SBC-AutoOpsMetaventions AI
Mission-critical voice had no pre-deployment safety net. Until now.

The independent
truth layer for
real-time voice.

Every business call crosses a Session Border Controller, the fragile, multi-vendor gateway between your network and the carrier. One misconfiguration and calls fail silently: one-way audio, dead trunks, hours of repair, and you hear about it from users, not a dashboard. SBC-AutoOps is the independent layer that reads any vendor's config and tells you exactly what will break, before you ship it. Local-first and air-gapped: raw configs never leave your environment.

You cannot monitor a call that is cryptographically blocked from ever starting. The only place to catch it is the config, before you ship.

5 vendor parsers · 8 validation domains · 145 tests in CI · air-gapped
sbc-validator · pre-deployment walk
$ sbc-validator walk sbc-teams-01.contoso.com.ini

  STAGE 1 · INGEST
  Teams leg   transport=tcp   mTLS=no   SRTP=no   roots 5/7

  STAGE 2 · VALIDATE
  [CRITICAL]  C.CA.ROOT_MISSING          2 of 7 Microsoft roots missing (DigiCert G5 pair)
  [CRITICAL]  D.NAT.PRIVATE_ADVERTISED   private IP advertised in SDP
  [HIGH]      B.SIP.TRANSPORT            transport is TCP, not TLS
  [HIGH]      C.TLS.MTLS_DISABLED        mutual TLS off
  [HIGH]      C.SRTP.DISABLED            media not encrypted

  STAGE 3 · VERDICT       risk 100/100   →   BLOCK

  STAGE 4 · PREDICT
  SBC  -->  TLS ClientHello
       <--  unknown CA / handshake failed      << call dies here
  Outcome: NO_CONNECT  (dies at TLS handshake)
80M+Teams Phone monthly users worldwide
20M+rely on Direct Routing for PSTN calling
4–12h+typical MTTR for an SBC misconfiguration
$1.1→2.4BSBC market, 2025 to 2035, ~8% CAGR
One model · every major SBC platform
AudioCodes Mediant·Cisco CUBE·Ribbon SBC Core·Oracle Acme Packet
Why now · the loudest example

The clearest proof is happening right now.

Take the most acute case. Teams Direct Routing runs entirely on mutual TLS, and in 2026 Microsoft retired the legacy roots, moved to new DigiCert and Microsoft 2017 root CAs, and the SBC certificate must carry the Server Authentication EKU. An SBC whose trust store is not updated does not degrade gracefully. It stops, completely. And this is not the last forcing event: public-TLS certificate lifetimes are compressing on a fixed schedule (200 days in 2026, 100 in 2027, 47 by 2029), the post-quantum migration is coming, and the carriers keep changing too. Trust now rotates faster than telecom. The migrations never stop. The fragile config underneath never gets an independent check. That is the permanent problem this layer exists to solve.

The failure mode

Instant TLS handshake rejection. SIP 408 timeouts. Total inbound and outbound outage. And nothing in Teams points at the certificate: you find out from users, not from a dashboard. Engineers call it the "scream test." Tracked as Microsoft Message Center MC1235747.

Feb 16
Microsoft test endpoint live for pre-deployment validation
Feb 28
Industry remediation deadline: firmware + root imports done
Mar 31
HARD deadline: legacy roots rejected outright
April
Microsoft rolls new certs onto the primary SIP proxy; unpatched peers drop
Jun 2026
Public CAs issue serverAuth-only certs; the SBC leaf must carry the Server Authentication EKU
We are here
Next forcing event · June 2026
--days
--hrs
--min
--sec
ENFORCEMENT LIVE
The 2026 root-CA deadline passed end of March; the serverAuth-EKU enforcement lands this month. Then the next one comes, and the one after that. This is the cadence, not the climax.
The cost of getting it wrong

It is already happening, in production.

Voipcloud
P1 · thousands of users

Microsoft stopped accepting their calls: a temporally valid but untrusted root presented at the mTLS handshake. Full outage across the customer base.

CallTower
Silent blackhole

Unpatched SBCs stopped sending SIP OPTIONS pings. Microsoft auto-deactivated the Direct Routing domain and blackholed every voice route, with no SIP 503 to warn anyone.

FusionConnect
One-way audio

Signaling succeeded, but media relays presented new DigiCert certs and silently dropped outbound audio (SRTP mismatch). The hardest class of failure to diagnose.

Regulatory

MiFID II / FINRA fines

An unrecorded trader call is a severe, heavily fined regulatory violation. If recording fails, the SBC must block the call outright.

Life-safety

99.999% or lives

Healthcare and emergency triage architectures demand five-nines uptime, where a failed call can literally mean the loss of a life.

Contractual

24/7 carrier SLAs

Tier-1 carriers are bound by strict around-the-clock SLAs. One core cert miss cascades into a mass tenant outage and SLA penalties.

Practitioners rate SBC config complexity 6 to 8 out of 10, and MTTR runs 4 to 12+ hours. Teams still learn about trunk outages from user complaints and LinkedIn posts. Until now, no cross-vendor, pre-deployment layer existed to catch this before it ships.

