From Subscriptions to Strategy: How AI Integration Reshaped a UK Robotics Engineering SME

Background

When this robotics engineering company engaged Beyond Touch through a government-funded AI support programme, it was already a survivor. The business had been hit hard by COVID, closed, and then re-emerged as a Phoenix company. Revenue had returned—not at a scale that offered comfort, but enough to keep the business moving forward.

The company manufactures and supplies commercial robotic cleaning systems for schools, hotels, and large corporate offices. Its client base includes well-known organisations across professional services, energy, defence, and logistics. At the centre of its offer is a fleet management system that stores, charges, and transports multiple robotic vacuum cleaners from a single trolley.

At the same time, the director was developing new products to move the business beyond distribution and into genuine UK-led innovation. These included a battery-powered robotic window cleaner and, more ambitiously, a fully proprietary external window cleaning robot designed and assembled in the UK. The business had already invested in AI tools, with subscriptions to both ChatGPT and Microsoft Copilot, but neither was delivering meaningful value. That gap between paying for AI and using it effectively was exactly what the engagement was designed to address.

Challenge

Paying for AI Without Real Operational Value

The company was already subscribed to both ChatGPT and Microsoft Copilot, but neither tool was being used in a way that justified the spend. The director had experimented with ChatGPT for circuit board design, only to end up with lengthy conversations and no exportable design file. Route planning had shown some promise, reaching around 85% accuracy, but that work had taken place six months earlier and had never been built into the business. Copilot remained largely untouched because the company had not upgraded to the full licence. Two AI tools, two ongoing costs, and very little measurable return.

Engineering IP Dependent on Overseas Supply

The long-term ambition was clear: move from reselling Chinese-manufactured robots to becoming a true UK design-and-manufacturing business. That required ownership of the PCB designs, firmware, and front-end software. It also required a manufacturing partner capable of producing the circuit boards. Four UK PCB manufacturers had been approached, and all four quoted lead times of at least twelve months. The only realistic option appeared to be an overseas supplier already experienced in similar production. For a business whose positioning relied heavily on UK engineering credibility and data security, that created a serious strategic tension.

Schools Market Without a Strong Commercial Narrative

Schools were already an important customer segment, but the sales approach lacked a compelling story. The company had recently scraped government data to build a contact database of 86,000 schools and was actively running outreach through an email platform. What it did not yet have was a message that connected with the priorities of headteachers, governors, and business managers. The outreach focused on the functional offer—robots that clean floors—rather than the wider institutional outcomes schools actually care about.

Cash Constraints Holding Back R&D Progress

A change to the company’s invoice discounting arrangement had created immediate financial pressure. The original lender had been acquired, and the new owner would no longer fund advance rental invoices. As a result, the director was preparing to inject personal capital as a director’s loan to cover both day-to-day cash flow and the company’s R&D programme. In that context, any tool, process, or system that could cut costs, improve efficiency, or accelerate revenue generation became commercially significant.

Solution

Demonstrating What AI Research Actually Looks Like

Before the first session even began, Grae at Beyond Touch used Perplexity AI to conduct a deep-dive on the company. Thirty-four sources were analysed, bringing together the business model, product range, client relationships, and wider public-domain intelligence. The director was visibly impressed when one of those sources turned out to be a report from the UK’s largest cleaning contractor, a client relationship he valued highly. That moment set the tone for the engagement: this is what AI research looks like when it is used properly.

The Perplexity research space was then shared directly with the director, giving him his own thread to continue the investigation independently. This was not a one-off demonstration. It was capability transfer in practice.

Consolidating the AI Toolkit

The approach to the company’s existing AI subscriptions was practical and commercially focused. The director was paying for ChatGPT while also holding a basic Microsoft 365 licence with only limited Copilot functionality. Grae demonstrated the full Copilot licence alongside ChatGPT, showing how Copilot could access diary entries, prepare meeting briefings, search internal documents, and generate talking points, all within the company’s existing Microsoft security environment.

The commercial logic was simple. The Copilot upgrade cost the same as the ChatGPT subscription it would replace. One subscription could be cancelled, one could be upgraded, and the business would gain stronger integration with its existing workflows at no additional cost.

