Learning to Walk with AI - Automating Workflows
Walking is all about automating business workflows with a sprinkling of AI to help move things along. There are things that we do within our professional lives that are repeatable and predictable. Recording contact details from business cards, recording monthly expenses, following up on missed phone calls, issuing invoices and onboarding clients or employees. They often don’t add a huge amount of value, but are necessary and time consuming manual tasks. By automating these processes, we can free up people’s time, enabling them to concentrate on more valuable tasks. Over the past 12 months we’ve seen the rise of the most hyped about topic in commercial life …. agents, which I think of as being workflow automation with some AI embedded into it. Now some people on social media, usually Marketers bless their little cotton socks, will have you believe that you automate so much of your professional life with Agents that you can watch Netflix all day while letting the agents do all the heavy lifting for you. In my opinion, this is complete nonsense, because if there was a scam to make that work I would have found it by now. So in this article we’ll explore some practical examples of how workflows with AI (aka Agents) can help benefit the business and separate reality from the social media fiction.
What are AI Agents?
I can remember it was only last summer whilst on holiday that I got a message from my marketing director asking me if I had heard anything about Agentic AI. My initial reply was “Nope”, swiftly followed by, “I will check the ICLR 2024 submissions to see how many papers are related to agentic ai. But I suspect that it’s probably just marketing bs”. ICLR is one of the big four AI conferences in the calendar. It’s where academics and industry go to submit their work in the hope that their research will get published at the conference. It’s a huge deal and the competition to get your paper published at a conference like this is intense, which I discovered first hand last year when I put forward my own research. Think of it as like the Oscars for nerds. You get the picture, so you would think that the submissions for 2024 would be dominated by Agentic AI from a frenzy of academics desperate to show off their Agents and yet how many of the thousands of papers published that year made any sort of reference ot Agentic AI……. none. Zero, nada, not a single nerdy sausage has said anything about Agentic AI.
The reason for that is quite simple. Agentic systems are primarily software engineering tasks and have very little to do with AI. I guess that someone one day was playing around with a business process automation tool and suddenly thought to themselves “oooh, what if I pass the text of that step in my process into a Generative AI LLM and use it’s answer as the input to the next step?” And hey presto, we have a valuable application of AI in business, the marketers start frothing at the mouth and social media goes crazy.
Which is all fine, but what do we mean by Agentic AI today? Now, I’m a big fan of YouTuber Yannic Kilcher, and not so long ago, he made a series of posts on LinkedIn in which he gave his interpretation of Agentic AI. It’s something that I happen to agree with, so let’s go over it. Agentic systems have the following three qualities:
They are capable of performing some useful task for us.
They have access to tools that enable them to do things like perform an internet search to get the most up-to-date results, or to browse a specific website that we give them and more.
They are given more autonomy to make decisions on our behalf than we might do with other forms of AI.
The good thing is that Agentic AI systems can perform more complex and useful tasks than the old AI Assistants, which only provide text to read. But the downside is that with great power comes great responsibility, and so the potential havoc that they can cause when they screw up also increases.
The bottom line is that whether you want to think of these systems as being old fashioned business process automation with some steps being performed by AI or by AI being augmented with tools to perform processes the net result are some very powerful business systems that can save us time and money and for small businesses allow us, in corporate language to “punch above our weight”.
So What?
Because these systems are essentially software there are several ways that we can create something that we can consider agentic. Including:
Purpose-built tools like clay.com for marketing research or retell for voice calls.
Big tech AI vendors are enhancing the capabilities of their AI Assistants.
Now in this kind of setup we’d use AI to perform one or more of the steps in the workflow to make decisions on our behalf. AI is very good at things like understanding intent, translating text from one language to another and even reading text from images using something called OCR. So if these AI tasks can be integrated into our workflow it can become very powerful.
Example 1. Personalised Landing Pages with Clay and Webflow
Imagine working in marketing where you want to create landing pages tailored to each prospect segment — for example, one for SaaS founders in fintech and another for marketing leaders in e-commerce. Traditionally, this would involve research, copywriting, and manual page creation for each target group — a time-consuming process that’s difficult to scale.
