Routes go bad at the worst times. A carrier's quality slips overnight, a rate changes mid shift, one prefix starts failing, and margin quietly leaks until someone notices and fixes it by hand. The latest TelecomsXChange (TCXC) release is built to close that gap.
Three things are new on the platform: per-prefix subrules and policies, routing by both the called and the calling number, and full Model Context Protocol (MCP) support, so a secure AI agent can manage your routing around the clock inside guardrails you define. You stay in control.
What's new
- Advanced subrules. Split a routing group by prefix and give each prefix its own routing policy. Two destinations in the same group can route in completely different ways.
- Route by destination and by caller. Choose a carrier based on where the call is going (CLD), where it is coming from (CLI), or both conditions together.
- Full MCP support. Connect a secure AI agent that watches your routing around the clock, proposes the best move, and acts inside the limits you set.
- Live context for every decision. The agent reads real supplier rates, quality stats, and CDRs from your own traffic, and it can test a route before it commits, so each decision reflects what is happening on your traffic right now.
CLD and CLI in plain terms
Every call works like an envelope. It has a "to" and a "from."
- CLD is the called number, the "to," the number being dialed. Routing by CLD means choosing a carrier based on where the call is going. For example, every call to UAE mobile, which starts with 9715.
- CLI is the calling number, the "from," the caller ID. Routing by CLI means choosing a carrier based on where the call is coming from. For example, calls that originate from a specific country or a specific customer.
Routing on both at once is where it gets useful. You can write a rule that reads "send all UAE mobile traffic to my best carrier, but if the caller ID belongs to my banking customer, send only that traffic down a premium route that guarantees CLI delivery." Looking at destination and origin together is how operators protect quality, guarantee CLI delivery, and manage cost and fraud at the same time.
How operators put this to work
1. Guaranteed CLI for high value calls, best price for everything else
A carrier terminates a large volume of UAE mobile traffic. Most of it is ordinary retail voice, where the cheapest route that still meets quality wins. A small slice is one time passcode and verification calls from a banking customer, where the caller ID has to arrive intact or the call is useless. With a CLD plus CLI subrule, all UAE mobile traffic (CLD starting 9715) takes the least cost route that passes quality, except calls carrying the bank's CLI range, which are pinned to a CLI certified carrier even though it costs more. One routing group, two outcomes, no second trunk to maintain.
2. A carrier degrades at 2am and the route corrects itself
A supplier's Answer Seizure Ratio (ASR) on UAE mobile falls from the mid forties to the high teens in the middle of the night. The AI agent sees the drop in your live CDRs, checks it against that supplier's current rate and quality, runs a test call down a backup carrier to confirm it answers cleanly, and proposes the switch. Within the guardrails you set, it moves UAE mobile to the backup, logs the change, and leaves a record for the morning team to review. The margin is protected before anyone reads the alert.
3. One group for a whole region, a different policy per country
A routing group covers Middle East mobile. UAE mobile needs premium quality because the customer pays for it. A neighbouring destination in the same group is price sensitive and is fine on a cheaper route. In the past that meant either splitting the group or keeping parallel rules in sync by hand. With subrules a single group holds both: UAE mobile on the quality carrier, the price sensitive prefix on least cost, each with its own policy.
An AI agent on your routing desk, with the whole stack as context
With MCP support you can connect an AI agent directly to your interconnect. What makes the agent worth having is not that it can change a route. It is that it has the context to change the right route for the right reason.
Through the unified TCXC wholesale stack, one agent can reach all of this at once:
- Read and write access to your routing groups, so it understands the current setup and can apply an approved change.
- Real time rates from every supplier, so it knows the current cost of each path instead of working from last week's rate sheet.
- Your CDRs, so it sees ASR, average call duration, and failures on your own live traffic as they happen.
- Route testing tools, so it can place a real test call down an alternative carrier and confirm it works before moving production traffic onto it.
That combination is the whole point. An agent that can only edit a route is guessing. An agent that reads live rates, reads live CDRs, tests the alternative, and then edits the route is making an informed decision on the spot, with the same information a senior routing engineer would collect, gathered in seconds rather than over an afternoon. Your engineers move to handling exceptions instead of watching dashboards. This is what the unified TCXC wholesale stack makes possible: the decision and the data live in the same place.
In practice you work with it in plain language. You describe the outcome you want, the agent reads the live stack, drafts the exact subrule change, tests it, and waits for your approval before anything goes live. This is what autonomous carrier wholesale routing looks like in the age of agentic AI: the work is automated, the decision stays with you.
You stay in control
Routing touches revenue, so this is supervised by design. The agent works inside guardrails you set. Changes are proposed, tested, and applied only on your approval. Every action is logged. Any change can be rolled back. The work is automated. The decision stays yours.
Talk to the team
We are opening this to carriers and aggregators who want to put it to work first. To see per-prefix subrules, CLD and CLI policies, and MCP routing running on your own traffic, contact the TelecomsXChange team. You can also read more about the TCXC MCP Server and the wider wholesale platform.
TelecomsXChange, the AI native interconnect.