Claude by Anthropic was also recommended for coding-related tasks, where it offered stronger performance than general-purpose tools. For a business working on firmware, API integration, and front-end software, that recommendation alone had the potential to save considerable development time.

Identifying AI-Powered PCB Design Platforms

The director’s frustration with UK PCB manufacturing delays was genuine. Four domestic manufacturers had been approached, and all four had quoted lead times of at least twelve months. During the Perplexity session, Beyond Touch searched for AI-powered PCB design systems and surfaced a range of viable options including DeepPCB, Flux AI, Quilter, Cadence Allegro X AI, Zuken, and Siemens EDA.

The director confirmed he had already subscribed to KiCad following an earlier search but had not been aware of these AI-enhanced alternatives. Each platform offered a different route forward, from cloud-based autonomous design to advanced layout validation. The research was shared as a live collaborative thread that the director could continue exploring in his own time.

Building a Value Proposition for Schools

This was the point in the engagement where live AI research materially changed the commercial conversation. Starting from the director’s observation that better-managed schools were more likely to invest in robotic cleaning systems, Grae ran a live search into whether Ofsted ratings take cleanliness into account. The answer was yes.

Under the newer inspection framework, cleanliness now connects to broader assessments around safeguarding, personal development, and pupil well-being. A second live search explored published academic research on cleanliness, student behaviour, and school performance. The findings supported the director’s instinct. Cleaner school environments were linked with greater pupil pride, stronger belonging, reduced vandalism, and improved behaviour.

The director immediately asked for the research links so they could be passed to the marketing colleague. In practical terms, the business now had a much stronger narrative for the schools market: not just cleaner floors, but a service linked to school culture, pupil well-being, and inspection readiness.

Fathom AI for Meeting Intelligence

Grae introduced Fathom at the start of the first session by showing the back end of a previous meeting. The director could see the recording, transcript, action items, and searchable AI summary all in one place. For a business that regularly meets suppliers, engineers, and cleaning contractors, the value was immediate.

Instead of relying on fragmented handwritten notes, the company could now create a searchable record of every important conversation. Adoption was immediate.

Timeline

Session 1 (November): Beyond Touch AI & Digital Transformation session. Fathom AI was demonstrated and adopted for meeting transcription. Perplexity AI was introduced through a live deep-dive on the company. A Microsoft Copilot upgrade was recommended, ChatGPT was marked for cancellation, Claude by Anthropic was recommended for coding and firmware tasks, AI-powered PCB design platforms were identified, and SEO analysis with content strategy was generated for the company website.

Session 2 (February): Follow-up session. The business confirmed a new part-time mechanical engineer and marketing hire. An overseas electronics partner had been identified for PCB and firmware development. A manufacturing location feasibility study was conducted live using Perplexity, while supply chain strategy, R&D planning, and the window cleaning robot development roadmap were reviewed in detail.

Session 3 (February): Final session. Live research was carried out on the Ofsted inspection framework and the relationship between school cleanliness and pupil outcomes. A new schools-market value proposition was developed during the session, AI-powered lead enrichment using Gemini was reviewed, and continued engagement beyond the funded programme was discussed.

Outcome

Immediate Capability Shift

Within the first session, the director moved from paying for AI tools he barely used to working with a more structured and purposeful toolkit across research, meeting management, and engineering design. Perplexity replaced ad-hoc Google searching. Fathom replaced notebooks. Copilot replaced ChatGPT at the same cost but with direct access to the company’s documents and calendar. The engagement did not simply introduce new tools. It showed how those tools could connect directly to the company’s real operational work.

Engineering and R&D Impact

Six AI-powered PCB design platforms were identified that the director had not previously known existed. For a business whose hardware development timeline was being measured in years, partly because of UK supply constraints, access to tools such as DeepPCB and Quilter opened up new ways to accelerate the design phase. The recommendation of Claude for coding tasks also gave the company’s part-time software resource a stronger option for firmware and API development.

