Vora
VOICE AI THAT COMPILES
The first voice AI
that picks up the phone
in every Arabic dialect.
Compiled from any URL in 2 minutes. 18 Arabic dialects. 100+ languages.
100,000 free inbound minutes per phone number.
voicevora.com / compile
WEBSITE URL
https://your-business.com
Crawl websiteDONE
Build knowledge baseDONE
Select voice & dialectCOMPILING
Deploy agent runtimeQUEUED
14
STAGE COMPILER
100+
LANGUAGES
18
ARABIC DIALECTS
100K
FREE MIN / NUMBER
Vora02 / 16 · WHO WE ARE
WHO WE ARE
Two builders. Three years in voice AI.
One gap nobody else would close.
THE STORY STARTED IN THE BALKANS
We became IBM technology partners while still very young. Three years ago we set out to deploy voice AI for SMBs in the Balkans through our firm Ventus — shipping chatbots and voice agents for insurance, banking, healthcare, and government clients across the region.
THE WALL WE HIT
Then we got stuck. For 1.5 years no voice model supported Serbian, Croatian, Macedonian properly. We tried Vapi. Retell. Synthflow. Botpress. Every chatbot builder was half-built. Every voice platform was missing the features production demands. Support was terrible. Integrations broke.
WHAT WE DID
So we built our own. Feature by feature, exactly as we'd build it for ourselves. Five months later: a working compiler that ships voice agents from any URL — 18 Arabic dialects, 100+ languages, 100,000 free inbound minutes per number. Then we realized the platform served both technical operators and non-technical SMB owners — so we split it into Vora and Vora Plus.
Stefan Stojanović
CO-FOUNDER & CEO · PRODUCT, GTM, COMPILE PIPELINE
Co-founder of Ventus. 3+ yrs deploying voice agents in production for Serbian and regional clients. Designed Vora's 14-stage compile pipeline and runs all of GTM.
Novak
CO-FOUNDER & CTO · VOICE PIPELINE, RAG, ML
Co-founder of Ventus. 3+ yrs in AI / voice systems. Owns Vora's multi-language architecture across 100+ languages, the RAG layer, and the agent runtime. Background in psychology informs conversational design.
VENTUS · WHAT WE'VE BEEN DOING
Through Ventus we've been building voice agents, running market research, and chasing enterprise deals across the Balkans and MENA. The work that became Vora started here. IBM technology partners · active Microsoft partnership development. Both co-founders 22 years old.
Vora03 / 16 · WHAT 5 MONTHS LOOKS LIKE
WHAT 5 MONTHS LOOKS LIKE
We built it feature by feature, for ourselves first.
Then realized the same platform served two buyers.
COMPILER · 14 STAGES
Your voice agent ships in 2 minutes — not 4 weeks of engineering.
Crawl → KB → dialect routing → prompt → tools → deployment → quality gate. URL in, live agent out.
ARABIC · 18 DIALECTS
Your hotel in Dubai answers in Khaleeji. Your clinic in Cairo in Masri. Without you configuring it.
Gulf · Egyptian · Levantine · Maghrebi · Iraqi. AR↔EN code-switch at 6.3% WER. The only platform that routes them.
LANGUAGES · 100+
Your business serves 100+ markets — through one agent, one URL, one bill.
Routed across Gemini, OpenAI, Anthropic, plus 13 TTS providers + 431 voices.
TELEPHONY · 100K FREE / NUMBER
Your customer hears ~$15,000 of inbound calls answered before paying us a dollar.
Our own telephony layer + Twilio · Telnyx · Plivo · Vonage · BYO-SIP fallback.
CUSTOMER MEMORY
Your returning callers get recognized. Without you programming the conversation.
bge-m3 embeddings on pgvector. Agent picks up where you left off — names, preferences, history.
AI PROVIDERS · 15 / 25
Your agent picks the best model per language. We pay. You don't manage it.
15 LLM provider types, 25 models — Claude · GPT-4o · Gemini · Llama · Mistral via OpenRouter.
MCP SERVER · LIVE
Build a voice agent just by chatting with Claude or ChatGPT.
