Why does a business need email in the age of messaging apps and AI?
Author
Redu
Date published
Because email is the only channel that works with everyone, belongs to your business and leaves a record. Messaging apps come and go. An email address stays on the business card, in the contract and on the invoice.
Email is not dying. It is growing
Every few years someone announces the end of email. Meanwhile more than 4.4 billion people use it, and by 2028 there should be more than 4.9 billion. Those are the figures of the research firm The Radicati Group in its Email Statistics Report 2024–2028.
The reason is simple. Email has no owner. You do not have to ask a customer which app they use. You write to an address and the message arrives: at a corporation, at a council office and at a one-person business.
That is why everything that matters in a business ends up in the mailbox anyway:
- Quotes, orders and contracts. Even if the conversation started on the phone.
- Invoices. You send yours and receive other people's.
- Accounts with other services. The bank, the wholesaler, the invoicing software: each of them sends password reminders to an email address. Whoever holds the mailbox holds the keys to the rest.
- Evidence. A messaging app is like a chat in the corridor. Email is like a letter with a date-received stamp. In a dispute with a supplier you search the mailbox, not the chat.
Email is not an inbox. It is contracts, invoices and customers.
Since everything goes by email, fraudsters take the same road
According to the UK government's Cyber Security Breaches Survey 2025, 43 per cent of businesses identified a cyber security breach or attack in the previous 12 months – around 612,000 businesses. Among those that did, phishing was by far the most common type: 85 per cent of them experienced it.
Phishing is a message that pretends to come from someone you trust: a bank, a courier, a supplier or your boss. What it wants is a password, data or a bank transfer.
A typical scenario goes like this. The accounts clerk gets an email from the "managing director". Urgent transfer, I'm in a meeting, don't call. The sender's address differs from the real one by a single letter. The English is flawless, because fraudsters have artificial intelligence writing for them too these days.
Ordinary filters check the technical side. Whether the sender's server has a good reputation. Whether there is a virus in the attachment. Whether the domain is protected by SPF, DKIM and DMARC – three mechanisms confirming that the sender is who they claim to be. UK government guidance tells public-sector organisations to protect their email domains in exactly this way.
That stops mass spam. But a carefully prepared, one-off email from a new domain is technically correct. The fraud is in the content. And an ordinary filter does not understand content.
With us, artificial intelligence reads the email too
That is why we have added another layer to the ordinary filters. Every incoming message is read by a language model – an AI "engine" that understands what the text says. It pays attention to the same things an alert employee would:
- time pressure and a request for discretion,
- a request for a transfer, a password or a change of bank account number,
- impersonation of the boss, a bank or a supplier,
- a sender address that does not match the signature.
The verdict is one of four: safe, suspicious, phishing, or phishing aimed at a specific person. A suspicious message gets a star. With phishing, a red warning also appears at the top of the message.
We did three things on purpose.
- We delete nothing. The message stays in the inbox. A person decides – only now a forewarned one.
- The content does not go outside. The model runs on our own graphics card in a UK or EU data centre. We do not send your emails to any external AI service.
- Email comes first. When the model is down, messages arrive as normal, just without the extra assessment.
The protection has been running in production on the first servers since September 2026. We extend it to further servers after a trial period. First we check that the model does not raise false alarms.
To be honest: AI does not replace common sense. The model will sometimes warn about an ordinary email and sometimes let something through. The rule "I confirm any change of bank details by phone" still applies.
The mailbox is also a desk for an AI agent
An AI agent is a program that carries out a task from start to finish and reaches for the data and tools it needs by itself. Since everything that matters comes into the business by email, the mailbox is a natural place for it to work.
An agent with access to a chosen mailbox can:
- sort the email and prepare draft replies to recurring questions,
- take invoices out of attachments and enter amounts, dates and numbers into a spreadsheet or your software,
- remind you about threads nobody has replied to,
- take instructions by email: you forward it a message with the note "prepare a quote" and you get a draft.
There is one rule: the agent proposes, a person approves. You are the one who sends the quote and the reply to a complaint. We build agents like this to order, for a specific mailbox and a specific task. And when your email and your agent are with one provider, your data does not travel between companies.
What this means for you
- Treat email as infrastructure. Your own domain, someone in charge and backups. It is not a free add-on to the internet.
- Check your domain. The National Cyber Security Centre offers a free tool, Check your email security. Within a minute it shows whether you have SPF, DKIM and DMARC in place.
- Ask your provider who reads your email, and where. An AI filter only makes sense when the content does not leave servers you trust.
RAG lets AI answer from your company documents and cite its sources. We show how to build it so that your data never leaves the data centre.
An AI agent is not software out of a box but an employee who needs a proper induction. You fit the technology to the people, not the people to the technology.