LevelChatLevelChatSeed · Investor deck

One SDK.
Any topology.

The real-time layer for modern apps — video, voice, and live streaming from 1:1 calls to 50,000-viewer broadcasts.

Mert Yıldız · Foundermertysta2@gmail.comApril 2026

Every modern app needs real-time. The infrastructure for it is still stuck in 2016.

Three structural problems, one migration event, a decade of accumulated glue code.

  • Telehealth startup

    Stitches Twilio Video (EOL Dec 2026), Deepgram, and Pusher. Four contracts, four SLAs, one on-call schedule that nobody wants.

  • Creator platform

    Runs LiveKit for chat, Mux for broadcast, Clerk for auth, Stripe for billing. Four vendors, one experience, zero leverage.

  • EdTech company

    Pays $18k/yr for Zoom, gets no branding, and the recordings are 6 GB per hour.

Topology silos

Vendors specialize in one shape — meeting, webinar, or broadcast — and get awkward outside it.

Opaque pricing

Per-minute bundles hide CDN, storage, and AI costs. Shock bills are weekly HN stories.

Cloud lock-in

EU, gov, health, and regulated customers can't run the same stack on-prem.

Twilio Video shuts down December 2026. 10,000+ applications need to migrate now. There has not been a moment like this in real-time infra for a decade.

A unified real-time SDK with honest pricing and on-prem parity.

One API surface across meeting, webinar, broadcast, and voice agent. Two meters. Same binaries for cloud and on-prem. A CLI that migrates Twilio for you.

One SDK, every shape

Room, Track, Participant primitives for 1:1 calls, group meetings, webinars, and 50,000-viewer broadcasts.

Two meters, not one

Participant-minutes for meetings, viewer-hours for broadcasts. Broadcasts inherit CDN economics, not SFU economics.

AV1 + SVC default

Server-side re-encode absorbs the CPU cost for low-power senders.

Recording compression on egress

1-hour classes are 500 MB, not 3 GB.

Same binaries

Managed cloud or Helm chart into your own Kubernetes cluster. Move between them when compliance changes.

Migration CLI

npx levelchat migrate twilio | agora | zoom — diff-PR ready.

A $48B market, a $22B addressable slice, a $23M plan-of-record.

Gartner CPaaS forecast (Mar 2025); IDC RTC market note (Oct 2025); Grand View Research (Feb 2026).

TAM 2025
$17.6B
TAM 2030
$48B
+22% CAGR
SAM 2030 (embed SDK)
$22B
29% CAGR
Twilio migration wedge
$6M
10k apps × 10% × $6k avg ACV
EdTech vertical
$6M
600 schools × $10k
Creator broadcast
$6M
200 × $30k
Core PLG dev teams
$5.4M
6,000 × $900

Adjacent expansion not counted in SOM but material: voice-agent builders, live commerce, sports streaming — each a $2B+ sub-category on its own trajectory.

Five minutes, one call, 500-viewer broadcast, one recording, one migration.

A single unbroken demo loop that we run live on stage.

  1. 00:05npx levelchat init in a fresh directory.
  2. 00:15pnpm dev — open two browser tabs, we are on a call.
  3. 00:30Add mode="broadcast", refresh — same SDK, now handling 500 simulated viewers from a load-test script.
  4. 00:45Toggle room.record(), click stop. Output: 1.2 MB/min, AV1-Pro.
  5. 00:55npx levelchat migrate twilio in a cloned open-source Twilio Video sample — diff PR opens, compiles.
  6. 01:00Public pricing calculator slides — meeting and broadcast side-by-side.

Four revenue lines. PLG dominant. Usage expansion compounding.

LineMotionY1 mixY3 mix
Managed-cloud subscriptionPLG55%42%
Usage-based overages (meters + AI)PLG expansion30%33%
Creator-mode plansPLG + partner10%13%
On-prem licensesSales-led5%12%
Gross margin Y1 → Y3
74 → 78%
CAC / LTV (blended)
15:1
Payback (median)
< 2 mo
NRR target Y3
130%

Numbers refresh weekly — current file at docs/metrics/weekly.md.

Marked placeholder until deck-usage time. Filled live at meetings with last-seven-day snapshots.

