ConnexAI (Athena) is a US-based cloud SaaS platform with AI voice agents, analytics, quality scoring, and a knowledge layer for contact centres. SARAH AI Suite is the complete inverse: SARAH — the Voice Conversational AI — lives at the customer's premises on an NVIDIA DGX B300 appliance, running its own sovereign LLM inference stack (Kimi K2.5-2.6, Gemma 4, HRM — 235B / 72B / 32B / 8B). SOPHIA — the Layer-3 API Hub with 34,792,085 Live Enterprise Features, Connectors and APIs — lives in the Data Centre. A Megaport 400GE dedicated private Ethernet circuit connects the two. Airlines, Banks, Telcos, Government, Healthcare, Hotels, Logistics, Insurance, and 1,000+ enterprise verticals. Zero borrowed code. Zero shared cloud infrastructure. Zero ongoing SaaS dependency. This brief documents exactly what ConnexAI delivers, absorbs every capability into SARAH's equivalent, and shows what SARAH delivers beyond it.
ConnexAI's entire product surface — Athena AI Agents, AI Voice, AI Analytics, AI Quality, AI Knowledge, Omni-Channel, and Flow automation — is contained within SARAH Voice Conversational AI as a subset. Each module is documented below with its SARAH equivalent and the dimension on which SARAH exceeds it.
ConnexAI's Athena handles outbound prospecting, lead qualification calls, appointment booking, and inbound inquiry resolution. SARAH Voice does all of this natively with multi-turn conversational reasoning across a sovereign multi-model stack (Kimi K2.5-2.6, Gemma 4, HRM 235B/72B/32B/8B), dynamic objection handling, real-time CRM writes, and autonomous scheduling — deployed on the customer's premises, not a shared US data centre. Every call, every intent, every outcome owned and auditable by the enterprise.
ConnexAI claims 100+ language support via cloud ASR/TTS. SARAH Voice runs local high-fidelity ASR and neural TTS on the DGX B300 appliance — no audio leaving the enterprise boundary. Multilingual switching mid-call, accent normalisation, domain-specific vocabulary injection (aviation, banking, medical), and SSML-level prosody control. Zero latency dependency on external speech APIs.
ConnexAI's AI Analytics surface offers conversational trend detection, sentiment scoring, and performance dashboards — all post-call, all cloud-processed. SOPHIA processes every interaction in-flight and post-call on-premises, with semantic clustering across the full 34.7M feature corpus, real-time sentiment feeds into supervisor dashboards, predictive churn modelling, and cross-channel correlation (voice + email + chat + WhatsApp simultaneously).
ConnexAI's AI Quality module scores call recordings against configurable rubrics and generates QA reports. SARAH's Quality Engine runs real-time compliance monitoring (not just post-call), automatic breach detection, full transcript + intent analysis, regulatory alert generation, and on-premises immutable audit logs. Banks, insurers, and government customers require the audit trail to never leave sovereign infrastructure — SARAH delivers this; ConnexAI cannot.
ConnexAI's AI Knowledge layer unifies CRM data and custom knowledge documents to ground agent responses. SOPHIA Layer-3 API Hub is a single Postgres-backed source of truth spanning all dispatcher modules, connectors, voice workflows, per-tenant data, and compliance attestation — 34,792,085 live entries — with real-time bidirectional sync to any CRM, ERP, government registry, airline GDS, banking core, or custom data source the enterprise operates.
ConnexAI supports omni-channel routing across SMS, live chat, WhatsApp, social, email, and voice — with PCI DSS payment capture. SARAH's Omni-Channel Orchestrator handles all of these channels with a unified context thread across every touchpoint, agent-handoff with full history, in-call PCI DSS payment flows, and a private WhatsApp Business API bridge — all on-premises. Channel parity with ConnexAI; infrastructure sovereignty beyond it.
ConnexAI Flow is a no-code automation builder for contact-centre workflows. SARAH's Workflow Engine goes further: trigger-based autonomous process chains across voice, messaging, CRM, payments, document generation, and compliance — with natural-language workflow definition (SARAH builds the workflow from a spoken description), cross-system orchestration across the full 34.7M connector corpus, and scheduled + event-driven execution. No drag-and-drop required; the AI builds the workflow.
ConnexAI is a US cloud SaaS with SOC 2, GDPR, HIPAA, and ISO certifications — compliance on shared infrastructure they manage. SARAH AI Suite operates in a split-sovereign architecture: SARAH lives at the customer's own premises, SOPHIA lives in the Data Centre — connected by a Megaport 400GE private Ethernet circuit. Data never traverses the public internet by design. Australian Privacy Act, EU GDPR Article 28, US HIPAA, PCI DSS Level 1, APRA CPS 234, ISO 27001 — all enforced at the infrastructure level, not contractually asserted by a SaaS vendor.
