Why new technology exposes old gaps instead of fixing them

Chris Witham • September 6, 2026

A pattern keeps turning up across different jobs, different clients and different bits of technology. Whether it's an AI agent, a website or just a bit of admin, the new thing rarely fixes a gap in how a business runs. It just makes that gap happen faster, or makes it impossible to ignore.


The agent copies the habit, not the intention

Set an AI agent up on top of a muddled process and it doesn't tidy the process up. It copies it, at speed. I've seen an agent handling quote requests inherit a habit that had been quietly costing time for years, quoting before checking whether the job was even in the right area. The agent did it perfectly, every time, much faster than a person would.



The fix wasn't a cleverer agent. It was going back and sorting the actual process first, then letting the agent follow the corrected version. The agent part is often the easier half of the job.


The website answers everything except the real question

The same thing shows up on websites. Plenty of small business sites are full of nice photography and warm words about the team, but a visitor scrolling for the price, or what happens after they enquire, finds nothing. So they email to ask, which is exactly the enquiry the website was meant to save someone from handling manually.


It happens because once you know your own business inside out, the obvious questions stop feeling obvious. The sites that generate proper enquiries usually start from a short list of three or four things a stranger genuinely needs to know before they'll get in touch, and build everything else around that.


Knowing when to say "I don't know"

Building more with AI agents lately, the harder part isn't getting a sensible answer out of one. That's the easy 80%. The harder 20% is getting it to recognise when it doesn't have enough information to answer at all. A confident wrong answer causes more trouble than an honest "I don't have enough detail, here's what I'd need". Testing the gaps, the vague questions, the missing information, tells you far more than watching the easy cases succeed.


Where the week actually goes

Underneath all of this sits the same admin pattern most small business owners recognise. It's rarely one big job eating the week. It's the same three questions answered by email a dozen times, the quote that takes twenty minutes because the information isn't in one place, the booking double-checked by phone because nobody quite trusts the form. None of it looks like a single problem. It just quietly takes the hours.


None of this is really a technology story. It's a clarity story. Before automating anything, it's worth writing down what only you currently know, sorting the process that's actually being followed, and being honest about the questions a stranger would ask on landing on your site. The technology part tends to be the smaller job once that's done.