We built an MCP that lets Claude/ChatGPT create, configure, and trigger Vora voice agents through normal conversation. No dashboard. No clicks. Just say what you want.
x402 · AGENT-TO-AGENT WORK
AI agents can hire Vora and pay us themselves — no human signs off.
We enabled real agent-to-agent commerce. An autonomous AI hires Vora, makes the call, pays in seconds. No voice AI competitor has shipped this.
INTEGRATIONS · 126 MCPs
Your agent books, charges, invoices, follows up — without you writing code.
126 MCPs catalogued across 25 industry categories (24 live today) + native HubSpot · Salesforce · Pipedrive · Zoho · Calendar · WhatsApp.
Every number on this slide is in our codebase right now. One backend, two front doors — Vora for SMBs, Vora Plus for operators. We couldn't fit both buyers in one product.
Vora04 / 16 · VORA · THE PRODUCT
VORA · FOR THE SMB OWNER
For the hotel manager who can't
and shouldn't have to learn voice AI.
THE BUYER
A mid-sized real estate brokerage in Dubai. 12 agents, 40 active listings, dozens of buyer inquiries a day in Khaleeji Arabic. Misses 60% of after-hours calls. Each missed call is a lost commission.
WHAT THEY DO WITH VORA
Paste their website. Two minutes later their main number is answered in fluent Khaleeji — the agent qualifies the buyer, books a viewing on the right broker's calendar, sends an SMS confirmation, and logs the lead in the brokerage's CRM.
THE PROMISE
"Paste your website. Get a voice agent. Forget about it."
No code. No engineering team. No consultant. No SDK. No prompt-writing.
voicevora.com / demo · live
CALL TRANSCRIPT · LIVE · GULF ARABIC
Agent: مرحبا، فندق البيت. كيف يمكنني مساعدتك؟
"Marhaba, fundoq Al-Bayt. Kayfa yumkinuni musa'adatuk?"
Caller: بغيت غرفة لليلتين من الجمعة
"Bagheet ghorfa li-laylatayn min al-juma'."
Agent: تمام. ٣٨٠ درهم في الليلة. أعمل الحجز؟
"Tamam. 380 dirham fil-layla. A'mal al-hajz?"
"Marhaba, fundoq Al-Bayt. Kayfa yumkinuni musa'adatuk?"
Caller: بغيت غرفة لليلتين من الجمعة
"Bagheet ghorfa li-laylatayn min al-juma'."
Agent: تمام. ٣٨٠ درهم في الليلة. أعمل الحجز؟
"Tamam. 380 dirham fil-layla. A'mal al-hajz?"
Sub-800ms · Gulf Arabic via Munsit
Booking saved to Calendar
SMS confirmation via Twilio
VERIFY IT YOURSELF · NOW
Compile your own agent at voicevora.com
Bring any URL. ~2 minutes. Platform ~90% production-ready — final 10% closes during diligence.
PRICING · PER BUSINESS
$0 · $50 · $99 · $299 / mo
Free tier: 20 min lifetime, no credit card.
Vora05 / 16 · VORA · WHY WE WIN
VORA · WHY WE WIN
For the MENA SMB buyer, every competitor
has at least one disqualifying gap. We have none.
SELF-SERVE
No-code setup the SMB owner can actually do alone.
Paste URL → live agent in 2 minutes. Vapi/Retell need a developer. Wittify/Sawt need a managed-services contract. Twilio needs an engineering team.
ARABIC THAT WORKS
18 dialects + real-time AR↔EN code-switching.
Auto-detect from phone prefix. 6.3% WER. Vapi/Retell ship MSA only. Wittify/Sawt cover dialects but no code-switch. Most route every Arabic call to Whisper Generic.
TELEPHONY INCLUDED
100,000 free inbound minutes per number — built in.
Number provisioning, carrier routing, call handling — all bundled. Every competitor on this list forces the SMB to bring their own Twilio and pay per minute from minute one.
PDPL-READY
Onshore-processable from day one for Saudi.