ConnexAI's marketplace claims "hundreds of popular app integrations." SOPHIA delivers 34,792,085 live enterprise features, connectors, and APIs — zero borrowed code — spanning Airlines (180 categories of GDS, loyalty, operations, cargo, and ground-handling connectors), Banks and Financial Institutions (236 categories of core banking, payments, FX, compliance, and lending connectors), and every other enterprise vertical in operation today. This is not a marketplace of third-party plugins. It is a hand-built, audited, production-grade enterprise intelligence network.
ConnexAI's customer base is contact-centre BPOs and mid-market operators. SOPHIA's deepest vertical coverage is in the sectors that demand the highest compliance, the most complex data integrations, and the strictest data sovereignty requirements — Airlines and Banks. These are not modules or add-ons. They are 416 industry categories of live enterprise connectors, hand-built to production standard.
Feature-for-feature assessment across the dimensions that matter to an enterprise technology decision.
| Dimension | ConnexAI (Athena) | SARAH AI Suite + SOPHIA |
|---|---|---|
| Deployment Model | US-based cloud SaaS — shared multi-tenant infrastructure | Split-sovereign architecture — SARAH (DGX B300 appliance) lives at the customer's premises. SOPHIA (Layer-3 API Hub, 34.7M features) lives in the Data Centre. Megaport 400GE dedicated private Ethernet connects the two. No shared infrastructure. No public internet. |
| Data Sovereignty | Data processed in ConnexAI's US data centres — contractual assurance only | Complete data sovereignty by architecture — no data leaves the enterprise boundary. Australian Privacy Act, GDPR, APRA CPS 234, HIPAA enforced at infrastructure level. |
| AI Voice Agents | Athena AI Agents — prospecting, lead qual, booking, inquiry — cloud-hosted, shared LLM inference | SARAH Voice Conversational AI — sovereign multi-model on-premises inference: Kimi K2.5-2.6, Gemma 4, HRM — 235B / 72B / 32B / 8B. Full model stack on DGX B300. Zero external API latency. Zero per-token billing. |
| LLM Infrastructure | Third-party cloud LLM APIs — provider, latency, cost, and model choice not disclosed | Own LLM inference stack — three-tier model architecture running on DGX B300. No API bills per token. Predictable cost, predictable latency, full model control. |
| Feature Depth | Hundreds of marketplace integrations — third-party plugins, not native connectors | 34,792,085 Live Enterprise Features, Connectors & APIs — zero borrowed code. Airlines (180 categories), Banks (236 categories), plus 1,000+ enterprise verticals. All hand-built, all production-grade. |
| Airlines Coverage | No airline-specific GDS, loyalty, cargo, IROPS, or aviation operations connectors | 180 airline connector categories — GDS (Amadeus, Sabre, Travelport), loyalty, codeshare, cargo, ACARS, revenue management, IROPS automation, ICAO/IATA compliance |
| Banking Coverage | Generic CRM connectors only — no core banking, no payment rails, no regulatory reporting | 236 banking connector categories — core banking (Temenos, Flexcube, Finacle), SWIFT/SEPA/NPP, KYC/AML, PSD2/CDR open banking, FX/treasury, APRA/ASIC/FCA reporting |
| Analytics | Post-call conversational analytics — cloud-processed, trend detection, sentiment scoring | Real-time + post-call analytics on-premises — in-flight sentiment, semantic clustering across 34.7M feature corpus, predictive churn, cross-channel correlation (voice + email + chat + WA simultaneously) |
| Quality Assurance | AI scorecards on recordings — rubric-based, post-call only | Real-time compliance monitoring — breach detection mid-call, automatic regulatory alerts, immutable on-premises audit logs, full transcript + intent analysis |
| Knowledge Layer | CRM unification + custom document ingestion — grounded responses from uploaded sources | SOPHIA Layer-3 API Hub — single Postgres source of truth, 34.7M entries, real-time bidirectional sync to any CRM, ERP, government registry, airline GDS, banking core, or custom data source |
| Omni-Channel | SMS, live chat, WhatsApp, social media, email, voice, PCI DSS payments — cloud-routed | All same channels + private WhatsApp Business API bridge — unified context thread across every touchpoint, agent handoff with full history, in-call PCI DSS flows — all on-premises |
| Workflow Automation | Flow — no-code visual builder for contact-centre workflows | SARAH Workflow Engine — natural-language workflow definition (SARAH builds the workflow from description), cross-system orchestration across all 34.7M connectors, scheduled + event-driven execution |
| Hardware Platform | No hardware — pure SaaS, customer has no hardware ownership | NVIDIA DGX B300 Appliance — customer owns the hardware. Full GPU compute for LLM inference, voice processing, and analytics. SARAH Spark 2 for single-location operators ($60,000 outright / $2,000 per month financed). |
| Energy Independence | Fully dependent on ConnexAI's cloud infrastructure and its energy sources | 5-Source Power Independence — (1) Magnetic Generator (zero fuel, zero emissions), (2) Hydrogen Generator BlueDM GEH2 (100 kVA, zero CO₂), (3) Microsonic Generator — ambient RF energy, (4) BLUETTI ES125 Battery Backup with Optional Solar System (257 kWh / 125 kW, LiFePO₄, scalable to 2,057 kWh), (5) Main Grid as final fallback. Near-zero recurring electricity opex. Customer owns the entire power stack. |
