While businesses rush to embrace artificial intelligence (if you listen to the mass-media AI is the silver bullet after all!) for operational efficiency, the reality is far more nuanced.

In our experience, the truth is that AI without proper process automation is like building a skyscraper on quicksand—impressive technology built on unstable foundations that will inevitably crumble (or be seen and never visited). Understanding this key relationship is, from my perspective, crucial for successful harnessing of the power available in the current generation of AI tools.

The Uncomfortable Truth About AI Implementation

Recent research (2022 through to late 2024) shows that up to 85% of AI projects undertaken by SME’s fail. Additionally, some 42% of companies have abandoned their AI initiatives faster than they could launch them.

The primary culprit isn't the technology itself—it's the lack of foundational processes that effectively support the implementation of the AI models (ie: they are a solution to a problem that has not been clearly identified, defined and tested).

Jumping into AI without structured workflows, will give predictable consequences: "garbage in, garbage out" data quality issues (or data in so many different places it is virtually unusable), automation without clear direction, and resistance from teams dealing with chaotic systems. As one expert noted, "AI depends on high-quality data. If your data is fragmented or inconsistent, AI-driven insights will be unreliable".

Why Process Automation Must Come First

Process automation serves as the essential foundation that makes AI implementation successful. Before AI can optimise decisions, predict outcomes, or enhance customer experiences, your business processes need to be standardised, documented, deployed, tested and streamlined. In tech speak, this is known as Robotic Process Automation (RPA) and it is THIS (RPA) that can actually transform business operations by handling 20% to 30% of organisational tasks, creating the stable environment that AI needs to thrive.

Consider this: businesses with clear AI strategies see up to 30% increases in productivity, while those without strategies WILL OFTEN ABANDON AI altogether. The difference lies in having automated processes that provide clean, consistent data with established workflows that AI can enhance rather than replace.

Dare I say it, the fundamentals of IT have not changed in 30+ years – process first, computerisation second – now we just replace process with RPA and computerisation with AI. Ie: there is NOTHING NEW HERE!

The Automation Ecosystem: Beyond AI Alone

Business optimisation in the age of RPA requires a comprehensive ecosystem of automation tools. Platforms like n8n, Zapier, and Make serve as the connection “tunnels” that link the various systems together. These tools handle the fundamental task of data integration and workflow coordination—this is the unglamorous but essential “stuff” that enables AI to function effectively and deliver true business value rather than “look at me” apps in an app-store.

Even without AI, process automation delivers immediate benefits! Enhanced productivity, cost reduction, improved accuracy, and better compliance can all be achieved through short, sharp focussed projects that never even need to consider AI security, privacy or the difference between models!

This then allows AI to provide the intelligence layer that can optimises these automated processes over time. Together, they then create a powerful combination that far exceeds what either can achieve alone.

Our 5-Step Framework for Automation-First Decision Making

To help businesses determine the right approach for their automation journey, here's our practical evaluation framework:

Step 1: Process Assessment and Documentation

Evaluate the current state of your business processes. Document workflows, identify bottlenecks, and map data flows. Ask: Is this process standardised? Are the steps clearly defined? Can the process be replicated consistently irrespective of who is operating it? How widely is the process used? What happens if the process stops/breaks/is interrupted?

Step 2: Automation Suitability Analysis

Assess each process against key criteria: volume (how frequently performed), complexity (rule-based vs. judgment-required), data structure (structured vs. unstructured), and stability (how often the process changes). High-volume, rule-based, stable processes are the ideal candidates for initial automation.

Step 3: Technology Matching

Match the right automation technology to each process. Simple, repetitive tasks suit RPA tools like Zapier or n8n. More complex processes requiring data analysis or pattern recognition may need AI integration in conjunction with an RPA tool. Consider the technical complexity and available expertise in your organization.

Step 4: AI Enhancement Evaluation

Determine if/where AI can add value to your automated processes. AI excels at optimising existing workflows, making predictions based on historical data, and handling unstructured inputs. However, it requires the stable foundation that process automation provides and needs exception handling and error condition checks.

Step 5: Implementation Prioritisation

Prioritise implementations based on impact, feasibility and risk profile. Start with high-impact, low-complexity and low-risk automations to build momentum and demonstrate value. This approach allows you to establish the operational discipline needed for successful RPA/AI implementation later.

Suggested Objectives - by Organisation Size

Small Organizations (1-50 employees): Focus on automating basic administrative tasks (invoicing, appointment scheduling, data entry) using tools like N8N, Make, Zapier or Microsoft Power Automate. Build data management discipline before considering AI solutions.

Medium Organizations (51-500 employees): Implement comprehensive process automation across departments. Integrate systems using platforms like n8n or Make. Begin pilot AI projects in areas where you have clean, automated data flows and data resources with known security, structure and risk profile.

Large Organizations (500+ employees): Develop enterprise-wide automation strategies combining RPA, workflow automation, and AI. Establish governance frameworks and change management processes to support organisation-wide transformation.

The Path Forward

To data, the Organisations displaying the most success with AI integrate it as part of a broader optimisation ecosystem rather than a standalone tool/solution. By prioritising process automation first, businesses can build the stable foundation that AI needs for success. This approach reduces risk, increases the likelihood of success, and delivers operational benefits in the short term.

Remember: automation first, AI second. Your future depends on getting this sequence right.