PDPL by default — no enterprise-tier paywall, no procurement bottleneck. US incumbents are locked out by default. Regional managed players gate it behind enterprise contracts.
SMB PRICING
$50 / $99 / $299 per month — sized for the buyer.
The hotel manager, the clinic, the brokerage — they can afford this without a CFO meeting. Vapi/Retell start at developer credit packs. Wittify/Sawt are enterprise sales. Twilio Flex is $150+/agent/month before telephony.
2-MINUTE COMPILE
URL in, live voice agent out. No prompt-writing.
14-stage compile pipeline does the work. Every competitor sells you a builder. We ship the compiler. Nobody else can do this — the orchestration isn't a feature you bolt on.
Vapi/Retell can't sell to SMBs — wrong GTM. Wittify/Sawt can't self-serve — wrong product. Twilio/Five9 are too expensive — wrong pricing.
Only Vora ships all six.
Vora06 / 16 · VORA · MOATS
VORA · MOATS
Two advantages competitors can't ship without rebuilding.
MOAT 1 · OUR OWN TELEPHONY
100,000 free inbound minutes per phone number — built in.
Vora ships telephony as part of the platform. Number provisioning, carrier routing, call handling — all included. SMB doesn't sign a separate Twilio contract. Doesn't worry about per-minute infra costs. The phone number arrives ready to ring.
WHY THIS IS A MOAT
Vapi, Retell, Synthflow all force the SMB to bring their own Twilio — sign a separate contract, pay per minute from minute one, configure SIP themselves. We bundle the whole stack and absorb the first 100K minutes. Competitors would have to rebuild their pricing model to match.
An SMB gets thousands of inbound calls answered before paying us a dollar. That closes the sale.
MOAT 2 · ARABIC THAT THE CUSTOMER WON'T QUIT
The customer stays in their own dialect. They don't hang up.
A buyer in Riyadh speaks Khaleeji. A patient in Cairo speaks Masri. A guest in Beirut speaks Levantine. Most voice AI defaults to MSA — so the customer switches to English in 30 seconds or hangs up. Vora detects the dialect from the phone prefix, routes to a model trained on it, code-switches with English in real time if the customer mixes. The call gets completed.
| Gulf | KSA · UAE · Qatar · Kuwait · Bahrain · Oman | Munsit |
| Egyptian | Egypt | Munsit |
| Levantine | Lebanon · Jordan · Syria · Palestine | Hamsa |
| Maghrebi | Morocco · Algeria · Tunisia · Libya | Hamsa |
| Iraqi | Iraq | Hamsa |
| AR ↔ EN code-switch | Bilingual flows | Speechmatics 6.3% |
WHY THIS IS DEFENSIBLE
Wittify and Sawt also offer Arabic dialects. Where we differ: real-time AR↔EN code-switching at 6.3% WER + automatic dialect detection from phone prefix. Most platforms route every Arabic call to Whisper Generic and the customer switches to English in 30 seconds.
Vora07 / 16 · VORA PLUS · THE PRODUCT
VORA PLUS · FOR THE OPERATOR
For the operator who's been let down
by every voice platform that came before.
WHAT WE HEARD
Operators tried Vapi, Retell, Synthflow, Botpress. The story was the same every time — templates that looked good in demos didn't survive a real production call. Memory missing. Integrations half-wired. Dialect support cosmetic. Telephony brittle. White-label was either a paid upgrade or a fiction.
WHAT WE BUILT
The platform we wished existed when we were running voice deployments ourselves. As easy as vibe-coding at the surface, with the depth a builder actually needs underneath. No half-features. No bait-and-switch tiers. Everything in production.
THE PROMISE
"Run your voice AI business. We power it quietly."
The agency owns the customer, the brand, the upside. Vora Plus is the engine they don't see, don't think about, and never have to explain to their clients.
WHO BUYS VORA PLUS
- → AI automation agency in Toronto serving 50 dental clinics
- → GoHighLevel reseller in Manila adding voice to their stack
- → MENA partner in Riyadh selling to insurance brokers
- → n8n studio in Berlin shipping voice flows for clients
- → In-house operator at a mid-market e-commerce company
THE FOUR FEATURES THAT MATTER MOST
FULL WHITE-LABEL
Their domain. Their brand. Their pricing to their clients.