Where it sits

One independent layer that reads every vendor.

Every SBC vendor ships a tool, and every tool sees only its own hardware. Packet capture sees everything and reads nothing without an expert. Between them is the gap no one fills: an independent, cross-vendor layer that checks the config before it ships. That is what this is.

The estate

Your SBCs

AudioCodes, Cisco, Ribbon, Oracle, Metaswitch. The voice-aware firewall at Layer 5 that every call passes through, and every vendor configures differently.

The independent layer

SBC-AutoOps

Reads any vendor's config before deploy and tells you, in plain English, exactly what will break. Never touches the live box.

Single-vendor

Vendor tools

OVOC, RAMP, Element Manager. Each sees only its own hardware, blind across a mixed fleet.

Manual

Packet capture

Wireshark sees everything and reads nothing without an expert. Not something an ops team runs before every change.

What we are not: we are not an SBC, and we do not compete with AudioCodes, Cisco, Ribbon, or Oracle. They are channel partners, not competitors. We are the independent layer that reads all of them.
What you get

Outcomes, not output.

Zero-outage deployments

Validate certs, trust stores, and HA pairs before production. The silent failure never ships.

MTTR from hours to minutes

Replace 4 to 12 hours of manual log forensics with an instant readout and the exact thing to fix.

One pane, every vendor

AudioCodes, Cisco, Ribbon, Oracle, Metaswitch, on-prem or cloud, in a single diagnostic layer.

SLA protection

Catch config drift before it breaks the uptime commitment your contracts and penalties hang on.

Self-serve independence

Stop escalating to vendors, ITSPs, or Microsoft support to learn what your own config already knows.

Security becomes the reason to buy

Local-first and air-gapped: raw configs never leave your environment. The security review that blocks every other tool becomes the reason to adopt this one.

How it works

Validate the config before the call, not the call after the outage.

Five real vendor parsers normalize any config into one model, and the engine runs validation domains over it. Deterministic verdicts, not LLM guesses. It runs inside your environment, fully air-gapped.

01 · Validate

Catch it from config

Eight domains: syntax, interop, TLS and CA (the 2026 roots, EKU, SRTP, trust-anchor chains), NAT and one-way audio, codec, topology leak, routing, security. The failure is caught before a single packet is sent.

02 · Simulate

Predict the call

Models the call as a chain (TLS, then SIP, then SDP, then media), predicts exactly how far it gets, names the user-visible symptom, and renders the SIP ladder up to the point of failure.

03 · Explain

Diagnose a capture

Reconstructs the SIP ladder from a packet capture, detects one-way audio and TLS alerts, and maps each failure back to the config change that fixes it.

Plus HA-drift diff against a known-good baseline, fleet readiness reports, and a CI/CD gate that catches non-compliant config before it reaches the change window, not after.

SBC-AutoOps architecture: four vendor SBC configs (AudioCodes, Cisco CUBE, Ribbon, Oracle Acme) are normalized into one vendor-neutral model, validated across eight domains in three modes, producing a PASS / REVIEW / BLOCK verdict. The whole pipeline runs air-gapped inside the customer VPC, with only a signed rule bundle inbound and only anonymized findings outbound. (A fifth vendor, Metaswitch Perimeta, has since been added.)
One independent layer, five vendors, eight domains. Configs never leave your environment: the only thing that crosses the boundary inbound is a signed rule bundle, and the only thing that can leave is anonymized findings you opt into.
See it run

One broken SBC. One fixed SBC. Run both.

This is real engine output, replayed in your browser. Run the broken config and the validator blocks it, then predicts where in the handshake the call dies. Run the fix and the same SBC passes with two-way audio. That closed loop is the product.

sbc-validator · live walk

Real sbc-validator walk output. The engine runs locally and air-gapped; nothing here leaves your browser.

That was a replay. Now scan your real SBC's edge, free, no config upload. Scan your SBC free Bring this to your fleet
Why independent + pre-deployment wins

Everything else is single-vendor, or watches the call after it fails.

ApproachWhat it doesAgainst the 2026 deadline
Post-deployment AIOps Inspects active SIP/RTP for jitter, latency, MOS Blind
The failure drops TLS before signaling exists. You cannot monitor a call that is cryptographically blocked from ever starting.
Vendor tools (OVOC, RAMP, EM) Manage their own hardware only Siloed
Reality is multi-vendor: LoopUp runs Ribbon + AudioCodes; Societe Generale runs Oracle + AudioCodes + Ribbon. No single pane exists.
SBC-AutoOps Validates any vendor's config before deploy, then predicts the call Catches it
The missing root, the wrong EKU, the SRTP gap: from the config, before traffic, across the whole mixed fleet.

And why it stays hard to copy

Moat 01

Vendor-agnostic Layer 5

True SIP and SDP normalization across mixed estates. A vendor is structurally biased to its own hardware; it cannot build the layer that reads everyone's.

Moat 02

Security-first, local-first

Air-gapped, local execution removes the raw-config-exfiltration objection that kills every other tool in security review. The review stops blocking adoption and starts driving it.

