At eight days, a project is usually still discovering how many meetings it needs. In March 2020, as Britain prepared to lock down, FourNet used those eight days to put nearly 2,000 npower contact-centre staff into their homes. Calls still arrived. Recording, workforce management and optimisation tools still worked. The emergency job was covered by the energy company's existing managed-service arrangement, so the extra project cost was zero. This is the neatest introduction to FourNet: a company whose work disappears when it succeeds. The phone rings, the citizen gets through, and nobody writes to an MP.
Founded in Manchester in 2005 as 4net Technologies, the business started with four communications specialists - Richard Pennington, Frank Jennings, Mike Jervis and Steve Tyrrell - and one customer. The original proposition was on-premise telephony delivered with more care than the big integrators could manage. Twenty-one years later, the equipment list has swollen to cloud contact centres, Microsoft Teams Voice, secure networks, cyber operations, customer-journey consulting and agentic AI. The care claim survives, but it now arrives with metrics and, in some engagements, a fee placed at risk.
A control room disguised as a phone queue
FourNet's customers occupy the institutional end of the address book: the Cabinet Office, HM Treasury, the Bank of England, 10 Downing Street, councils, universities, housing providers, financial firms and more than half of Britain's ambulance trusts. These are places where “communications” means rather more than a pleasant video call. It can mean a 999 operator hearing a caller, a vulnerable tenant reporting mould, or a university help desk remaining reachable while decades-old exchanges are dismantled.
The company treats the contact centre as an operational control point. Voice, chat, email and messaging enter; staffing, compliance, case systems and human judgment meet behind the screen. When it works badly, the first visible failure is rarely the cloud. It is the morning queue, the abandoned call, the agent typing the same summary twice or a specialist receiving a simple case while an exception bounces between teams.
“It’s changed how we plan every day.”Project lead, UK government agency
One unnamed government agency offered an unusually clean demonstration. Its leaders believed an end-to-end legal process took about 28 weeks but could not see where the time went. Work was distributed largely at random. Callers could spend more than 20 minutes waiting, talking and holding. FourNet's engineers cleaned millions of case-system events with Python and built a trustworthy view of work. The revelation was awkward and useful: only about eight weeks were active handling; much of the rest was waiting on citizens or third parties.
The team changed telephony planning and created a daily work list prioritised by skill, service level and risk. Customer engagement time fell from roughly 1,400 seconds to 795 - a 42 per cent reduction - while abandonment dropped about 60 per cent, even with fewer hours on the phones. What failed first was visibility. The fix was not a heroic replacement programme but a reliable dataset and a queue people could manage.
Fix the operation, then move the machinery
The method is less “big bang” than a controlled series of small noises. FourNet diagnoses demand, queues, wrap time, recontact and risk. It looks for early capacity - automated call summaries, better forecasting, less manual quality assurance - and tracks the gain in a benefits ledger. That released time or cost funds the next stage. FourNet calls this self-funding transformation. Its published ranges are 40 to 60 per cent less wrap time, 50 to 60 per cent less quality-assurance effort and a five to 10 per cent lift in first-contact resolution. Those are claimed typical improvements, not promises for every buyer.
At the University of Manchester, caution was the acceleration mechanism. The university's Siemens telephony estate had roots around 2000, supported tens of thousands of ports and was increasingly expensive to maintain. It also fitted hybrid work poorly. Rather than flip a campus-sized switch, the programme began with 1,000 representative users. Feedback exposed integration wrinkles while the old service remained available. The next phase moved almost 30,000 ports in managed batches, using 500 Gamma SIP trunks with active-standby resilience. Some 21,000 numbers ultimately moved to Teams Voice, and NICE CXone Mpower became the contact-centre layer.
The machines did not get the final vote. FourNet says most users liked the new platform, while people with advanced requirements continued to shape it after launch. That admission is more persuasive than a frictionless case study. The pilot changed minds because it made migration concrete: staff could test the future without betting the university on it.
Owning the seams
FourNet sits between a consultancy and a managed technology provider. It will recommend NICE, Microsoft, Cisco, Avaya, Content Guru or another supplier when that fits; it also owns several pieces of the stack. Agile Cloud provides UK-hosted communications for regulated organisations with a stated 99.99 per cent availability target. ANTENNA, designed with the Cabinet Office, is the five-nines option on Tier 4 infrastructure. UC3 bundles voice, video, messaging, meetings and service-centre functions. FourNet Connect supplies the managed network underneath them.