AGENT SIMULATOR
Test the agent with persona-driven callers before deploying to real customers.
MCP SERVER
Claude / ChatGPT can call their voice agents as a tool.
CUSTOMER MEMORY
Their clients' callers get recognized across calls.
+ A/B flow testing · SIP BYOC · data residency · 33 MCP integrations · x402 / USDC ready
PRICING · PER AGENCY
$0 · $499 · $999 · $1,999 / mo
Free tier: 20 min lifetime, no credit card.
Vora08 / 16 · VORA PLUS · WHY WE WIN
VORA PLUS · WHY WE WIN
We built the depth other platforms
never bothered to ship.
FULL WHITE-LABEL DAY 1
Your domain. Your brand. Your pricing.
Custom domain, full UI rebrand, no Vora mention anywhere. Vapi/Retell don't offer it. Synthflow gates it behind enterprise. GHL ships a chrome reskin, not the platform.
SUB-ACCOUNTS + PER-CLIENT BILLING
Bill your clients separately on day one.
Isolated tenants, RBAC, SSO, audit logs, per-tenant billing rules. Every other platform leaves the agency to build this layer themselves — months of work before they can resell.
CUSTOMER MEMORY
Returning callers get recognized — across all calls.
3-tier memory: short-term + working + long-term + intelligent forgetting. No competitor in this set has shipped persistent cross-call memory. Each call starts from zero on every other platform.
MCP SERVER
Claude and ChatGPT can call your agents as a tool.
126 MCPs catalogued · 24 live · agent-callable on day one. Nobody else in the voice AI category has shipped MCP. The agency that resells Vora Plus sells "agent-ready" the moment Claude/ChatGPT users ask for voice.
x402 / USDC PAYMENTS
Autonomous agents can pay us — no human in the loop.
Stablecoin settlement on Base, 3-second finality. Zero competitor ships this. Every other platform's billing assumes a human signs the invoice — that breaks the moment agents hire your agents.
AGENT SIMULATOR
Test before deployment with persona-driven callers.
Stress-test agents with simulated personas — angry buyer, confused caller, hostile prospect — before they go live. Built because we shipped agents ourselves and got burned by "looked good in demo, died in production."
Vapi sells SDK access. Retell sells API access. Synthflow sells a builder. GHL sells a CRM with voice bolted on. None of them sells a full platform an agency can flip on day one.
We do — because we were the operator first.
Vora09 / 16 · VORA PLUS · MOATS
VORA PLUS · MOATS
Two compounding advantages.
MOAT 1 · OPERATOR-GRADE DEPTH
Features competitors never thought to ship.
Every feature in Vora Plus exists because we needed it ourselves while building real voice systems at Ventus. We were the customer first.
- → 3-tier customer memory — short-term + working + long-term + intelligent forgetting
- → MCP server — Claude / ChatGPT call Vora agents as a tool
- → x402 / USDC settlement — agent-to-agent stablecoin payments shipped
- → TCPA DNC.gov compliance loader — almost no voice-AI competitor ships this
- → Agent simulator — persona-driven test callers stress-test agents before deploy
- → 5-layer provider failover — circuit breakers across STT, TTS, LLM, SMS, realtime
- → 8 parallel decision tracks in the compiler — 22 distinct decisions per agent
- → Custom domain white-label + sub-account isolation + RBAC + audit + SSO
- → Webhook DLQ + 5 webhook services · semantic + query + audio caching
- → 126 MCPs catalogued across 25 categories · 24 live today
MOAT 2 · ZERO-CAC SCALE THROUGH AGENCIES
Each agency we sign is 30-100 SMBs we never had to find.
Voice AI in MENA is sold by people, not bought online. The agency already has the SMB's trust, the contract, and the local language. We are the engine inside their offering — invisible to the end customer, recurring on our side.
