The questions people actually ask about AI.

Most quotes fall into two shapes. Agencies charge a one-off build fee, typically £3,000 to £15,000, then bill separately for hosting, changes and support. Or you pay a monthly platform fee for a self-serve tool you configure and maintain yourself, usually £30 to £300 a month, which is cheap until you need it to do something the template does not cover.

Ostfro works the second way with the first way’s quality: a flat monthly plan from £299, no upfront build fee. Building, hosting, monitoring, fixes and monthly improvements all sit inside it. If the chatbot needs changing in month seven because your pricing changed, that is not a new invoice.

The honest variable is scope. A bot answering twenty repeat questions from your existing documents is quick. One that checks live stock, takes payment and writes to your CRM is a bigger build, but it is the same monthly plan, and the scope is written down before anything starts.

There is a simple test, and it has nothing to do with company size or how technical you are. Is there a task that happens repeatedly, follows rules you could explain to a new starter in ten minutes, and involves information that already exists somewhere in writing? If yes to all three, it is automatable today.

Businesses that struggle are usually automating something with no settled process yet. If five people do the same job five different ways, automation freezes the confusion rather than fixing it. The fix is deciding the process first, which a scoping call usually does anyway.

You do not need clean data, a big budget, or a technical person on staff. You need one task that is genuinely eating hours. Start with that one, prove it, then widen.

The realistic list is narrower than the headlines and more useful than the sceptics suggest. AI is very good at reading documents and pulling out what matters, answering repeated questions from a known body of information, drafting text a human then approves, moving information between systems, spotting patterns in numbers, and handling phone and message traffic around the clock.

It is not good at things needing genuine accountability, physical presence, real relationships, or judgement where there is no precedent. Anyone selling you AI that runs the business is selling something that does not exist.

For most small businesses the wins are unglamorous and large: the phone answered at 3am, the quote drafted in four minutes instead of forty, the invoices posted without retyping, the follow-up that actually happens.

Goldman Sachs estimated in 2023 that around 300 million full-time jobs' worth of tasks worldwide are exposed to AI automation. That figure is about tasks, not people, and the distinction matters: most roles are a bundle of tasks, and AI takes some of them, not all.

What we see in practice with small UK businesses is redeployment rather than redundancy. The admin hours come back and get spent on work that actually grows the business, quoting more jobs, calling customers back, being on site. Nobody has hired us to make someone redundant; they hire us because they cannot afford to hire in the first place.

The uncomfortable half of the truth: a business using these tools does more with the same headcount than one that isn’t, and that gap compounds. The risk is not AI replacing you, it is a competitor with AI outpacing you.

A single chatbot or automation, scoped tightly, is typically live within two to four weeks. Something spanning several systems, a CRM with a dashboard and automated quoting, runs longer, and the scope document says so before you commit to anything.

The part that takes longest is rarely the technology. It is getting agreement on what the process actually is, and getting access to the systems the thing needs to talk to. Businesses with their rules written down move fastest.

You see working versions as they land rather than waiting through a silent gap for a reveal. If something is heading the wrong way, that is when you catch it.

It depends entirely on what is built and where it runs, which is why the question deserves a direct answer rather than reassurance. Your records stay yours, exportable in full at any time, in a format you can read. Nothing is designed to make leaving expensive.

The practical safeguards that matter: know which provider processes your data and where it sits; keep customer personal data out of consumer chat tools; use business tiers where inputs are not used for training; and keep a human in the loop for anything with legal or financial consequence.

For UK businesses this is a GDPR question as much as a technical one. If a build touches personal data, that gets addressed in the scope, not after launch.