That last piece explains the acquisition strategy. After Palatine Private Equity bought a majority stake in 2021, FourNet acquired ComputerTel, infrastructure and security firm C>Ways, Cisco specialist Nowcomm and global network operator 01T. The acquisitions follow the anatomy of a failed call: application, security, local network, long-haul connection. Buying 01T reduced reliance on third-party network aggregators. Fewer seams mean fewer places for a supplier to point when the line goes quiet.
The model earns money from consultancy, implementation, licenses, subscriptions and recurring managed service. Public buyers can procure it through government frameworks. Commercial terms range from per-user cloud billing to tailored contracts. A healthcare customer moved 10,000 users and 850 agents onto a hybrid Avaya subscription that FourNet said would cost less over five years. A claims business received a Teams-integrated contact centre on a cost-neutral basis, partly by avoiding Microsoft calling plans. The headline number is therefore rarely the project fee. It is the infrastructure retired, agent time released or outage avoided.
Measure the queue before shopping for software. Pilot with difficult users, run old and new systems together where continuity matters, and make each saving pay for the next controlled change.
AI enters through the service door
IntellAIgent, launched in 2025, is FourNet's move from integrating AI to owning some of it. The platform creates governed software agents for voice and back-office tasks, with more than 300 pre-built integrations. Its sales pitch is end-to-end action: schedule an appointment, update a CRM and continue a conversation across channels, rather than merely produce a clever answer.
A live project for a global appliance manufacturer is the better description. Customers report faults, check warranties and move engineer visits through a conversational “front door” spanning voice, webchat and WhatsApp. The system uses Druid AI with Azure OpenAI's GPT-4o mini in an EU-hosted environment. A proof of concept went live in July 2025; production followed by November, with two-week hypercare periods at each stage. It now handles about 45,000 monthly interactions, routes 90 per cent of inbound contact through the AI layer and resolves 20 to 30 per cent - more than 13,000 issues - without an agent.
The useful design choice is the exit. When a machine reaches a complex or sensitive case, the human receives a transcript and structured summary. The customer does not begin again. At Alzheimer's Society, FourNet and NICE applied the same attention to vulnerability in a dementia-friendly contact centre that won a European industry award in 2024. Automation earns its place by removing repetition and recognising when empathy matters, not by trapping everyone in a more articulate menu.
The constraints are the point
FourNet competes with carriers such as BT and Vodafone, large integrators and focused UK specialists including Maintel, Kerv, Cinos, Wavenet and Sabio. It is smaller than the giants and cannot plausibly win every procurement on footprint or price. Its distinction is an accountable bundle: operational consultants, customer-experience designers, security-cleared engineers, a network operations centre and a security operations centre, plus owned infrastructure where sovereignty or assurance matters.
That bundle is most convincing under particular conditions. The service must be important enough that downtime, compliance and failed hand-offs have a measurable cost. Leaders need access to honest operational data and authority to change workflows. A staged migration needs time for pilots, parallel running and user feedback. Savings must be tracked rather than absorbed invisibly. If the buyer wants the cheapest commodity license, has no usable baseline, refuses to alter a broken process or needs a global hyperscale footprint, the economics become less obvious.
FourNet's own growth also carries integration risk. It expanded from 35 to more than 100 employees in the three years before its 2020 Tech Track ranking and reported more than 200 staff after four acquisitions. The culture it advertises - training, promotion from within, mental-health support and practical problem solving - now has to travel across inherited teams. Its carbon emissions rose after acquisitions before falling 12.14 per cent in 2025. Scale always sends an invoice.
“The modernisation has given us the agility we need to support our community in a changing world.”Patrick Hackett, Registrar and COO, University of Manchester
The company has come a long way from installing a good phone system for one customer. Yet the original idea remains legible. A FourNet project begins with somebody trying to get through: to an energy company in lockdown, a legal service with a morning queue, a university shedding an old PBX, or an appliance maker whose customer would simply like the washing machine fixed. Cloud, cyber and AI are supporting characters. The result that matters is a completed conversation - preferably on the first attempt.