PIPELINE TODAY
6,000+
agencies sourced & graded
→
OUR ARR PER AGENCY
$12K–$24K
recurring · sticky
→
SCALE BY YEAR 3
5K-10K SMBs
on Vora through agencies
WHY THIS COMPOUNDS
Each Plus agency reaches 30-100 SMBs we'd otherwise have to find one-by-one. CAC is the agency's cost. Margin is ours. The more SMBs run on Vora through agencies, the better our dialect data, the harder we are to displace.
Vora10 / 16 · WHY NOW
WHY NOW
The four shifts that built the runway under us.
MAY 2026
CATEGORY VALIDATED
Voice AI is now venture-validated. Big tech is buying in.
Vapi $50M Series B at $500M valuation — Amazon Ring picked it over 40 rivals for 100% of inbound calls. Retell hit $40M ARR, profitable, 20-person team, 40M+ monthly calls. ElevenLabs hit $500M ARR / $11B valuation, IPO signaled. Google DeepMind absorbed Hume AI. The category transitioned from "interesting" to "every major fund is positioning."
JAN 2026
ARABIC LLM
Sovereign Arabic LLMs went production-grade.
TII Abu Dhabi launched Falcon-H1 Arabic — the 34B model beats Llama 3.3 70B and Qwen 2.5 72B on the Open Arabic LLM Leaderboard. Saudi PIF's HUMAIN deployed ALLaM as a sovereign LLM via Azure AI Foundry. Two state-funded, production-grade Arabic LLMs are now plug-and-playable — and Vora is the voice layer they need.
APR 2026
AGENT ECONOMY
MCP is the agent standard. x402 has real volume.
MCP: 97M monthly SDK downloads, 10K+ public servers, 78% of enterprise AI teams have MCP-backed agents in production (Linux Foundation, Dec 2025). x402: 69,000 active AI agents, 165M transactions, $50M USDC volume (Coinbase, April 2026). Cloudflare joined the x402 Foundation. We are the only voice platform that shipped both.
2025–26
MENA CAPITAL
MENA AI capital doubled. The buyer of our Series A is being capitalized right now.
MENA AI funding hit $858M in 2025 — 2× 2024, 22% of all MENA VC, Series A AI grew 6× YoY (MAGNiTT FY 2025). HUMAIN (PIF) is launching a $10B AI venture fund in Summer 2026 with a $1.2B financing framework for AI data centers. Saudi PDPL still locks foreign incumbents out by default.
Sources: TechCrunch (Vapi Series B, May 2026) · GlobeNewswire (Retell, Jan 2026) · TechCrunch (ElevenLabs Series D, Feb 2026) · TII Falcon-H1 launch · Microsoft Tech Community (HUMAIN ALLaM) · Anthropic / Linux Foundation MCP stats · Coinbase Developer Platform (x402) · MAGNiTT FY 2025 · PIF / Arab News (HUMAIN $10B fund)
Vora11 / 16 · MARKET
MARKET
Two markets. One backend.
MENA-anchored. Globally expandable.
TAM · GLOBAL
$17B → $50B
Conversational AI 2025 → 2031
MarketsandMarkets · 19.6% CAGR.
SAM · VORA (MENA SMB)
$400M → $1.2B
MENA SMB voice slice
MEA conv. AI (Fortune Business Insights) × 30% voice × 40% SMB filter.
SAM · VORA PLUS (GLOBAL)
$8B+†
Global AI-automation agency platforms
† ~100K+ AI automation agencies globally × ~$80K avg annual platform spend on voice + agent infra. Conservatively sized against our 6,000+ graded agency pipeline.
SOM · YEAR 3
$25–40M ARR
Combined Vora + Vora Plus
1% MENA SMB capture + 3% global agency capture.
VORA SOM MATH
3M MENA SMBs × 1% capture × $1,500 ARR = $45M reachable · Year-3 target $10–18M
VORA PLUS SOM MATH
50K global agencies × 3% capture × $9K ARR = $13.5M reachable · Year-3 target $15–22M
Sources: MarketsandMarkets Conversational AI 2025-2031 · Fortune Business Insights MEA Conversational AI · IMF MENA SME Baseline · agency market — internal estimate from our pipeline
Vora12 / 16 · BUSINESS MODEL
BUSINESS MODEL
Two products. Two pricing ladders. One margin profile.
