Examples and evidence
Check the public company record, try a working demonstration, inspect sample documents and read the independent research behind the service logic. Every item explains what it can and cannot show.
One part of the business · your team approves important decisions · written instructions included
At a glance
What you can check
Register v1.0 · checked 27 July 2026. We label each item honestly and do not call a sample a customer result.
The Companies House register shows Alternate Machines Ltd as an active private limited company, company no. 17326261.
You can check the legal name, company number, status and filed details at Companies House.
It does not show the quality of our work, certify our skills or prove a customer result.
You can check the legal name, company number, status and filed details at Companies House.
It does not show the quality of our work, certify our skills or prove a customer result.
A fictional request moves through a responsible person, a missing-information check, a human decision and a recorded result.
A working example has been built, and you can see how its decisions and checks behave.
It is not connected to a customer system or AI service, and it does not show a customer result or software compatibility.
A working example has been built, and you can see how its decisions and checks behave.
It is not connected to a customer system or AI service, and it does not show a customer result or software compatibility.
The sample shows how we map a process, check the software you already own, recommend what to do next and label every estimate.
You can inspect the report structure and level of detail before buying a Workflow Decision.
It contains no client data and shows no saving, live implementation, software compatibility or delivery time.
You can inspect the report structure and level of detail before buying a Workflow Decision.
It contains no client data and shows no saving, live implementation, software compatibility or delivery time.
Sample documents show how we record responsibilities, approved access, testing, handover and removal of access.
The project controls are written down in a format you can inspect before buying.
It is not a certification, independent security audit, service-level agreement or record of work for a client.
The project controls are written down in a format you can inspect before buying.
It is not a certification, independent security audit, service-level agreement or record of work for a client.
Current client evidence
We do not imply a customer result, saving or return where none has been independently documented and approved for publication.
Work for a related company can appear only when source records and written permission are available. We will clearly explain the relationship and will not present that work as independent proof.
External research and official evidence
These are findings about other organisations—not Alternate Machines customer results, forecasts or guarantees. They explain why the services may be useful and why each project still needs its own tests.
Research register v1.0 · sources checked 6 August 2026
Among 1,890 UK SMEs already using digital technology, 40% said it saved time that could be spent on other tasks. Thirty-six per cent reported better work quality and 32% reported more effective business operations.
The possibility of practical time, quality and process benefits, and the reason to begin with one observable operational problem.
It does not establish a causal effect, an AI-specific result, money saved, an Alternate Machines result or what one proposed workflow will achieve.
Difficulty identifying suitable activities or business use cases was the most commonly reported AI-adoption barrier, cited by 39% of firms in 2023. Cost was cited by 21% and AI expertise or skills by 16%.
The reason to check the business task and software already in use before recommending a build.
It does not prove demand for Alternate Machines, willingness to pay £950 or that every business has a valuable AI use case.
Among 700 UK businesses using AI, 57% said they had developed new, improved or more flexible processes or operations, and 34% reported lower production or operating costs. However, 77% reported no revenue change and 10% reported no impact so far.
Process improvement as a credible objective and the need to measure an operational result instead of promising revenue or return on investment.
It does not establish causation, audited savings, Alternate Machines capability or a predicted result for a proposed project.
In 2025 to 2026, 21% of 1,870 AI-using UK businesses said their AI tools were connected to existing business systems.
It shows that many reported AI uses remain separate from existing business systems.
It does not mean the remaining 79% need their tools connected, that connecting them always creates value or that every supplier or system permits a connection.
A 2025 peer-reviewed study of 5,172 customer-support agents found that access to a generative-AI assistant increased successfully resolved customer issues per hour by 15% on average.
The potential of an assistant that helps staff find information and draft replies while people retain control.
It does not establish UK SME performance, autonomous chatbot effectiveness, an Alternate Machines result or compatibility with a buyer’s systems.
In a three-month cross-government trial involving 20,000 licence holders, 7,115 survey respondents reported an average saving of 26 minutes per day with Microsoft 365 Copilot. More than 70% agreed that it reduced time spent searching for information and on mundane tasks.
Potential uses within Microsoft 365 and the importance of adoption, information management and access controls.
It does not establish an objectively measured causal saving, an SME result, a monetary return or an Alternate Machines outcome. Seventeen per cent reported no clear time saving.
In an experiment with 758 BCG consultants, people using GPT-4 completed 12.2% more of 18 suitable knowledge-work tasks and finished them 25.1% faster on average. On one deliberately unsuitable task, AI users were 19 percentage points less likely to reach the correct answer.
The need to choose one clearly limited task, test it against written checks and keep people responsible for important decisions.
It does not establish live operational performance, results outside consulting or how a different model will perform on a buyer’s workflow.
How we publish a result
A testimonial tells you what someone thought. A measurable claim also needs records showing where the number came from and what may have affected it.
Name the process, team and period covered by the evidence.
Record the starting figure, how it is defined and where it came from.
Explain what was introduced and who was responsible for it.
Show the measured change, time period, calculation and important limits.
Keep written permission, a contact who can confirm the evidence and a clear note about any business relationship.