By Chris Witham August 31, 2026
Why the unglamorous groundwork behind websites, AI agents and admin systems matters more than the finished result most people see.
By Chris Witham August 25, 2026
Why AI agents and automation only help once the underlying workflow, form or service page is properly understood. Practical notes for small business.
By Chris Witham August 13, 2026
Small business workflows and websites drift out of shape through small changes. Here is why periodic review matters more than adding more tools.
LucidSynergy Content Engine logo with a green circular icon and gold text on white background
By Chris Witham April 30, 2026
Content Engine: a new writing tool built for in-house marketing teams
Abstract blue glowing digital circuit network on a dark background.
By Chris Witham February 20, 2026
Not every week ends with a certificate. This one did.
Blurred colorful computer code on a screen, with lines of red, blue, green, and white text.
By Chris Witham December 11, 2025
Where AI really helps your Business If you spend any time on LinkedIn or X, you’ll have seen bold claims about how AI can help you build software in a matter of days. There’s a lot of excitement, a lot of big promises, and a fair bit of confusion for business owners trying to work out what’s real. A new term doing the rounds is “Vibe Coding” —the idea of describing what you want to an AI assistant and having it generate the code for you. It’s becoming popular because it can move things forward quickly and help people explore ideas they wouldn’t have been able to create alone. And the truth is, it does have its place. The challenge isn’t the technique. It’s the expectation that AI will automatically deliver finished, reliable, production-ready tools without any real design or thinking behind them. AI accelerates the work you already do well Used properly, AI can: • Remove huge amounts of repetitive work • Speed up drafting and iteration • Generate working prototypes in hours • Help non-technical people explore ideas • Improve documentation, planning and communication This is where it shines. But it still needs clarity, structure, and well-designed processes around it. It’s like having a very fast assistant rather than a fully formed development team. Why many AI projects don’t deliver what people expect Independent research this year showed a clear pattern: • Many early AI initiatives failed to produce measurable business value • Companies abandoned AI ideas because they couldn’t scale or integrate them • The gap between an impressive demo and a reliable tool is larger than people thought This doesn’t mean AI is overhyped. It means teams jumped straight to execution without the groundwork. The technology isn’t the issue. It’s the approach. Small businesses don’t need Enterprise Platforms Most UK small businesses don’t need to build a full software product. What they actually need is: • Better workflows • Faster content generation • Clearer communication • Improved customer support • Tools that reflect the way they work • Consistency and repeatability AI is perfect for this. A custom GPT trained on your tone, your documents and your processes can become: • A writing assistant • A customer support helper • A knowledge base navigator • An internal guide for staff • A quality-control layer • A process automator No engineering team needed. No complex infrastructure. No stress. Where AI builds real value right now AI works best when it’s part of a thoughtful, guided approach: • Define the outcome you want • Build a lightweight prototype (AI helps here) • Add structure, rules and guardrails • Connect it to your real workflow • Test it with real users or staff • Iterate until it feels natural You can still move fast. You just avoid building something brittle that breaks the moment it’s needed. The key insight: AI doesn’t replace expertise, it amplifies it AI is at its strongest when someone knowledgeable decides: • What it should do • What it shouldn’t do • How it should behave • What tone it should use • How it fits into the business • What checks and constraints matter That’s where tools like custom GPTs genuinely shine. They’re not software products in the traditional sense. They’re flexible assistants shaped around your business. With the right design, they can save huge amounts of time and deliver consistent, practical value without any of the complexity of building a full system. A more useful way to think about AI in 2026 Instead of “AI will build everything for you”, a healthier mindset is: AI speeds up the work, but you set the direction. For small businesses, that’s more than enough to make a real difference.
Banner reading “INTERVIEWED BY” above a white “WEBSITE PLANET” logo on a dark background
By Chris Witham August 21, 2025
Website Planet Interview
January 16, 2025
THOUGHT OF THE DAY: 'An investment in knowledge pays the best interest.' - Benjamin Franklin EXPLANATION: This quote reminds us of the enduring value of learning and education as a pathway to success.
January 15, 2025
WORD: Egregious DEFINITION: Outstandingly bad or shocking. USAGE: 'The error in the report was so egregious it couldn’t be ignored.' ORIGIN: From the Latin 'egregius,' originally meaning 'distinguished' or 'remarkable,' later used ironically to highlight glaring faults.
January 14, 2025
TRADE FOCUS: Chef AI APPLICATIONS: - MENU DESIGN: AI creates innovative recipes based on available ingredients. - SUPPLY MANAGEMENT: AI predicts ingredient needs to minimise waste. - CUSTOMER INSIGHTS: AI analyses feedback to improve dish offerings. - SKILL TRAINING: AI provides virtual cooking lessons and techniques.
January 13, 2025
On 13th January 1898, Émile Zola published 'J'accuse,' an open letter defending Alfred Dreyfus, which became a pivotal moment in the Dreyfus Affair.
January 12, 2025
IDIOM OF THE DAY: Bite off more than you can chew FIGURATIVE MEANING: To take on a task that is too difficult to manage. TYPICAL USAGE: 'She bit off more than she could chew by agreeing to organise the entire event.' ORIGIN: Likely from the literal difficulty of chewing an overly large piece of food.
January 11, 2025
IDIOM OF THE DAY: Throw in the towel FIGURATIVE MEANING: To give up or admit defeat. TYPICAL USAGE: 'After hours of trying, he threw in the towel and called for help.' ORIGIN: From boxing, where a towel is thrown into the ring to signal surrender.
January 10, 2025
AI GENERATED HUMOUR: Why did the AI bring a notepad to the beach? To take down sandy data! This joke connects AI’s data-driven nature to a playful seaside scenario.
January 9, 2025
THOUGHT OF THE DAY: 'The only way to do great work is to love what you do.' - Steve Jobs EXPLANATION: This quote highlights the importance of passion in achieving excellence in one’s work.
January 8, 2025
WORD: Ineffable DEFINITION: Too great or extreme to be expressed in words. USAGE: 'The beauty of the sunrise was ineffable, leaving everyone speechless.' ORIGIN: From the Latin 'ineffabilis,' meaning 'unutterable,' often used to describe profound experiences.
More Posts