Vora
MENA SMB
Subscription. Per-business pricing.
| TIER | $/MO | OVERAGE | FOR |
|---|---|---|---|
| Free | $0 | — | 20 min trial |
| Starter | $50 | $0.14 | Single-location SMB |
| Growth | $99 | $0.12 | Multi-location SMB |
| Pro | $299 | $0.10 | High-volume SMB |
EVERY TIER INCLUDES
18 Arabic dialects · 100+ languages · 100K free min / number · Calendar & CRM · Customer memory
Vora Plus
GLOBAL OPERATOR
Subscription + minutes pool. Per-agency pricing.
| TIER | $/MO | MIN/MO | OVERAGE |
|---|---|---|---|
| Free | $0 | 20 lifetime | — |
| Pro | $499 | 2,000 | $0.12 |
| Business | $999 | 5,000 | $0.10 |
| Enterprise | $1,999 | 15,000 | $0.08 |
EVERY TIER INCLUDES
Full white-label · Sub-accounts · Per-client billing · SIP BYOC · MCP server · Data residency
UNIT COST (ALL-IN)
$0.01–0.03 / min
STT + TTS + LLM + telephony at our scale
GROSS MARGIN AT SCALE
70–80% target
Achievable with our provider mix + caching
VORA OVERAGE
$0.14 → $0.08 / min
Drops by tier · all below Vapi sticker
PLUS OVERAGE
$0.12 → $0.08 / min
Plus Pro → Plus Enterprise
Vora13 / 16 · AGENT ECONOMY
THE AGENT ECONOMY · x402
AI agents need to make money to survive.
The ones that need to make phone calls — pay us.
CONCRETE EXAMPLE · REAL TODAY
Take Automaton (Conway Research, github.com/Conway-Research/automaton) — an open-source autonomous AI agent that must earn its own existence to keep running. It generates a wallet, picks up work, and gets compute time only if it has credits. If it stops earning, it dies.
When that agent gets a task that requires a real phone call — book a hotel, negotiate with a supplier, qualify a lead — it discovers Vora through our MCP server. It calls our tool. Vora rings the number, completes the task in the right language and dialect, returns the result. The agent pays us $0.42 in USDC via x402, settled in 3 seconds on Base — no Stripe, no card, no human in the loop.
Autonomous agents have to monetize their work to survive. The ones that need to make phone calls — pay us. That's a transaction with no approval bottleneck.
REAL TRACTION · APR 2026
69,000 active AI agents · 165M+ transactions · $50M USDC volume through x402 alone (Coinbase, April 2026). Cloudflare joined the x402 Foundation. The agent economy is no longer speculation.
WHY VORA WINS THIS
We are the only voice AI platform with MCP server + x402 in production. No competitor — Vapi, Retell, Bolna, Synthflow — has shipped both. And critically: the agent doesn't need a human's approval to pay us. That's a frictionless transaction nobody else in the MCP market can serve.
WHY THIS MATTERS FOR THE INVESTOR
A second revenue stream on top of SaaS subscriptions. Every Claude Computer Use, every Operator, every autonomous agent shipping in 2026 — when they need voice, they hire us. An option on the agent economy that we already paid to build.
Vora14 / 16 · MARKETING PLAN
MARKETING PLAN
Two channels. Pipeline built.
Execution is the only thing left.
VORA · MENA / INDIA / SEA SMB · PLAN
3,000+ influencers and influencer agencies. IG + TikTok.
- → 3,000+ micro-influencers + influencer agencies already sourced across MENA, India, SEA (in our CRM)
- → IG + TikTok — dialect-matched creators per market · activation starts at close
- → AEO / GEO content engine — first blogs & research already shipped, optimized for ChatGPT / Perplexity / Google AI Overview
- → Regional event presence — LEAP (Riyadh, March) · GITEX (Dubai, October)
- → Founders on the ground in UAE from close — direct field motion in Dubai + Riyadh
VORA PLUS · GLOBAL OPERATORS & AGENCIES · PLAN
Affiliates. Educational communities. Technical influencers.
