At this year’s Capacity Europe, I sat on a panel called “Put Your Money Where The Revenue Is.” Together with me fellow panelists, we quickly drifted towards AI.
I remember just a year ago I was saying that AI is like teenage sex – everyone is talking about it, but very few are actually doing it, let alone doing it right.
However with the rapid pace of development, AI today is at a whole different stage of maturity. The exciting news is that we see more and more B2B use cases coming to life.
The point of the session was very direct: if you are going to spend money on AI, spend it where it moves revenue or reduces costs.
We were invited to that panel as practitioners in lead-to-cash automation for wholesale carriers. So I didn’t go on stage to talk about “the future of AI.” I talked about what’s already live, in production or in a PoC phase, tested with results. Below is a summary of what we have developed as Enxoo and we see as true game changers for Carriers.
The quoting agent
This is one of the most effective places we’ve seen AI create measurable value in wholesale. We have seen the rise of portals and APIs, which are absolutely great. But still the majority of quote requests are being received by email. And now we have an agent to handle just that.
Here is the normal reality today for most carriers:
- A customer emails: “Can you quote me for 1 Gbps between London and Amsterdam for 12,24 and 36 months?”
- An account manager reads it, retypes the request into CRM/CPQ
- If it’s an extensive Excel file, they spend time getting it into an uploadable format
- Presales checks whether that route is standard or off-net
- They pull the pricing, apply the discount policy, and check margin
- Someone formats it into an official quote or offer
- Someone sends it back to the customer
- Hours or even days go by
None of that is strategic work. It’s admin. We’ve now put an AI agent in the middle of that process.
The agent:
- reads the inbound email
- understands what the customer is asking for (endpoints, bandwidth, term)
- looks up the right product SKU from the catalog
- applies the correct pricing and margin rules
- generates a clean quote that’s ready for human review and approval
The outcome: for one account manager, it gave back around 35 hours per week of manual work. That is basically a working week of “copy/paste, format, resend” removed. The other outcome: response time went from “we’ll get back to you” to “here’s your quote” in minutes.
Intelligent off-net quoting
The second case is off-net quoting. Anyone in wholesale knows this pain.
If you’re delivering a service where part of the path depends on a partner network, you have to stitch together:
- partner availability (can they serve this location?)
- partner cost
- SLA terms
- your own commercial margin and contractual rules
Traditionally, this is slow. You email or message partners, you wait for answers, you merge the answers into your own pricing model, and only then you go back to the customer. Sometimes it takes days. Sometimes the customer has already gone elsewhere.
Now we’re changing this. An AI agent handles the process end-to-end: The agent calls the partner-facing APIs or interfaces (ideally standardised, such as MEF LSO Sonata), retrieves availability and cost data, applies your pricing and margin logic, and instantly generates an off-net quote ready to send to the customer.
But the challenge isn’t just about speed. It’s also about data diversity and orchestration. Today, carriers have to deal with multiple data sources: different partner portals, APIs, spreadsheets, marketplaces, and historical cost data. The real art lies in using them intelligently and agilely: managing complex business rules, aggregating inputs from many vendors, and learning from past outcomes to continuously improve.
That’s where intelligent off-net quoting becomes truly powerful: a multi-source, AI-driven, rule-aware system that selects the best option across vendors in real time, ensuring both competitiveness and profitability.
“My Internet is slow”
This one is more on the service / ISP side, but I like it because it illustrates the capabilities to connect to a CPE and orchestrate a series of certain actions. One of the most common inbound requests to an ISP is the phrase: “My internet is slow” or “My Wi-fi is not working”.
Historically that means: open a ticket, talk the customer through basic steps, maybe dispatch someone.
Now we’re seeing an agent that is connected directly to the customer’s CPE (the router or access device). The agent can work together with monitoring systems:
- run diagnostics,
- push basic fixes (restart, channel change, config update),
- confirm whether performance is back to normal,
- If it’s a matter of moving closer to the device, recommending to do so
- and if the device itself is underpowered for what the customer is trying to do, recommend an upgrade
- Even complete the purchase and add it to the customers plan
This reduces load on support teams and truck rolls. It also quietly creates a controlled upsell path when the device really is the bottleneck.
Why is this relevant to wholesale? Because it shows the same pattern: an agent that is allowed to act, not just “answer a question.” That is the point. AI is moving from “assistant” to “operator.” This case is relevant to any kind of CPE.
Now, let’s talk about how all of this is possible technically, because this is where a lot of people misunderstand it.
From proprietary APIs to AI wrappers on standards
For years, the way carriers integrated systems was to build proprietary APIs and wire them together one by one. You had an API for quoting, another one for inventory, another one for order status, another one for partner access, and so on. Every time you wanted to automate something new, you wrote a new connector or you extended an old one.
That approach is rigid, slow to maintain, and expensive to scale. What’s changing now is not “no APIs.” APIs still matter. What’s changing is how they’re used. The industry is slowly aligning around API standards in areas like product, ordering, serviceability, and partner interaction — for example, MEF-style interfaces in wholesale for quoting, ordering and service delivery between carriers.
Once you have those standard interfaces in place, you can put an AI wrapper- an agent- on top of them.
Here is what that actually means in practice:
- Instead of a human reading an email, asking engineering for availability, asking procurement for partner cost, asking finance for target margin, and then writing a quote in Word…
- The agent understands the request (“1 Gbps London ↔ Amsterdam, 36-month term”), calls the product catalog via standard API, calls the partner access API for off-net cost and SLA, applies your margin rules, generates the quote, and returns it for human approval.
So the logic that used to live in people’s heads and inboxes now lives in the agent. And the agent is only able to do that because those systems expose predictable, structured interfaces. This is what we mean by “AI wrapper.” The agent sits above the systems. It orchestrates them. It doesn’t replace them.
This allows you to stop hardcoding every single process flow. The amount of “Excel plus tribal knowledge” you need to get a quote out the door is significantly reduced.
This is also why standards like MEF are important. The cleaner and more consistent the interface between you and your partners, the easier it is for an agent to generate an off-net quote without waiting for someone to email somebody else.
So when we say “move from proprietary APIs to AI wrappers on API standards,” we’re not saying throw away your systems. We’re saying: expose them in a standard way, and then let an agent sit on top and run the process end to end.
That’s the architectural shift that unlocks quoting in minutes instead of days.
The Agentic Carrier
It’s not a marketing label. It’s a carrier where the core workflows in sales, delivery, and service are handled by AI agents that can interpret intent and act across systems.
The next-generation carrier is not only automated- it’s autonomous. This marks a new era for wholesale, where we can finally optimize complex processes that have remained manual for years — simply because wholesale was considered too difficult to automate until now.
The goal is not to replace people. The goal is to take the repetitive work off them so they can spend time on higher-value work: managing the relationship, structuring bigger deals, fixing real problems.
To be clear: AI is no longer a hype. It’s already in production in wholesale quoting, off-net sourcing, and first-line service. If you’re going to invest, invest there first. That’s what “put your money where the revenue is” actually means.
