When Not to Automate: Balancing Efficiency and Authenticity in the AI world

This article was inspired by a conversation with a client. In that conversation, we made a strong case against automating. In today's world, where you can theoretically automate almost everything, the better question is when I should not automate.

There is a huge opportunity, especially for smaller, more nimble businesses, to adopt a technology-driven business environment, and automation is one of the cornerstones of operational efficiency. This is particularly true as it becomes easier to place AI-generated content within automated workflows, we must now address new questions on authenticity.

However, determining when to automate—and perhaps more importantly, when not—requires careful consideration of both technical possibilities (which are getting easier and easier to solve) and human connection. This article explores what we feel is the nuanced decision-making process behind an effective automation strategy, helping a business strike the right balance between efficiency and authenticity, in particular when it comes to AI-generated content.

Before diving into the specifics, let's acknowledge a fundamental truth: while automation can dramatically improve productivity and reduce costs, it should never come at the expense of meaningful human connections in high-value interactions. This is where the integration of AI-generated content should be inverted and rebalanced within an automation.

The Fundamental Approach to Automation

Let's reexamine some of the fundamentals of automation before we address the 'new' kid on the block, generative AI, which opens the door to a whole new field of content automation.

Identifying Low-Value Work for Automation

The primary goal of automation should be to eliminate tasks that add little value yet consume significant human time and resources. As highlighted in research by Appian, this includes processes characterised by repetitive tasks, manual steps and excessive handoffs between teams, which create inconsistent outcomes[1]. Before implementing any automation solution, ask yourself:

• Is this task repetitive and rule-based?
• Does performing this task manually add any unique human value?
• Would automating this process free up time for more strategic work?

Modern automation tools excel at handling structured, predictable workflows. According to IFTTT research, automation is particularly valuable when it saves time, reduces errors, boosts productivity, and increases scalability. These benefits make automation ideal for back-office operations, data entry, scheduling, and basic information requests.

We recommend adding another rule here, especially if you pride yourself on being a nimble organisation that reacts quickly to market shifts.

• How often would you want to change the automation?

It will help determine the 'how' of implementing your automation.

The Business Case for Strategic Automation

Successful automation isn't about replacing humans but optimising where human attention is directed. Forbes Tech Council notes that determining whether to automate should consider factors including:

• Cost-benefit analysis of implementation
• Business readiness and technical capabilities
• Comfort with delegating control to automated systems
• A clear understanding of what automation can realistically achieve

Let's explore that last point a bit more. As artificial intelligence has rolled into the mainstream, we have seen all sorts of 'wrapper' platforms that say things like, "Put in a few sentences, and we will give you a production-ready blog post." While they do, they trap you into a volume and velocity game when it comes to content automation—which will fail.

Integrating generative AI into content workflows makes a lot of sense, but it should be tuned to create value for the author or content creator, not replace them.

When Automation Falls Short: The Human Element

On Digitising Events, we have been conducting a year-long experiment on where and how to apply artificial intelligence in the content creation process. Great content, which ranked and got high engagement, was one where human oversight was put at the core of the content production line. Don't get me wrong; we automated a lot, but the subject-matter expert plays a key part in adding emotional intelligence and genuine connection.

In the context of B2B, relationships with audiences and customers are built over time, which is very different from consumer-based marketing, which is where a lot of AI-driven marketing is focused. In this type of marketing, the relationship and trust are built over a shorter time period and are much more transactional.

The Risk of Inauthentic Communication

Where is the danger of misguided automation more apparent than in customer and prospect communications? Take, for example, the seemingly endless receipts of supposedly "personalised" outreach messages that immediately feel generic and inauthentic:

"LinkedIn messages that we are all recipients of, which are trying to be personal... but you can see through it a mile off, and it comes across as disingenuous and inauthentic."

This sentiment is echoed in market research, which finds that 86% of customers prefer speaking with human representatives rather than automated systems[4]. When communications lack genuine personalisation, they risk damaging rather than building relationships.

A Value-Based Automation Framework

Industry research supports our view that the level of human oversight should directly correlate with the value of the interaction. ModSquad's research confirms this approach, warning that losing the human touch is a primary pitfall of customer service automation[5].

"the higher the value of the potential interaction that you're trying to initiate or reignite... the more care should be taken about it."

So, breaking this down, we have:

• Low-value, transactional communications (payment reminders, registration confirmations, badge pick up and so on) are ideal for full automation
• Medium-value interactions may benefit from partial automation with human review
• High-value or high CVO (Commercially Valuable Outcomes) communications (significant sales opportunities, relationship management with key clients) warrant substantial human involvement, and automation should be engineered to enable and support them, not replace them.

Intelligent Automation: A Better Approach

The closer the automation is to your customer or attendee. Where you know you will benefit significantly from an emotional connection, the more cautiously and carefully you should approach the addition of automation.

Automation with Intent and CVO aligned

Rather than viewing automation as an all-or-nothing proposition, modern approaches emphasise "automation with intent"—using technology to enhance rather than replace human connection.

CloudFactory's research on AI implementation emphasises that "human oversight is critical for managing the unpredictable elements of real-world data" and that the ideal approach combines "the speed of automation with the nuanced judgment of human insight".

Leveraging AI for Enhanced Personalisation

We need to invert our thinking on leveraging AI-generated content to support the personal touch and deepen human interactions. Not using AI tools to replace personal outreach, but to strengthen it through deeper research and understanding:

"The clever use of [AI] is to help you craft that authentic outreach by leveraging their ability to do a much more detailed and amazing amount of research, which you would never have time for..."