- → Affiliate network — 30% lifetime commission, no cap, no clawback · structure built, first partners closing post-raise
- → Educational community partnerships — n8n / Make / GoHighLevel / AI automation cohorts and bootcamps
- → Technical influencers — YouTubers, newsletter writers, Twitter operators in the AI-automation space
- → AEO / GEO content engine — research and first content already shipped, B2B-tuned for the operator buyer
- → Agency partnership program — co-marketing for any agency white-labeling Vora Plus
OUR OUTREACH ENGINE · BUILT
We built the tools. Now we need to use them at scale.
- → Our own CRM — purpose-built for our outreach motion, lead grading + scoring built in
- → Automated email outreach — LLM-personalized at scale, tracked open / reply / conversion
- → LinkedIn outreach automation — for Vora Plus agency operators
- → WhatsApp at scale — early test: 471 LLM-personalized messages already shipped in MENA
- → Reddit + X monitoring — real-time inbound from operator complaints about competitors
We've done the research, built the CRM, sourced the leads, started the content. We are about halfway there.
The $475K GTM budget funds the execution — not the plan.
Vora15 / 16 · RISKS & MITIGATIONS
PROACTIVE ANSWERS
Five questions a sharp IC will ask.
Our answers — before you have to ask.
RISK · TRACTION
Pre-revenue. No signed customer logos on the deck.
MITIGATION
Live working product at voicevora.com — compile an agent during this meeting on any URL you pick. 6,000+ agencies graded, 471 LLM-personalized WhatsApp messages in flight. Underlying voice systems shipped in production at Ventus for 3 years before Vora was named.
RISK · FIRST-TIME FUNDRAISERS
Two 22-year-old co-founders. First raise. No operating playbook for $2M deployment.
MITIGATION
Three years of production deployment via Ventus (IBM partner) — not zero operating experience, just first time deploying venture capital specifically. Tranched close. $1M at close + $1M at milestone — the structure shares execution risk with the investor.
RISK · PROVIDER CONCENTRATION
Arabic dialect routing depends on Munsit, Hamsa, Speechmatics — third parties we don't own.
MITIGATION
Routing logic is the moat, not the providers. Failover already built — if any provider degrades, we route to the next. Q3 2026 roadmap: self-trained dialect models for Gulf and Egyptian, reducing dependency on Munsit specifically.
RISK · US INCUMBENT RESPONSE
What if Vapi ships Arabic dialect routing in Q3 2026?
MITIGATION
Vapi's developer-first GTM cannibalizes if they pivot to SMB no-code. Their $20M war chest doesn't translate to the MENA-SMB wedge — it's the wrong product for this buyer. Realistically we have 12-18 months before any serious response. By then, agency channel + Vora Plus depth become the structural defense.
RISK · TWO PRODUCTS IN PARALLEL
Two ICPs (MENA SMB + Global agencies) with two GTMs and two founders. Where's the focus risk?
MITIGATION
One backend. Two front doors. Vora Plus is the productization of operator-experience we already had at Ventus — not a separate roadmap. If forced to choose: Vora Plus wins. The agency channel is the leverage. Vora direct is the MENA wedge that buys mindshare for the platform. Plus's unit economics are stronger and the buyer is more sophisticated.
Vora16 / 16 · THE ASK
THE ASK
$1.7M pre-seed.
