Telcos.
Cut churn and grow ARPU, on subscriber data that never leaves the perimeter.
In telecoms, growth is a retention game: winning subscribers, cutting churn, and lifting ARPU through personalised care and cross-sell. All of it runs on the subscriber graph, who connects to whom, where, and how. No other industry holds behavioural data this granular, and none can afford to hand the customer stack to a third-country model.
- R01Personalised marketing on traffic and location data (ePrivacy Directive)
- R02Subscriber-data lawful basis and transfer scrutiny under Schrems II
- R03Churn and next-best-action models trained on subscriber behaviour
What's actually
reshaping the sector.
The shifts our research is tracking, named, current, cited.
TM Forum reports that agentic-AI customer care can cut churn 15-25% and billing-related call-centre inquiries 50-70%, with a ~5% revenue uplift from better upsell and cross-sell. Care is now a growth lever, not a cost centre.
TM Forum →Vodafone's TOBi assistant handles nearly 45 million customer questions a month across 13 countries; after the SuperTOBi upgrade, Vodafone Portugal's first-time resolution rose from 15% to 60% and online NPS climbed 14 points. Agentic care is live, not a pilot.
Vodafone →McKinsey reports a leading European telco used generative AI to hyper-personalise upselling for a 5 to 15% increase in ARPU by segment. The subscriber graph is the input, and the reason it stays home.
McKinsey →Deutsche Telekom is building an EU AI factory with sovereign LLMs (Teuken/SOOFI); Orange and Mistral offer GenAI-as-a-service in a sovereign, secure framework. Sovereignty is monetizable, not just defensive.
Deutsche Telekom →In telecoms, the subscriber relationship, not the network alone, is where AI and sovereignty compound: churn, care, and cross-sell run on data too granular to hand over.
Cut churn on the one dataset a rival can't buy.
Telco growth is a retention problem now: every point of churn is subscribers a rival wins, and the models that predict, prevent, and personalise around it run on the subscriber graph, the most granular behavioural data in any economy. Hand that graph to a third-country model and you are training someone else on the one dataset a rival cannot buy.
Run churn, care, and cross-sell on sovereign, first-party subscriber data and you lift retention and ARPU on the workloads where a few points move the whole P&L, and, at subscriber-base volume, sovereign inference undercuts per-token frontier pricing anyway. The moat and the margin point the same way.
We build the sovereign customer stack: churn and retention, agentic care, and personalised cross-sell decisioned inside the perimeter, with frontier speed at the non-identifying creative edge. Protect the subscriber relationship and you compound it. Leak it and you fund a competitor's.
Where each AI workload
should live.
| Workload | Placement | Rationale |
|---|---|---|
| Churn prediction & retention offers | Sovereign | Trained on subscriber behaviour; must not leave the perimeter. |
| Personalised cross-sell & ARPU growth | Sovereign | Runs on the subscriber graph, the telco's moat. |
| Agentic customer care | Hybrid | Frontier reasoning; subscriber memory and PII kept sovereign. |
| Marketing creative | Frontier | Non-personal, fast iteration. |
What is yours to encode,
and what you are renting by accident.
Before any platform decision, one question settles the rest: how much of what your organisation knows can be written down precisely enough for a machine to act on it, and who ends up owning that writing. We run a short engagement to answer it, and it produces working artefacts rather than a report.
Two artefacts come out of this. An operating ontology, the nouns: the entities, relationships, and states your business actually runs on, described once and precisely. And agentic skills, the verbs: the procedures, thresholds, and judgments your best people apply, written down, versioned, and testable instead of retold.
A placement map addressed to the CTO or CIO office: what runs on your compute, what is bought, what is rented, who holds the keys, and what each costs. One test cuts through most of the debate. If a supplier vanished on a Friday, what stops working on Monday, and how long would it take you to replace it?
Where an open-weight model fine-tuned or adapted on your own corpus beats a frontier call, where a small purpose-built model for one narrow task beats both, and where the frontier still earns its fee. Geographic sovereignty falls out of that answer rather than having to be argued for on its own.
