Run AI across every cell site and core. Govern it from one control plane.
CDRs, signaling, and subscriber data are licensed to stay inside the operator's environment. Iftah runs models inside the network core, edge, and BSS/OSS — one control plane enforcing residency, governance, and audit across every cloud and on-prem domain, at sub-millisecond latency where the 5G core and MEC edge demand it.
Built for licensed network operators
- CDRs stay in the licensed environment
- Mapped to CST CRF, NCA ECC, PDPL
- Sub-ms inference at core & MEC edge
- Air-gapped lawful-intercept & OT
Designed for the regimes that license the network.
Iftah does not certify compliance — that stays the operator's obligation. In-licence deployment, edge isolation, and tamper-evident lineage give your assessors concrete evidence.
CST CRF
Telecommunications Act and Cybersecurity Regulatory Framework for the ICT sector.
NCA ECC / CCC / CSCC
Essential, Cloud, and Critical-Systems controls for Critical National Infrastructure.
TDRA + UAE PDPL
TDRA regulation with federal personal-data protection.
CRA + PDPPL
Communications Regulatory Authority with the National Data Privacy Law.
CITRA
Communication & IT Regulatory Authority and its data-privacy regulation.
TRA + PDPL
National Telecommunications Regulatory Authorities and personal-data laws.
Operator AI objections, resolved by design.
CDRs leave the licence
SaaS AI requires CDRs, subscriber location, and billing data to leave the licensed environment — breaching CST, TDRA, CRA, and CITRA conditions.
Every model and endpoint is pinned to a chosen jurisdiction; CDRs and subscriber data never leave the licensed environment.
CNI controls unmet
Critical National Infrastructure status mandates OT/network controls multi-tenant AI can't evidence.
Tamper-evident audit and lineage are mapped to NCA ECC and CST CRF evidence requirements for CNI reviews.
Latency breaks RAN & fraud
RAN optimization and fraud blocking need sub-millisecond inference a distant cloud can't meet.
Low-latency inference runs on bare-metal GPUs at the 5G core and MEC edge, with air-gap for lawful-intercept and OT.
Network IP becomes training data
Topology, traffic models, and pricing logic feed vendor models, eroding competitive IP.
Models and fine-tuning stay private to the operator; proprietary network data never trains a shared model.
Fragmented per-market AI
AI deployed per cloud and per market yields no unified policy, lineage, or audit across the footprint.
One control plane governs models, access, and guardrails across every cloud, on-prem, and market.
Core, edge, and BSS/OSS — one governed plane.
Match each function to its domain — and govern all of it from one plane.
Sovereign / telco cloud
BSS/OSS and customer AI in the operator's licensed jurisdiction.
Multi-cloud per market
Markets across AWS, Azure, OCI, and private telco cloud governed as one.
Bare metal at core & edge
Sub-ms inference on operator GPUs next to the 5G core and at MEC edge sites.
Air-gapped LI & OT
Isolated lawful-intercept and network OT with no external data path.
Network and customer AI, in-licence.
- RAN & network optimization
Predictive capacity, energy-saving cell shutdown, and self-healing fault detection at the edge.
- Fraud & revenue assurance
SIM-box, subscription, and bypass-fraud detection on live signaling and CDR streams in-network.
- GenAI customer care
Bilingual Arabic-English agent assist, call summarization, and deflection with PII guardrails.
- Churn & next-best-action
Across BSS/CRM on first-party subscriber data without external export.
- Field & NOC copilots
Retrieval over runbooks, tickets, and topology for faster incident resolution.
- SOC automation
Anomaly detection across network telemetry aligned to NCA ECC and CST CRF.
Next step