$8–11M post-money · ~17% dilution · priced round, optionally tranched
STRUCTURE
1 anchor lead + 3–5 family offices / angels
Optional tranche: $1M close + $700K at milestone (50 paying SMBs OR 3 signed agencies within 6 mo)
Optional tranche: $1M close + $700K at milestone (50 paying SMBs OR 3 signed agencies within 6 mo)
WHERE THE $1.7M GOES
- 40% · $680K Engineering — UAE/Belgrade hybrid team (voice infra + Arabic depth + Plus operator features)
- 28% · $475K GTM — UAE BD lead, influencer-agency partnerships, content engine, MENA events
- 15% · $255K Ops & Compliance — DIFC / ADGM entity, PDPL + DPO, SOC 2 audit
- 8% · $135K Founder relocation — both co-founders to UAE: visa, housing, on-the-ground setup
- 6% · $100K Infra / COGS — cloud + STT / TTS / LLM provider costs at scale
- 3% · $55K Reserve
DAY-90 PLAN
- → DIFC / ADGM SPV formed
- → Both co-founders relocated to UAE
- → UAE BD lead hired · first agency partnerships activated
- → 10 paying SMBs on Vora · 3 agency LOIs on Plus
- → SOC 2 audit kicked off
SERIES A IN 18 MONTHS
- → $500K–$1M ARR · ~300 paying SMBs · ~25 agencies on Plus
- → Target raise: $4-7M seed at $25-35M post
THE LEVERAGE
A few hundred dollars in our pockets.
Two builders. Tristenik, Serbia.
Five months. Live in production.
Two builders. Tristenik, Serbia.
Five months. Live in production.
Imagine eighteen more months
with $2M and the team it builds.
with $2M and the team it builds.
CONDITIONS PRECEDENT
DIFC or ADGM entity formation within 90 days · both co-founders relocate to UAE within 6 months · UAE BD lead hired within 6 months. Same operating model as NodeShift ($1M seed, Presight–Shorooq Fund I).
ADVISORS · BEING ENGAGED
Targets being engaged via warm intros: ex-Etisalat enterprise sales, ex-IBM MENA solutions, Endeavor MENA mentor. We're not naming what isn't signed.
VoraAPPENDIX A · ROADMAP
APPENDIX A · 12-MONTH ROADMAP
What the $1.7M unlocks in 12 months.
Q1 · MONTHS 1-3
Foundation
- → ADGM / DIFC entity
- → Founder relocation
- → Hire MENA BD lead
- → First 10 paying SMBs
- → First 3 white-label agencies
Q2 · MONTHS 4-6
Scale signals
- → 50 paying SMBs
- → 10 agencies
- → Arabic dialect quality v2
- → Multi-agent orchestration (Plus)
Q3 · MONTHS 7-9
Channel depth
- → 150 paying SMBs
- → 15 agencies
- → First named enterprise channel partner
- → Per-tenant white-label theming live
Q4 · MONTHS 10-12
Seed-raise ready
- → ~300 paying SMBs
- → ~25 agencies on Plus
- → $500K-$1M ARR run rate
- → Seed-round materials prepared
VoraAPPENDIX B · ARCHITECTURE
APPENDIX B · TECHNICAL ARCHITECTURE
The stack behind the compiler.
BACKEND · RUNTIME
- — TypeScript + Express + Prisma
- — Postgres + pgvector (RAG)
- — Python asyncio agent runtime
- — LiveKit WebRTC (voice transport)
- — Hetzner + Oracle infra
AI / VOICE PROVIDERS
- — 15 LLM provider types, 25 models
- — 13 TTS providers, 431 voices
- — 8–10 STT providers + dialect routing
- — bge-m3 embeddings + Xenova ONNX
- — Quality-gated agent versioning
TELEPHONY · INTEGRATIONS
- — 5 SIP providers (Twilio, Telnyx, Plivo, Vonage, BYO)
- — Our own telephony layer · 100K free inbound min mechanism
- — Google Calendar, Calendly, HubSpot, 4 CRMs
- — 126 MCP integrations catalogued (24 live)
- — x402 / USDC settlement (EIP-3009)
COMPILE PIPELINE · 14 STAGES
Intelligence · KB Analysis · Parallel Decisions (Tracks B/D/E/F/G/H) · Sequential Decisions (Track C providers → Track A voice) · Pre-Prompt Coherence · System Prompt · Greeting · Call Analysis · FAQ Seeding · Action Generation · Output Generation · Post-Output Coherence · KB Persistence · Quality Gate
DEPLOYMENT
Per-agent
isolated runtime
isolated runtime