Moat 03

A compounding data moat

To be unambiguous for your security team: the raw config never leaves. What can leave, only if you opt in, is anonymized findings (check IDs and severities, never config text or IPs). Those become a cross-vendor benchmark no single-vendor tool can assemble. The pattern compounds; your configuration never does.

The market

Large, urgent, and structurally underserved.

Direct Routing saves enterprises 50 to 75 percent on global telephony versus Microsoft Calling Plans. That entire saving, and the uptime it rides on, depends on the SBC being right. SBC-AutoOps is cheap insurance on a high-value, fragile, compliance-bound setup.

$2.4BSBC market by 2035, from $1.1B in 2025 (~8% CAGR)
$100B+UCaaS market, growing 15–20% a year
20M+users on Teams Direct Routing for PSTN calling
50–75%telephony savings that ride on the SBC working

Who we sell to

Tier-1 carriers · SBCaaS

AT&T · Verizon · BT

Multitenant cloud SBC fleets. One cert mismatch cascades into simultaneous outages across hundreds of enterprise tenants.

Cloud-voice MSPs · OC

SIPPIO · CallTower · NWN Carousel

API-driven virtual-SBC fleets behind white-label voice. A failure severs the downstream channel revenue.

Systems integrators

Continuant · SoftwareOne · Insight

Own the multi-vendor PBX-to-Teams migration, and the responsibility that hybrid estates survive the cryptographic shift.

Enterprises

50+ mixed-vendor SBCs

Direct Routing or BYOC estates with no cross-vendor, pre-deployment check before every change window.

Who cannot afford to get this wrong

The orgs that run their own SBCs do it because they have to.

Compliance keeps them off managed cloud calling, so they own the certificate problem directly. For them, downtime is regulatory or life-safety, not inconvenience.

Financial · MiFID II / FINRA

Societe Generale · TP ICAP

HA Oracle clustering for trader voice with WORM-compliant recording; multi-vendor cloud migration on AudioCodes + Cisco CUBE. If TLS fails, unrecorded calls must be blocked, or it is a fine.

Health · Public safety · HIPAA / CJIS

UCF · Falck EMS

CJIS-compliant police routing, HIPAA clinics, and global emergency triage that cannot drop a packet. Survivable branch appliances must trust the new DigiCert roots.

Defense · Manufacturing

US DoD · Bosch Group

JITC-certified Ribbon with FIPS-140-2 at the DoD; HA AudioCodes pairs running Local Media Optimization across Bosch factory floors worldwide.

What runs today

A working product, not a slide.

4 real vendor parsers, one model 8 validation domains live 126 automated tests in CI 7 Microsoft roots, signed authority-sourced rules air-gapped Docker image validate · simulate · explain · diff · fleet

The honest gaps, stated up front

Routing and security checks for Cisco, Ribbon, and Oracle, per-config cipher matching, and live probing all stay silent until they are validated against a real config for that vendor. The tool refuses to guess. A wrong verdict, telling a customer to fix the wrong thing, is the one thing this product cannot afford. That discipline is the difference between a diagnostic and a liability.

Where it goes

The deadline is the entry point. The lifecycle is the product.

Everything in the first column ships today. Everything after it earns its way in the same way the rest of the tool did: validated against real configs, never guessed.

Shipped · running today

The validator, in production form

  • Local-first validator: 4 vendors normalized to one model
  • 8 validation domains, including the 2026 CA and EKU checks
  • Validate, simulate, and explain, plus HA-drift diff
  • Fleet readiness reports and a CI/CD gate
  • Signed rule bundles, air-gapped Docker image
  • 126 automated tests in CI
Next · with design partners

Depth across the whole fleet

  • Routing and security checks live on Cisco, Ribbon, and Oracle; each needs one real config
  • Per-config cipher matching
  • The cross-tenant readiness benchmark
  • Deeper CI/CD and change-management integration
Vision · the lifecycle

The independent layer for the whole SBC lifecycle

  • Continuous drift detection across the fleet
  • An LLM explainer for SIP ladders
  • Assisted remediation, applied by engineers, never auto-pushed to production
  • STIR/SHAKEN checks and a deepfake-voice sidecar

Next and Vision are roadmap, not product. Each check stays silent until it is validated against a real config for that vendor.

The team · Metaventions AI

Telecom domain depth, meet AI build velocity.

Co-founder · Telecom

Philip Drammeh

Ex-Microsoft Telecom Spec Lead with deep SBC and Teams Direct Routing architecture expertise: the ground truth every validator is modeled against.

Co-founder · AI builder

Dico Angelo

AI builder and systems architect. Builds production AI infrastructure and autonomous diagnostic platforms; ships the engine and the local-first distributable.

The ask

Two design partners.
One real config per vendor.

We are partnering with one or two MSPs or enterprises running 50+ SBCs. We need one real, sanitized config per vendor: that is what turns routing and security checks live for that platform, the same path that made AudioCodes real.

In exchange: a 2026-migration readiness audit across your entire mixed-vendor fleet, run air-gapped inside your environment, before the next failure finds you.