This belief aligns with HeroHunt.ai's findings on automated personalisation, which suggests that effective automated outreach should include genuinely personalised elements like references to the recipient's specific experience, skills, and achievements.

The Next Level - Human-in-the-Loop Automation Design

A consensus is rapidly building around the critical role of human oversight in automated systems, and human involvement is a key part of this.

1.   The understanding of nuances and context

2.   Addressing the ethics of automated systems

3.   Identifying areas for improvement and experimentation

4.  Managing change and driving adoption

5.   Balancing efficiency with empathy and connection

This "human-in-the-loop" (HITL) approach ensures that our automation enhances rather than diminishes the quality of interactions. It also removes significant friction from the implementation of automation as the teams realise they are a baked-in and integral part of this human-in-the-loop automation design.

For clarity, our view on HITL is that it must not be rubberstamping LLM output but a much more integrated, thoughtful and modernised approach to process design.

Practical Applications and Where to Draw the Line

The rush to automate and roll out your GPT service agent needs a handbrake. If poorly done, it will damage a relationship. For example, a recent fraudulent transaction on my Revolut account, which I documented here, means I will never use that business again.

Or, in the case of delivering service during high-stress times, like the buildup of a show, your customers may not want to be engaged in a frustrating chat with a joyfully optimistic and overly friendly robot.

Customer Service and Support

While chatbots and automated systems can efficiently handle straightforward queries, research shows they falter with complex issues that require nuanced understanding.

Best practice involves creating clear paths for customers to access human support when needed. As ModSquad recommends: "If you use customer service bots, keep the sequence short. Give your customers a clear way to indicate they need to speak to a real human. Don't leave them stuck without a lifeline".

In the context of Digitising Events, we must not ignore the context of time. No organiser will get any brownie points for putting an exhibitor into an AI chat loop when they have an urgent, high-stress situation onsite.

Sales and Marketing Outreach

In sales outreach, the tension between scale and authenticity is particularly acute. While tools like ZELIQ can automate sequences of emails, connection requests, and messages, As Kaspr's research indicates, effective personalised outreach requires truly understanding the prospect's needs and pain points, not just inserting their name into a template[10]. This might involve:

• Prospects into meaningful segments
• Focusing on solving specific problems
• Building a genuine social media presence before outreach
• Guiding prospects through a thoughtful buyer journey

Existing Customer Communication

"These are people you ought to know a lot about... it does hit the wrong note if not done well."

LinkedIn research on authenticity suggests that for existing relationships, communication should reflect a shared history and demonstrate a genuine understanding of the customer's situation[11]. Automated systems should draw on customer data intelligently to reference specific interactions, challenges, and successes in the relationship.

A Framework for Automation Decisions

So, when should you automate, and most importantly, how should you automate in a world where we are edging closer and closer to the reality of being able to automate anything?

While blue-collar workers have been the recipients of technological redundancy through robotics and other advances, the GPT revolution is a technology that has the white-collar and knowledge worker squarely in its sights. So, how do we embrace it and retain human value?

Four Key Questions to Ask Before Automating

Here's a starting framework for evaluating when to automate:

Value assessment: What is the potential CVO value of this interaction or process?

Complexity evaluation: How much contextual understanding and judgment is required?

Relationship impact: Will the automation strengthen or potentially damage essential relationships?

Human augmentation: Could technology enhance rather than replace human capabilities?

The Automation Spectrum

The world is not black and white - a binary attitude which is harmful. So, rather than an "automate or don't automate" decision, consider a spectrum of approaches:

Full automation: For high-volume, low-complexity, low-value interactions

Guided automation: Systems that suggest responses but allow human review and modification

Augmented human work: AI tools that enhance human capabilities through research, suggestions, and analytics

Fully manual: Reserved for the highest-value, most nuanced interactions

The risks of AI-Generated Content and its role in automation

Let's face it: Generative AI has created a tsunami of drivel on social media and the internet. It also erodes trust in content, as the assumption that, for example, a long article was written with a lot of care and attention is now no longer true.

However, if approached correctly, small businesses can quickly gain a competitive edge over larger rivals. To achieve this, the approach of using Powered tools and AI-driven content must be anchored in the goal of maintaining an authentic connection, especially when it comes to customer interactions.

We have been experimenting extensively with the role of AI content, particularly when applied to social media posts. What has worked best is to use AI to create initial drafts but not publish unsupervised—instead, placing them in a queue to easily delete or finely edit in order to tune the posts for that authentic connection.

AI-driven digital marketing is already here, and many tools claim that you can hand off close to 80% of a marketing team's jobs. However, we urge caution, as nearly all the tools we have evaluated are tuned to work for a high-volume use case like e-commerce or other more consumer-oriented businesses.

To achieve a similar outcome in B2B, you will need to invest in building a memory for your AI-driven digital marketing, which has particular domain knowledge, allowing it to create meaningful connections with audiences and customers.

A balance must be struck between automation and a genuine commitment to high-quality content that creates engagement and meaningful connections. As B2B relationships are built over time, the higher the transactive value, the more care must be taken.

Remember to Automate with Intention and HITL

Automation works best when implemented with clear intention and awareness of its limitations. LinkedIn's analysis states, "Maintaining authenticity in every interaction is not just a nice-to-have—it's a necessity".

The most effective automation strategies recognise that technology should serve relationships, not vice versa. Organisations can achieve efficiency and authenticity by focusing automation on low-value tasks while preserving or enhancing human connection in high-value interactions.

As we navigate an increasingly AI-driven business landscape, the winners will not be those who automate most aggressively but those who automate most thoughtfully, with a clear understanding of when the human touch remains irreplaceable.