AI-powered routing & revenue optimisationAI-based call routing: ML-driven least cost routing and quality scoring.
Routing powered by intelligence
Traditional LCR engines rely on static rate tables and manual route configuration. Our AI-powered routing platform goes far beyond that: it learns from billions of historical CDRs, analyses real-time network conditions, and makes dynamic routing decisions that optimise simultaneously for cost, quality, and revenue. The result is higher ASR, lower PDD, better ACD, and more profit on every minute.
Neural network engine
Our deep learning model is trained on billions of CDRs spanning years of global traffic data. It identifies patterns that are invisible to traditional routing logic, predicting route quality minutes before degradation occurs and adjusting routing tables proactively.
Real-time scoring
Every active route is continuously scored across multiple quality dimensions: ASR, ACD, PDD, MOS, and CLI delivery rate. Scores are updated in real time as new CDRs flow in, ensuring routing decisions always reflect the current state of the network.
Revenue optimisation
Go beyond cost minimisation. Our algorithms balance the cost of termination against the revenue potential of each call, factoring in customer SLAs, quality commitments, and margin targets to maximise your profitability on every single minute.
Smart Routing and Revenue Turbine, on PacketExchange.
Smart Routing
The cheapest working route, call by call, with failover inside Switch route groups.
Revenue Turbine
Revenue-boosting route blending: several routes become one sellable product, with one weighted rate, pooled capacity and a blended quality profile.
Measured quality
ASR, ACD and NER measured on real buyer calls, with route reports and an automatic re-test when a seller marks a fix.
How our AI transforms routing
Six pillars of intelligence that work together to deliver superior routing performance.
ML-driven least cost routing
Our LCR engine replaces static rate-based routing with a dynamic, multi-dimensional optimisation model. For each call, the system evaluates every available route across cost, quality metrics, historical performance, current load, and time-of-day patterns, selecting the optimal path in under 10 milliseconds. The model retrains continuously on new data, adapting to market changes faster than any human team could manage.
- Multi-factor route evaluation beyond simple cost comparison
- Sub-10ms routing decisions for zero-impact call setup
- Continuous model retraining on live production data
- Configurable optimisation targets: cost, quality, or balanced
Predictive analytics
Our predictive models forecast route quality up to 30 minutes into the future by analysing trends in ASR, PDD, and error codes. When the model detects early warning signs of degradation, such as a gradual increase in PDD or a subtle decline in ASR: it preemptively shifts traffic to alternative routes before your calls are affected. This proactive approach prevents quality issues rather than reacting to them.
- 30-minute quality forecasting for proactive route management
- Early warning detection for network anomalies and outages
- Traffic volume prediction for capacity planning
- Seasonal and time-of-day pattern recognition
Automatic failover
When a route fails or degrades beyond acceptable thresholds, our system instantly reroutes traffic to the next best alternative, with zero manual intervention and typically zero dropped calls. The failover engine maintains a continuously updated ranked list of backup routes for every destination, pre-validated and ready to receive traffic at a moment's notice.
- Sub-second failover with zero call drops in most scenarios
- Pre-ranked backup routes for every active destination
- Configurable quality thresholds that trigger automatic failover
- Automatic route recovery and traffic restoration on issue resolution
Optimising every quality dimension
Our AI simultaneously optimises across all key performance indicators that define voice quality and profitability.
ASR optimisation
Answer-Seizure Ratio optimisation ensures the highest possible call completion rates. Our AI learns which routes deliver the best ASR for each specific destination and time of day, continuously improving connection success rates across your traffic.
ACD enhancement
Average Call Duration analysis identifies routes that deliver longer, higher-quality conversations. Short ACD often indicates audio quality issues or one-way audio: our AI detects these patterns and routes away from problematic suppliers automatically.
PDD reduction
Post-Dial Delay is minimised by routing through suppliers with the fastest call setup times. Our system tracks PDD per route in real time and penalises slow routes in the ranking algorithm, ensuring your callers hear ringing as fast as possible.
MOS scoring
Mean Opinion Score estimation using machine learning models that predict audio quality based on network metrics. Routes with consistently high MOS scores are prioritised for premium traffic, while degrading routes are automatically demoted.
CLI delivery
CLI pass-through rates are tracked per route and supplier. Our AI ensures calls requiring CLI delivery are only routed through suppliers with verified, tested CLI pass-through capability, never through routes that strip or modify the caller ID.
Cost vs quality balance
A configurable slider between pure cost optimisation and pure quality optimisation lets you define the perfect balance for each customer or traffic type. Premium customers can be routed on highest-quality paths while bulk traffic optimises for margin.
Beyond traditional routing
Capabilities that give you a competitive edge in the wholesale voice market.
A/B route testing
Scientifically evaluate new routes by splitting a controlled percentage of traffic between your current routing and a test configuration. The system measures performance differences with statistical significance and recommends the winning route, eliminating guesswork from route management.
Traffic pattern analysis
Deep analysis of your traffic patterns reveals optimisation opportunities. The system identifies peak hours, seasonal trends, destination shifts, and anomalies, then automatically adjusts routing strategies to match. Weekly insight reports highlight revenue opportunities you may be missing.
Dynamic supplier selection
Suppliers are ranked dynamically based on real-time performance, not static preferences. A supplier delivering excellent quality today but degrading tomorrow will be automatically deprioritised, and restored when performance recovers. No manual intervention required.
Historical performance analysis
Access years of historical routing data with powerful analytics tools. Compare route performance over time, identify long-term trends, benchmark supplier quality, and generate evidence-based reports for rate negotiations and capacity planning.
The AI routing pipeline
A look inside the decision-making process that happens in under 10 milliseconds for every call.
Ingest
The call arrives and is classified by destination, customer profile, quality requirements, and margin targets. The routing engine fetches all available routes for the destination prefix and current real-time quality scores.
Score
The neural network model scores each candidate route across cost, ASR, ACD, PDD, MOS, and CLI delivery. Scores are weighted according to the customer's configured optimisation profile and current SLA commitments.
Select
The highest-scoring route is selected as primary, with the next three routes pre-loaded as failover candidates. If the primary route fails the call attempt, failover happens instantly to the next candidate with no additional scoring delay.
Learn
After the call completes, the CDR feeds back into the training pipeline. The model updates its route quality predictions, the historical database grows richer, and the next routing decision benefits from this new data point.
Measurable impact
Potential results based on platform capabilities from traditional LCR to our AI-powered routing engine.
Up to +12%
ASR improvement
Typical ASR improvement within the first 30 days of switching to AI routing.
Up to -40%
PDD reduction
Typical reduction in post-dial delay by routing through the fastest-connecting suppliers.
Up to +18%
Margin increase
Typical revenue margin improvement through intelligent cost-quality balancing.
<10ms
Decision time
Average time for the AI engine to evaluate all routes and select the optimal path.
Maximise every minute
Our revenue optimisation algorithms ensure you extract maximum value from every minute of traffic flowing through your network.
Margin-aware routing
The routing engine knows your sell rates and buy rates for every destination. It factors margin into every routing decision, ensuring you never route a call at a loss unless explicitly configured as a strategic decision.
Volume-Tier optimisation
Automatically direct traffic volumes to meet supplier commitment tiers and unlock better rates. The system tracks your volume progress against each supplier agreement and adjusts traffic distribution to maximise tier discounts.
Revenue dashboards
Real-time visibility into your revenue, costs, and margins per destination, customer, and supplier. Identify your most profitable routes, your highest-margin customers, and opportunities to improve underperforming traffic segments.