Sales and Marketing Breakthrough

The Ofsted research conducted during the final session gave the business something it had not previously had: an evidence-based narrative for the schools market. Instead of positioning its robots simply as cleaning equipment, the company could now speak to wider outcomes including pupil well-being, institutional pride, and inspection readiness.

The director’s own instinct—that stronger-performing schools were more likely to become customers—was reinforced by the academic literature surfaced during the session. Research links, summaries, and follow-up search threads were passed to the marketing team the same day.

Financial Efficiency

The ChatGPT subscription was cancelled and the Copilot licence was upgraded at equivalent cost. In return, the director gained stronger functionality including document search, calendar integration, and meeting preparation without increasing monthly spending. For a business closely managing cash flow while funding R&D through director’s loans, that efficiency mattered.

Client Reflection

Frequently Asked Questions

Q: How can AI integration help a small engineering company that already has AI subscriptions?

A: Many SMEs are already paying for AI tools they barely use. These sessions begin by auditing what is already in place, then showing how to extract practical value from those subscriptions. In this case, one redundant subscription was removed and another was upgraded at the same cost, giving the business better security, stronger document access, and improved meeting preparation without additional spend.

Q: What technical knowledge do I need for AI integration with a business mentor?

A: Very little beyond basic computer literacy. The tools in this engagement were demonstrated live, applied to real business challenges, and shared in a way the director could continue independently afterwards. The process is designed to build confidence, not assume prior expertise.

Q: Can AI mentoring help with product development and engineering?

A: Yes. In this case, Beyond Touch identified six AI-powered PCB design platforms the client had not previously encountered, alongside a coding-focused AI recommendation for firmware development. The value came not from abstract advice, but from live research tied directly to the company’s engineering ambitions.

Q: How quickly can AI integration show results for a manufacturing SME?

A: Often within the first session. In this engagement, Fathom AI was adopted immediately for meeting transcription, a redundant subscription was identified for cancellation the same week, and commercially useful research was forwarded to the marketing team within hours of the final session.

Q: Is AI integration relevant for companies outside of tech?

A: Absolutely. This business manufactures physical products with metal chassis, electronics, and battery systems. AI was not replacing engineering expertise. It was improving research, accelerating market intelligence, and making existing digital tools genuinely useful. The same principles apply across manufacturing, services, nonprofits, and early-stage companies.

Q: What does an AI advisory session actually look like?

A: Practical, conversational, and tailored to the business in front of you. Grae typically demonstrates tools using the client’s own company as the subject, so the value becomes visible straight away. Sessions involve live research, shared outputs, and direct capability transfer rather than abstract presentations.

Q: How is data privacy handled during AI advisory sessions?

A: Privacy is treated as a core consideration. Meeting recordings are handled through Fathom AI with consent. Microsoft Copilot was recommended in part because it operates within the company’s existing Microsoft security environment. For a business already concerned about data routing through overseas systems, that mattered.

Q: Can this AI integration approach work alongside government-funded business support?

A: Yes. This engagement itself was delivered through a government-funded AI one-to-one support programme. Similar work can sit effectively within Growth Hub support, local authority contracts, and funded business mentoring schemes, making expert guidance more accessible for SMEs.

Ready to Make Your AI Subscriptions Actually Work?

If your business is paying for AI tools but not getting meaningful value from them, a single session with Beyond Touch can change that. Whether you need help choosing the right platforms, consolidating subscriptions, or building AI-powered research into your sales and engineering workflows, the focus is always on practical implementation using what you already have.

Products & Services Reference

Products Used in This Case Study:

• Power Hour - Intensive 1:1 mentoring session providing immediate, practical AI implementation guidance

Services Demonstrated:

• AI & Digital Transformation - Helping organisations discover and implement AI tools that solve real operational problems without requiring technical expertise
• Business Mentoring - One-to-one guidance that builds capability through hands-on practice with your actual business materials
• Business Training - Practical sessions that transform participants from nervous beginners to confident independent users in under an hour
• Business Process Improvement - Identifying and removing bottlenecks that constrain service delivery, particularly for budget-conscious organisations