With an agentic AI workflow, you could automate much of this. A tool like Clay can collect and enrich data about potential customers — such as company size, industry, tech stack, or recent funding news — by pulling information from LinkedIn, Crunchbase, or other online sources. Once this data is gathered, the AI agent can use it to generate personalised marketing copy, headlines, and imagery suited to each audience segment.
Next, the workflow connects to Webflow through its API. The AI automatically generates new landing pages, populates them with the personalised content, and even adjusts design elements like colour schemes or testimonials to match the tone or brand of the target industry.
The outcome is a fully automated system that can research, design, and publish customised landing pages at scale — something that would normally require coordination between marketing researchers, copywriters, and designers. This makes campaigns far more dynamic, responsive, and cost-effective, allowing marketers to test ideas faster and target new audiences almost instantly.
Example 2. Handling Expense Receipts with n8n
Now imagine you’re in operations, and your team is constantly processing expense receipts — scanning them, verifying details, and entering them into accounting software. It’s repetitive work that eats up time and often leads to delays or data entry errors.
An agentic AI workflow built in n8n (a low-code automation platform) could handle this end-to-end. The system watches a shared inbox or a cloud folder for new receipts. When one arrives, the AI reads the content of the image or PDF using OCR, extracts key information like the merchant name, date, amount, and VAT, and classifies the type of expense.
It can then automatically cross-check this against company policy, flag anything unusual, and push the verified data directly into your accounting system (like Xero or QuickBooks). Finally, it can notify the submitter that their claim has been processed — or request clarification if the document was incomplete.
The result is a workflow that runs continuously, never tires, and ensures accuracy and compliance — reducing admin time while maintaining a clear audit trail.
Example 3. Triaging Messages from a Contact Form
Consider a company website with a contact form. When someone submits a message — perhaps asking about job opportunities, pricing, or technical support — it usually lands in a shared inbox. A team member has to open the email, read it, decide what it’s about, and then forward it to the right person.
An agentic AI workflow can take over this process entirely. It monitors the inbox for new messages, analyses the text to understand intent (for example, whether it’s a technical issue, sales enquiry, or HR request), and routes it automatically to the right destination — such as a helpdesk system like Zendesk, a CRM for sales leads, or a specific department’s inbox.
The system can also send a polite acknowledgement to the sender, confirming receipt and letting them know when to expect a response.
The net result is that messages are handled instantly and consistently, reducing manual effort and improving response times. It’s more reliable, more scalable, and unlike humans, the system doesn’t take breaks, get sick, or go on holiday.
Example 4. Handling Incoming Calls with Voice AI
Now imagine a busy restaurant during peak hours. Staff are serving customers, managing bookings, and handling payments — all while the phone keeps ringing with new reservation requests. Missed calls mean missed business.
A Voice AI agent can step in to manage this seamlessly. When a customer calls, the AI answers the phone, speaks naturally, and understands the caller’s intent — for instance, to book a table, modify a reservation, or ask about opening hours. It can then check availability, confirm the booking in the restaurant’s reservation system, and send a confirmation via SMS or email.
This frees up staff to focus on customers in the restaurant while ensuring no opportunity is lost. The experience feels professional and responsive — and the business gains the capacity to handle every call, even during its busiest times.
Conclusion
As we’ve seen, workflow automation combined with a touch of AI can transform how businesses operate - not by replacing people, but by amplifying their capacity to focus on what truly matters. Whether it’s building personalised landing pages, processing receipts, or triaging messages, these systems remove friction from everyday tasks and create space for deeper thinking and better service.
And the same principle applies to communication itself. Voice AI is now extending automation beyond the screen - allowing businesses to answer every call, engage every customer, and deliver a consistent experience 24/7. It’s not about building robots to replace human connection; it’s about using AI to make sure that connection never gets missed in the first place.
Call to Action
At Neural Aspect, we’re helping businesses in taking their first steps into this new era of automation - from smart workflows to conversational Voice AI that communicates in your brand’s language. If you’re interested in exploring how Voice AI or workflow automation could streamline your operations, reduce administrative burden, or enhance customer engagement, contact us. Let’s see how we can help your business walk - and perhaps even run - with AI.