- →Tariff and eligibility logic, including every legacy plan nobody dares touch
- →Retention offer ladders, and the point at which a save is not worth making
- →Care scripts, fault triage, and the escalation thresholds behind them
- →The network diagnostics your field engineers perform by instinct
- →Complaint and vulnerability handling, in the form the regulator expects to see
Network telemetry and care transcripts are a private corpus at enormous scale. A small model trained on them triages faster and cheaper than any general model, on compute you already own.
- +A register of encodable know-how, ranked by value and by how fast it decays
- +A draft operating ontology in an open format, not inside a vendor's platform
- +Two or three working skills, built and evaluated on your real work, not demonstrated on a slide
- +A bring-it-in-house plan the CTO or CIO office can actually staff
- +A model strategy: open weights, fine-tuning, purpose-built small models, and where the frontier stays
- +An evaluation harness that shows it works, and keeps showing it after we leave
This is a CTO and CIO office engagement, not a procurement exercise. Your own people have to end up running it, because the moment the encoding is delegated, the thing being encoded quietly stops being yours.
Palantir calls this layer an ontology, and the idea is right: a governed model of your objects, links, and actions that agents can act on. It lives inside their platform. 8090 will design, build, host, and maintain the software around it for you. Both are serious, and both leave the same question open. At the end of it, who owns the layer that holds your judgment? We build the same artefacts in open formats, on compute you control, and hand your CTO the keys.
The subscriber stack,
inside the network.
The same four layers we ship in legal, instantiated for an operator sitting on the most continuous behavioural record any company holds.
A care copilot grounded in your own tariffs, coverage, and fault history, churn and next-best-offer decisioning on network and CRM signal together, and a consent and preference hub the whole estate reads from.
Tariff and eligibility rules, retention offer ladders, complaint and vulnerability handling, and regulatory scripts written once as versioned skills, so the answer a customer gets in the app matches the one they get on the phone.
Location, usage, and network telemetry are pinned to sovereign compute by architecture. Frontier capability is reachable for non-personal work only, and every routing decision is logged where a regulator can follow it.
In the operator's own data centres, at the edge where latency demands it, or on a managed isolated instance for the teams that would rather not run it themselves.
The application layer here is built for you, not bought. The pattern is the one we already ship, as a product, in legal.
Four postures.
One codebase.
The same software in every posture, so a team can start hosted and end up air-gapped without a migration project. You move along the ladder when you decide to, not when a contract renews.
Your stack stood up on our infrastructure in days and pointed at real work, so it is judged on your matters rather than on a demo. The pilot fee credits against the hardware when you migrate.
Teams that want proof before capital expenditure.
The open-source cores installed on a machine you own, by a script anyone can read first. No account with us, no telemetry, no phone-home. Code flows one way, from us to you, and only when you ask for it.
Small teams, and anyone who wants to inspect the source before trusting it.
An isolated instance per client on sovereign infrastructure, European by default, run, patched, and monitored by us, with the security and backup credentials in your hands rather than ours.
Organisations without an infrastructure team to spare.
Dedicated AI hardware inside your building, with the applications and the open-weight models you choose installed and tested before it ships. It runs with the network unplugged, and it keeps working whatever happens to us.
For an operator the appliance is how a subsidiary or a market gets its own capability quickly without waiting on a central platform programme.
Work that cannot leave the building, ever.
- +Deployment and hardening, on your machines or ours
- +Knowledge encoding: your playbooks and positions compiled into installable skills
- +Model selection, tuning, and an evaluation harness built on your own work
- +Remote maintenance and monitoring at a published monthly rate
- +Updates as ordered, readable migrations, always after an encrypted backup
- +Backup, restore, and continuity drills you can actually rehearse
- +Enablement for the people who use it daily, not only for IT
- +Governance evidence: routing logs, audit trail, and an attestation pack
- Reach into your instance or see your work
- Revoke your software or disable a licence remotely
- Force an update on you
Not promises. Consequences of the architecture: your keys and your data are generated on your machine and stay there, and nothing calls home.
Get the full
telcos
placement map.
We send a sector-specific brief with workload-by-workload placement guidance and a reference architecture for sovereign deployment.