VICIfast
About VICIfast

The face behind VICIfast.

Founder-led. Seven homegrown VICIdial engineers.

Eighteen years of VICIdial - running call centres, deploying it for clients, fixing it at the weekend. VICIfast is the version I wanted to use myself, built and run with seven full-time VICIdial engineers we trained in-house. Servers come up in under 40 seconds, measured on every one. A real dashboard. Audit on everything. Pricing you can read on the website. Cancel any time, no friction.
Talk to us

7-day free trial · Cancel anytime · Pay with card or USDT

18 yrin cloud telephony · since 2008
2,500+VICIdial deployments
7full-time VICIdial engineers
<40smeasured deploy median
Kaushal Kumar, founder of VICIfast

Who actually runs this

Kaushal Kumar

VICIdial architect & cloud telephony engineer

“I build the systems call centres run on.”

Eighteen years in cloud telephony, working in the space since 2008. Over 2,500 VICIdial deployments, the largest a single floor of 1,200 concurrent agents, migrated with no downtime. Platforms built on this stack carry more than ten million calls a day.

The day job has always been the awkward part of VICIdial rather than the demo: Asterisk and OpenSIPS, carrier routing that survives a bad hour, cluster scaling, TCPA and DNC compliance, WebRTC, CRM integration, and more recently AI call analysis. VICIfast exists because that work should not need to be repeated by hand for every new floor.

Kaushal runs VICIfast with seven full-time VICIdial engineers, all homegrown and trained in-house. They answer the support threads and run the Managed Floor accounts; he takes anything commercial or complicated.

Also behind Adoptiv Inc, Callix LLC and Prishal Technolabs — all of them pointed at the same problem from different angles.

Why we built this

Managed VICIdial used to mean “a guy with an SSH key and a spreadsheet.”

We've worked with VICIdial for over a decade - running call centers, installing it for clients, fixing it on weekends. Every time, the same friction: fast machines, slow setup, no real dashboard, no real audit, no real escape hatch.

The pattern repeats whether you're self-hosting or paying a managed provider. You wait. You get a server with a default password and a spreadsheet of credentials over email. There's no obvious way to invite your colleague. You can't see what state your server is in without SSH-ing into it. Pricing is mysterious. The vendor lives in your inbox.

VICIfast is what we wanted to use ourselves: a real product around VICIdial. Server delivery in 40 seconds, measured. A dashboard that shows every state change. Sub-users with proper roles. Audit log on everything. Pricing you can read on the website. Cancel any time, no friction.

We don't tie you to our minutes. We don't bundle agent licensing. We don't lock you in. Bring your own carrier with no markup from us, or use VICIfast Voice, our own US termination and DIDs, from $0.0085/min. If we ever cease to be the right fit, your data is yours and we'll help you migrate out.

How we run the business

Small, technical, self-funded.

An engineer working at a terminal on the platform

Senior engineers only

Every customer talks to someone who understands the stack. There is no outsourced tier-one desk reading from a script. The seven engineers on your support threads are the same people who patch the platform.

No outside funding

Revenue funds the product. We ship what helps customers, not what helps a board deck. There is no exit pressure to suddenly become a different company.

Operator-built

The team has hands-on time running production VICIdial floors. Every default in the platform exists because we saw the alternative break a real customer.

For nerds

The real stack, by layer.

If you're evaluating us against a build-it-yourself plan or doing a vendor due- diligence pass, here's the inventory. Versions are current as of the latest deploy. We rebuild the golden image every 30-60 days and bump these.

Customer dialer · the box

What ships to each customer. Single-tenant. Their root, their disk.

  • VICIdial2.14-b0.5
  • Asterisk22.5.1
  • PJSIPws · wss · udp transports
  • ConfBridgefor monitor / barge audio
  • MariaDB10.11 · 412 tables
  • Apache + PHP7.4 · security-patched builds
  • Perl AGITCPA scrub · custom dialplan
  • Ubuntu LTS22.04 · hardened image
  • systemd · cloud-initvendor-data overridden
  • ufw · fail2banfirewall + intrusion jails

Platform · the dashboard

The Next app you sign in to. Server-first React, strict TypeScript.

  • Next.js16 · App Router · RSC
  • React19 · server components first
  • TypeScriptstrict · no implicit any
  • Prisma7.8 · schema-managed migrations
  • PostgreSQL17
  • Redis7 · cache + queue backplane
  • BullMQworkers · scheduler · Bull-Board
  • shadcn/ui + TremorTailwind v4 design system
  • Zodevery API input validated
  • Vitest + Playwrightunit + e2e harness

Infrastructure

The bones. Boring, well-tested, no surprise vendors.

  • Cloud serversshared + dedicated vCPU
  • Object storageS3-compatible · logos · snapshots
  • CloudflareDNS-only · gray cloud · DNS-01
  • Let's Encryptper-customer TLS · auto-renew
  • 6 regions liveeu-fra · nbg · hel · ash · hil · sin
  • Provider-level snapshotsdaily 03:00 · 7-day retention
  • Cloudflare APIper-customer subdomain DNS
  • systemd-cronplatform scheduler · queues

Identity + payments

How customers sign in and how money moves.

  • Better-Authsession + RBAC
  • WebAuthnsecond factor for owners
  • Stripe + Stripe Taxsubs · invoicing · portal
  • PayPalsubs + one-shot wallet top-ups
  • USDT-TRC20TronGrid match + suffix routing
  • Resend + React Emailtransactional · per-reseller domain
  • Wallet ledgerappend-only · idempotency-keyed
  • Reseller dual-debitretail + wholesale in one tx

Observability + audit

How we know what is happening on every box, every payment, every state change.

  • Sentryplatform + per-customer error budget
  • PostHogfeature flags + product analytics
  • Incident logper server · opened by the probe
  • AuditLogappend-only · 7-yr retention
  • BillingEventevery money mutation logged
  • Live deploy timerper-stage wall-clock visible
  • Health probesevery 60s · HEALTHY / DEGRADED / UNREACHABLE
  • Recorded SSH sessionsasciinema · replayable in dashboard

What we don't run

Tools we considered + rejected. Listed so you know what trade-offs we made and why.

  • Kubernetesoverkill for single-tenant VPS
  • Multi-tenant DBno shared schema; one DB per box
  • Per-agent licensingVICIdial does not. We do not.
  • Cloud-vendor CDN proxyorange cloud breaks SIP / RTP
  • LLM-generated docsevery doc grounded in real config
  • Outsourced supportno script-reader tier

How we're different

Four principles. Real ones.

Speed

Timed on every server, from the datacenter handing it over to VICIdial running. The median is under 40 seconds, and the full histogram is published and recomputed nightly from production.

Transparency

Audit log, per-server health and incident history, sub-processor list, DPA pre-purchase. If you have to email us to find out something, we consider it a bug. Source for every customer-visible state change is in the dashboard.

Boring tech

Postgres. Redis. BullMQ. Prisma. Next. Stripe. Resend. Cloudflare. Let's Encrypt. Nothing surprising - just composed well. We deploy what we can debug at 3 AM.

No vendor lock-in

Your server runs stock VICIdial, so its database, recordings and configuration move to any other VICIdial host. Download a backup any time; when you cancel we keep one for 90 days. No exit fee. Only our agent theme and hosted features stay behind.

Want the procurement-ready version?

Security posture, compliance status, retention windows, DPA, sub-processor list, incident disclosure contact - one page that procurement teams can hand to legal.

Same VICIdial. Built by operators.

Get a dedicated VICIdial server in 40 seconds. Free for 7 days. Pricing you can read on the website.

Talk to us