Where Most SMEs Really Are On Their AI Journey

Most SMEs are already "doing AI", even if they wouldn't describe it that way. In conversations across Kinabase's customer base, a familiar pattern shows up: teams are genuinely excited, they've experimented with tools like Claude and ChatGPT, and they've seen glimpses of what's possible. The gap isn't enthusiasm. It's the translation.

AI adoption for SMEs is the process of moving from ad hoc tool experimentation to repeatable, governed use cases that are designed around how the business actually operates.

The biggest blocker tends to be imagination, and not in the creative sense, but in the practical sense of converting a powerful general-purpose tool into a real-world operational advantage. SMEs often know AI can write, summarise, brainstorm, and analyse. What they're less sure about is how to apply it to the messy, cross-functional reality of running a business: handovers, approvals, inconsistent data, edge cases, and unwritten protocols.

That's where the AI journey often stalls with plenty of experimentation, but not enough systems thinking and process redesign.

The Common Early-Stage Trap: Optimising a Broken Process

A theme we notice at Kinabase is that many SMEs start by trying to improve one step in their existing workflow, often the noisiest or the most time-consuming step, rather than rethinking the workflow end-to-end. The intention is sensible: "Let's automate the admin", "Let's speed up proposals", or "Let's add a chatbot." But if the underlying process is inefficient, AI can end up accelerating the wrong thing.

This is why process mapping matters so much. The most effective AI implementations usually begin with a simple but rigorous exercise: map the journey from first customer contact through to the delivery and invoicing, including every decision point and dependency. Then ask the key question: "If the business were designed today, with AI available, how would this process work?"

This idea aligns with broader thinking in process redesign: AI isn't only a layer of automation, but a catalyst for re-engineering end-to-end workflows, shifting focus from incremental efficiency to operational improvement.

Why AI Feels Overhyped but is Still Valuable

AI is undeniably a buzzword. That creates two opposing risks at once:

• Overconfidence: assuming AI can replace complex judgement, domain expertise, or nuanced customer conversations.
• Underuse: treating AI as a novelty for marketing copy and meeting summaries, while leaving deeper operational value untouched.

The truth is more practical than either extreme. Generative AI is powerful, but it comes with constraints that matter in business settings. Large language models (LLMs) can produce fluent, convincing outputs, yet they may generate incorrect information because they are optimised to produce plausible text, not to guarantee truth.

This isn't a reason to avoid AI, but it's a reason to place it carefully. In Kinabase customer environments, the best results tend to come when teams choose AI use cases where "almost right, quickly" is valuable, and where humans can review, validate, and make final decisions.

Chatbots Are Not the Main Event

A lot of SMEs begin with chatbots, like ChatGPT, Claude, or Gemini, because they're visible, easy to picture, and feel like an obvious "AI move". Sometimes they help, especially for triaging routine requests. Even so, chatbots are rarely the highest-leverage starting point, and they can disappoint when businesses expect them to behave like expert staff members.

For many SMEs, bigger wins come from less flashy applications:

• Drafting first versions of proposals, reports, or client updates
• Turning meeting notes into action plans and task lists
• Summarising long email threads into decisions and next steps
• Extracting structured data from unstructured inputs (with oversight)
• Building internal knowledge workflows that help teams find answers faster

These uses share a theme: they remove repetitive, time-consuming work while keeping people responsible for decisions, quality, and client outcomes.

The Real Misunderstanding: AI Isn't Set and Forget

Another misconception is that adopting AI means switching on a tool and letting it run. In reality, value comes from designing the system around the tool.

In practice, that includes:

• Giving AI the right information: context, constraints, templates, brand voice and tone, and examples
• Defining clear tasks: what success looks like, what to avoid, how to handle edge cases
• Building review steps: where humans verify facts, approve outputs, and correct errors
• Capturing learning: reusing prompts, improving templates, and documenting what works and what doesn't.

This approach mirrors what responsible AI guidance emphasises: organisations need to identify and manage risks, implement appropriate controls, and treat generative AI as a capability that requires governance, not just software to install.

Augmentation Beats Replacement, Especially for SMEs

The strongest AI outcomes Kinabase sees are rarely about removing people. They're about increasing capacity.

AI is particularly good at handling work that teams already know how to do, but struggle to make time for: polishing, summarising, structuring, drafting, and turning raw inputs into usable outputs. When those tasks shrink from hours to minutes, SMEs gain something far more strategic than cost savings; they gain time and mental capacity to focus on client relationships, service quality, innovation and growth.

This is consistent with how major policy and research organisations frame AI's impact on work. AI is changing tasks and roles, creating opportunities alongside risks, and organisations benefit most when people and AI are designed to work together rather than treated as substitutes.

What 'Good' Looks Like: Starting with the Sales-to-Invoice Flow

Across SMEs, one of the most effective ways to start an AI programme is to walk the entire "sales to invoice" path:

• Lead and enquiry handling: What information is captured, where does it go, and what gets lost?
• Qualification and scoping: Where do judgement calls happen, and what data supports them?
• Proposal and pricing: Where do the errors most commonly occur? What takes the most time to generate?
• Delivery and handover: Where do delays happen? How do teams chase updates?
• Invoicing and follow-up: What triggers billing? What causes disputes or late payment?

Mapping this flow makes it much easier to spot where AI genuinely fits, and where the business needs process clarity first. Often, the "AI opportunity" isn't an isolated tool; it's a redesign of how information moves through the organisation.

A Practical View of the AI Journey for SMEs

From Kinabase's vantage point, most SMEs sit in one of three stages:

• Experimentation: people use publicly available tools ad hoc; results vary, and there's little consistency or reusability.
• Operationalisation: teams define a smaller number of repeatable use cases; templates emerge, and quality controls appear.
• Transformation: end-to-end processes are redesigned with AI in mind, data is treated as a core asset, and governance is embedded.

The biggest leap is from experimentation to operationalisation, because it requires changing behaviour, not just buying software.

For SMEs, the most reliable path forward is simple and structured: start with a process map, pick high-frequency use cases, build human review into the workflow, and create reusable assets (prompts, templates, checklists) so benefits compound over time.

If you're working through where AI fits into your operations, Kinabase can help you map the workflows, structure the data, and build the review steps that make AI genuinely useful. See how it works in practice.

AI doesn't reward vague ambition. It rewards clear processes, good inputs, and thoughtful implementation.