$-43,786 | US$-446,146 |
| Q8 | US$66,735 | US$99,505 | US$-32,770 | US$-478,916 |

- Monthly recurring revenue (MRR) at month 24: **US$23,400** (ARR US$280,800)
- Total 24-month spend: **US$737,081**; cumulative revenue: **US$258,165**
- Breakeven falls after month 24; the round covers the full runway without relying on revenue

**Revenue assumptions:**
- Current revenue attributed to SAVANT: ~US$2,160/month (assumption: 2/3 of revenue shared with KYC) with 3 brands.
- One new brand or carrier every 2 months starting month 5: 13 brands by month 24.
- Average revenue per brand grows from US$720 to US$1,800/month through volume and additional channels.
- Telephony and AI costs grow 3% monthly with volume.

## 13. The round
- Amount: **US$825,000** for **24.8%** (pre-money US$2,500,000, post-money US$3,325,000).
- Use of funds detailed in *06_Modelo_Financiero_y_Ronda* (financial model and round).

## 14. Risks and mitigation
| Risk | Mitigation |
| --- | --- |
| Concentration in one client group (INXEL) | Commercial priority: 2 new carriers in 6 months; annual contracts |
| Dependence on Twilio, Vapi and OpenAI | Provider abstraction layer; telephony migration plan |
| Persistence in JSON files | Migration to Postgres in the first 90 days |
| Privacy of telecom users' data | Privacy notice and data processor agreement per client |
| Scattered brand | Unify under SAVANT AI with IMPI (Mexican trademark office) registration |