8 reference pages answering the questions we are actually asked before a project starts: what an AI agent is, what automation costs in 2026, how to calculate the return, whether software with no interface can be automated, and how long a build takes. Each page answers its question in the opening paragraph and carries the figures behind it, including third party market ranges that are not ours.
Last reviewed . Written and maintained by Oxford Crown Technologies.
The complete index. The grouped sections below give the opening line of each answer.
An AI agent is a system that is given a goal, works out the steps itself, and then carries them out inside your real software, reporting what it did. A chatbot only produces text: it answers a question and stops. Workflow automation sits between the two, following a fixed sequence a person designed in advance, with no judgement of its own.
Read the full answerLast reviewed
An AI receptionist is a voice system that answers your inbound calls, holds a natural conversation, answers questions from your own documents, books appointments into your calendar and routes or escalates anything it should not handle itself. It works around the clock and takes every call at once. It handles routine, factual and transactional calls well, and it should never be the last stop for a complaint, a vulnerable caller or anything with legal or financial consequence.
Read the full answerLast reviewed
In 2026, a single automation project from an AI automation agency typically costs between USD 1,500 and USD 12,000 or more. On a retainer, small businesses generally pay USD 1,000 to USD 3,500 a month for two or three workflows, mid-market organisations USD 4,000 to USD 10,000 a month, and enterprises USD 8,000 to USD 25,000 a month. Running costs sit on top of all of those and are usually quoted separately.
Read the full answerLast reviewed
Take the hours the process consumes each week, multiply by the fully loaded hourly cost of the people doing it, and multiply by 52 to get the annual cost. Do the same for the hours the process will still consume after automation. The difference, minus annual running costs, is the annual net benefit. Divide the build cost by one twelfth of that figure and you have the payback period in months.
Read the full answerLast reviewed
Hire in-house if automation is a permanent, continuous function of your business, because a salaried engineer who understands your operation will out-perform any external supplier over a two to three year horizon. Use an agency if you have a defined set of processes to automate, want the first one running in weeks rather than months, or cannot yet justify a full-time salary. The deciding question is whether the work continues after the first few builds are finished.
Read the full answerLast reviewed
A single agent automating one clearly defined process takes two to three weeks from the start of discovery to running in the live path. A suite of three or four connected agents takes four to eight weeks. A department-wide rebuild takes three to six months and is delivered in stages rather than as one release. Roughly a third of that elapsed time is normally spent waiting for access, decisions and sign-off rather than on the build itself.
Read the full answerLast reviewed
Yes. Four routes exist, and they should be tried in order of durability: an undocumented or paid vendor API, direct read access to the underlying database, a scheduled file exchange such as a nightly export, and, last, driving the interface itself the way a member of staff would. Each is legitimate. Each carries a different risk, and the last one carries the most.
Read the full answerLast reviewed
Yes. Open-weight models can be run on hardware your organisation owns, inside your own network, so that no document, recording or record is ever transmitted to a third party. It costs more up front than calling a hosted model, and the very best models are not available this way, but for extraction, classification, summarising and routine drafting the gap is small enough that most regulated workloads run perfectly well.
Read the full answerLast reviewed
Four rules govern this section, and they are published so that a reader can hold us to them.
Disclosure, and the only sales pitch on this page
Oxford Crown Technologies builds voice and operations agents for organisations headquartered in London and delivered across the Gulf, including the systems that have no way to connect to them. If a question you need answered is not here, the discovery call is the fastest place to get it answered properly.