Active customers (beta)
[N]
Monthly participant-min
[N]M
Enterprise pilots signed
[3]
Pilot TCV combined
$[X]k
Waitlist
[N]
Migration CLI ports
[N] repos
GitHub stars
[N]
Discord members
[N]
MoM usage growth
+[X]%

Four compounding moats, none copy-paste-able in 12 months.

  1. 1
    Unified topology architecture

    Cascade SFU + AV1/SVC + recording compression + viewer-hour CDN routing — a coordinated engineering bet that takes 18+ months to replicate.

  2. 2
    On-prem parity

    US hyperscaler vendors will not sell a 10-file Helm chart that runs on Hetzner; it conflicts with their AWS/Azure partnership economics.

  3. 3
    Developer ecosystem flywheel

    SDK surface, migration CLI, open-source upstream contributions, public benchmarks. Each lowers acquisition cost for the next developer.

  4. 4
    Pricing transparency

    Calculator, per-dimension meters, no-surprises credit guarantee. Brand equity hard to copy without re-architecting pricing.

Five-year vision-of-moat: federated rooms (“Stripe Connect” for real-time), marketplace of third-party add-ons with a 15% take rate, spatial audio and 3D rooms for events and embedded experiences.

Six tailwinds. One migration event.

  • Twilio Video EOL December 2026

    Single largest migration event in real-time infra this decade.

  • AV1 patent licensing clarified 2024

    Now safe to ship at scale. 30–40% bandwidth savings unlocked.

  • Safari 17+ AV1 decode

    Last major browser blocker fell in 2024.

  • Voice agents primary use case

    LLM + TTS stacks need a real-time video/voice foundation layer.

  • EU AI Act + GDPR enforcement

    Enterprises now actively reject US-only data residency.

  • Hetzner + Bunny priced to disrupt

    Baremetal and CDN 5–10× cheaper than US hyperscaler equivalents.

We built this on nights and weekends with people who have done this before.

Mert Yıldız
Founder · CEO / CTO

Previous infra/RTC work, scale systems shipped, public open-source contributions.

Co-founder · Eng #1
Distributed systems

SRE, kubernetes/helm-fluent, ex-[Company].

DevRel Lead
Joins Q3 2026

Ex-[dev-tools company]; prolific conference speaker; author of [artifact].

Design Lead
Brand sensibility: Linear / Apple

Ex-[product-company]; shipped [notable app].

Advisors: [RTC-industry names + EdTech or Creator-platform operator]. Hiring next 6 months: 2 systems engineers (Go/C++), 1 mobile, 1 full-stack, 1 DevRel, 1 SE.

Raising $4M seed @ $20M post-money.

24-month plan. Milestones below, committed to on the record.

Use of funds
  • Engineering (10 FTE by M12)$2.1M
  • DevRel + Docs$400k
  • Infra (Hetzner + Bunny + GPU pool)$500k
  • Compliance (SOC2 II, HIPAA, ISO 27001)$200k
  • GTM (content, sponsorships, conferences)$500k
  • Legal + G&A$300k
Milestones
  • M6 — GA launch100 paying · 5 Enterprise pilots
  • M12 — $1.2M ARRSOC2 II letter · HIPAA BAA · Helm v1
  • M18 — $4M ARR120% NRR · LTV/CAC > 5 → Series A
  • M24 — $7M+ ARR10 on-prem Enterprise customers

In five years, LevelChat is what “video, voice, and live” defaults to — the way auth defaults to Auth0/Clerk and payments defaults to Stripe.

The next generation of software is voice-first, video-native, and live by default. Classrooms, creator tools, voice agents, customer service, virtual-presence apps — all need a unified real-time primitive that doesn't lock them into a hyperscaler and doesn't make them pick between meeting and broadcast.

We are building that primitive.

How to help

Request an intro.

We're raising a $4M seed and actively shortlisting. If you lead infra/devtools pre-seed/seed — get in touch. The full financial model is available under NDA.

LIVE
“Every decade, one layer of the stack becomes inevitable. Stripe took payments. Auth0 took identity. Vercel took deployments. The real-time layer is next. We intend to take it.”
— Mert, founder.