| Global Network | Cloud CDN and US data centre egress — latency dependent on geography | Megaport 400GE Dedicated Ethernet — SARAH lives at the customer's premises. SOPHIA lives in the Data Centre. All traffic between them runs over a dedicated private Megaport 400GE Ethernet circuit — no public internet, no shared infrastructure. WireGuard encrypted overlay available for branch and remote endpoints. |
| Vendor Dependency | Fully dependent on ConnexAI platform continuity, pricing, and roadmap decisions | Zero vendor lock-in — customer owns the appliance, the software stack, and the data. SARAH AI Suite continues operating indefinitely regardless of third-party decisions. |
| Code Provenance | Commercial SaaS — code provenance, security audits, and sub-processors not disclosed | Zero borrowed code — entire SARAH AI Suite and SOPHIA stack is hand-built by IDESKS ONLINE. Full source auditability for security, compliance, and regulatory review. |
| Target Clients | Contact-centre BPOs, mid-market operators — RefiJet, Exact Medicare, TIH/Telesure | Tier-1 and Tier-2 Banks, Airlines, Federal and State Government, Telcos, Global Enterprise — any organisation requiring sovereignty, depth, and scale |
| Implementation | Cloud onboarding — timeline and data migration risk determined by ConnexAI's process | 30-Day Implementation — hardware delivery, network integration, workflow configuration, and full go-live within 30 days. 36-month warranty included. |
ConnexAI has no hardware story. Every customer is a tenant on shared US cloud infrastructure. SARAH AI Suite is the opposite of that in every dimension.
ConnexAI's marketplace advertises "hundreds of popular app integrations." SOPHIA delivers 34,792,085 live enterprise features, connectors, and APIs — hand-built, zero borrowed code. The scale difference is not marginal. It is the difference between a contact-centre tool and a complete enterprise intelligence network.
Airlines (180 categories) and Banks (236 categories) alone represent more domain-specific integration depth than ConnexAI's entire feature catalogue. When a Tier-1 Bank evaluates an AI voice platform, they need core banking access, KYC/AML automation, and regulatory reporting — not a generic CRM connector. SARAH delivers this out of the box.
ConnexAI's compliance posture relies on contractual data residency commitments and third-party certification. SARAH AI Suite enforces data sovereignty at the infrastructure architecture level — the data physically does not leave the enterprise boundary. For regulated industries, this is not a preference; it is a procurement requirement.
ConnexAI's AI agents run on undisclosed third-party LLM APIs — cost, latency, model quality, and availability are outside the enterprise's control. SARAH runs a sovereign multi-model stack on-premises — Kimi K2.5-2.6, Gemma 4, HRM at 235B / 72B / 32B / 8B — all on DGX B300. The enterprise owns the inference stack, owns the cost profile, and owns the capability roadmap.
ConnexAI customers are permanent tenants. They own nothing — not the compute, not the software, not the data infrastructure. SARAH AI Suite customers own the NVIDIA DGX B300 appliance, the software stack, and the data. The 30-day implementation is a one-time event; the asset lives on the balance sheet.
SARAH AI Suite and SOPHIA are hand-built end-to-end by IDESKS ONLINE — every connector, every voice handler, every workflow engine, every analytics layer. Zero borrowed code means full source auditability, zero supply-chain compromise risk, and complete control over the security posture — a requirement for government and Tier-1 bank procurement.
ConnexAI's own performance claims — 10x interactions, 70% handle time reduction, 65% inbound reduction — describe a tool that augments human agents. SARAH AI Suite replaces the contact-centre headcount model entirely, on sovereign infrastructure, at a fraction of the total cost of ownership.
| Scale Point | Human Contact Centre | ConnexAI (Augmentation) | SARAH AI Suite (Replacement) |
|---|---|---|---|
| 10-Agent Operation | $40,000–$60,000/mo loaded labour | SaaS subscription + remaining human agents | SARAH Spark 2 — $2,000/mo financed. 2 concurrent AI voice calls. Zero agent headcount. |
| 50-Agent Operation | $200,000–$300,000/mo loaded labour | SaaS seats + reduced headcount | SARAH AI Suite on DGX B300. Full autonomous contact centre. One appliance. Unlimited concurrent AI voice capacity on-premises. |
| Data Sovereignty | N/A — internal systems | US cloud — contractual GDPR/HIPAA commitments | Architecture-level sovereignty. No data leaves the enterprise boundary. Required for banks, airlines, and government. |
| Hardware Asset | None — labour-based model | None — perpetual SaaS tenancy | NVIDIA DGX B300 on the customer's balance sheet. Appreciating operational asset. 36-month warranty. |
If your organisation operates in Airlines, Banking, Government, Telco, Healthcare, or any regulated enterprise vertical — and requires data sovereignty, sector-depth connectors, and a hardware asset on your balance sheet — this is the conversation worth having.
Book a Discovery Call