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

# Y Combinator application (raw draft)

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


Answers ready to paste into the YC form (Winter 2027 closes on **Nov 2, 2026**, 8 pm Pacific time; confirm at ycombinator.com/apply). Only real data as of Oct 28, 2026; where there is no data we say so. At the end, the critique a partner would make.

Criteria used: ycinsight.com/apply (what partners look for in each question), ycinsight.com/yc-deal and ycinsight.com/yc-request-for-startups.

## Company

### Describe what your company does in 50 characters or less.

AI agents that resolve telecom customer support (47 characters)

### What is your company going to make?

SAVANT AI answers a mobile operator's customers by voice, WhatsApp and email and resolves the request inside the operator's own systems: balance and line lookups, number portability, top-ups, theft suspension, tickets. When it can't, it hands off to a human with a summary. We measure contention and first-contact resolution, not 'messages answered'.

## Progress

### How far along are you?

Live with 3 mobile brands of one group (INXEL: inXel, Jamachulel, XAP) since mid-2026. Last 30 days (public panel, 27-Oct-2026): 430 calls and 665 minutes handled by AI, 63% contained without a human. Revenue: ~US$3.2k/month combined with our KYC product sold to the same client; we have not split it per product. Public plans: ~US$805 / $1,560 / $2,700 per month.

### How long have each of you been working on this? How much of that has been full-time?

~4 months in production (first flows June 2026). ~6,500 lines of JS plus per-brand n8n flows. Vladimir built it; the founders also run other products.

## Idea

### Why did you pick this idea to work on? Do you have domain expertise? How do you know people need what you're making?

An operator we already worked with needed 24/7 support without hiring a bigger call center. We built it for them and they pay monthly. Evidence of need: they kept paying and added a third brand. We have no second customer yet.

### Who are your competitors? What do you understand about your business that they don't?

BPO call centers (human agents by the hour); generic WhatsApp bots (Yalo, Treble, Botmaker); suites like Zendesk/Genesys; and new AI support companies (Sierra, Decagon) that sell to large US enterprises. What we understand: MVNOs in Mexico run on the same shared network, so integrations for portability, balance and suspension are repeatable from one operator to the next.

### How do or will you make money? How much could you make?

Monthly plan per brand (~US$805–2,700) plus usage. Mexico has 21 M+ MVNO lines (IFT, 2024). At ~US$1,800/month per brand, 14 brands is ~US$300k ARR — a niche. The bigger bet is the same 'resolve inside the system' agent for utilities, fintech and retail.

### Tell us something surprising or amusing that one of you has discovered.

Fill in from the call logs: the most common reason people call. (Pending — we will not invent it.)

## Founders

### Founders

Three co-founders: Rafael Herrera (CEO/CFO), Vladimir Díaz (CTO, software engineer, 20 years, PhD in business competitiveness and innovation from IQS–Ramon Llull, advised Ford, CANTV, Google and Globant) and Francisco D'Angelo Ohep (lawyer, LLM in international business law, AI-law specialist from Universidad de Salamanca). Vladimir wrote the code.

## Numbers YC will work out

| Question | Honest answer |
| --- | --- |
| Default alive? | Not yet. Current revenue is ~US$3.2k/month shared with KYC; the funded plan reaches ~US$23k MRR in 24 months, still below its cost. |
| Does it fit a 2026 RFS? | Fits 'Vertical AI for compliance-heavy industries' and 'AI-native replacements for SaaS incumbents' only partially; don't force it. |
| Valuation we would ask for outside YC | US$2,500,000 pre-money |
| What YC offers | US$500k: US$125k for 7% (≈ US$1.79 M post) + US$375k on an uncapped SAFE with MFN |

## Blunt critique (what a partner would attack)
- This is the product with the best real evidence (paying customer, public containment metrics) and the best candidate to apply to YC.
- Fatal weak spot: a single customer group. YC will ask 'who else pays?'. You need 2–3 new operators, even as paid pilots, before applying.
- Separate SAVANT revenue from KYC revenue. 'US$3.2k combined' sounds like you don't know your numbers.
- Market: MVNOs in Mexico are a small market. You must explain in one line how this gets to US$100 M (more verticals, more countries).
- Team: YC requires full-time commitment. With 10 products, the partner will ask what the founders work on on Monday. The right answer: this.

## What to do in the next 8 weeks
- Split revenue by product and publish real MRR.
- Contact 20 MVNOs and close 2 paid pilots.
- Pull the real 'surprising' finding from the call log.
- Measure 48-hour FCR and cost per resolved conversation.