The Cognitive Marathon: Applying Systems Thinking to Tame the SEND Admin Beast

Before I stood at the front of a classroom in Stoke-on-Trent, I spent my days knee-deep in peat bogs. As a natural scientist, my world was governed by the carbon cycle—analysing the complex exchange of greenhouse gases between the soil, plants, and the atmosphere. I measured flux, calculated variables, and sought to understand the "homeostasis" of natural systems. I spent years radiocarbon dating samples, looking for patterns in the chaos of nature.

When I transitioned into teaching Special Educational Needs and Disabilities (SEND), I realised something profound: a Further Education (FE) college is just another ecosystem. But instead of carbon, we are trading in cognitive energy.

And right now, that system is leaking energy at an alarming rate.

I am a 41-year-old teacher who happens to be dyslexic and dyspraxic. For me, and for many of my students, standard literacy is what I call a "Cognitive Marathon." It is the exhausting process of decoding symbols just to get to the meaning. Recently, while gardening—my hands in the soil, listening to an audiobook—I had a "lightbulb moment." By removing the act of visual decoding, I could absorb complex information effortlessly. I wasn't fighting the text; I was engaging with the ideas.

If we apply this logic to our professional lives, specifically the crushing weight of SEND administration, we find a solution that doesn't just save time—it saves our sanity.

The Entropy of Administration

In thermodynamics, entropy is the measure of disorder in a system. In the UK Further Education sector, "admin" is our entropy. We all know the reality: Education, Health and Care Plans (EHCPs), RARPA (Recognising and Recording Progress and Achievement), funding bands, and endless observation records.

According to the Department for Education’s (DfE) School Workforce Census, teachers consistently report working nearly 50 hours a week, with a significant portion of that time dedicated to non-teaching tasks like data management and general administration. In the SEND sector, where every interaction requires robust evidence to justify funding and track subtle progress, this workload is amplified.

This isn't just annoying; it is inefficient. Every hour I spend agonising over the phrasing of a written report is an hour I am not planning a lesson, supporting a student, or simply recovering my own cognitive resources. For a neurodiverse teacher, this "admin tax" is even higher.

The Solution: Contextualised Speech-to-Structured Text

To fix this, I stopped thinking like a tired teacher and started thinking like a pragmatic scientist. I needed a workflow that bypassed the "literacy barrier" and moved directly to data capture. I needed to hack the system.

I call this process Contextualised Speech-to-Structured Text AI.

Here is the methodology I have developed to track progress, particularly in dynamic environments like community visits, Supported Internships, or hospital placements where sitting down with a laptop is physically impossible.

1. The Input: Wireless Data Capture

I wear a wireless lapel microphone (or lavalier) clipped to my lanyard. It connects to my laptop or tablet, allowing me to be completely hands-free. As I float around the room or the work placement, I record verbal feedback.

> The Scientific Principle: In field research, you don't wait until you're back in the lab to remember the data. You record it in situ. If you wait, you lose fidelity.

> 

I walk up to a student—let's call him Student X—and I speak the feedback directly to him, knowing the microphone is listening. I might say:

"Well done Student X, you stayed on task for 15 minutes today without support. You used the equipment safely, which hits your EHCP outcome for independence."

This achieves two critical functions:

 * Immediate Feedback: As per the Education Endowment Foundation (EEF), feedback is most effective when it is immediate and specific. The student hears it instantly and can react.

 * Data Capture: The raw audio is transcribed into a rough block of text on my device.

2. The Processor: The AI "Filter"

At the end of the day, I am left with a "raw dump" of text. It is messy. It is full of "umms," "ahhs," and my own dyspraxic stumbles. If I were to paste this directly into a RARPA record or an observation form, it would look unprofessional.

This is where the AI comes in. I feed this raw text into a Large Language Model (LLM) using a customised, rigid prompt.

The prompt acts like a filter in a chemistry lab. It is designed with strict constraints to prevent hallucinations (the AI making things up). The instructions I have built are clear:

 * Role: Act as a professional SEND editor.

 * Task: Restructure the raw speech into formal observation records.

 * Constraint: Only use facts present in the text. If a student is not mentioned, do not generate feedback for them. Map evidence specifically against EHCP outcomes or SMART targets.

3. The Output: Structured Evidence

The AI takes my rambling monologue and converts it into a pristine table:

 * Activity: What was done (the context).

 * Progress: What was achieved (the evidence).

 * Next Steps: What comes next (the pedagogy).

It transforms a 5-minute verbal brain dump into hours' worth of typed paperwork, perfectly formatted for our college systems.

Overcoming the "Dyslexia Fatigue"

This system is professional, but for me, it is also deeply personal.

My dyslexia manifests most aggressively when I am tired. By 4:00 PM, after a full day of teaching and managing complex behaviours, my brain is foggy. I start mixing up "is" and "are." I miss words entirely without noticing. Attempting to write quality observations at that time is a recipe for errors and frustration.

By using this speech-to-text workflow, I change my role. I am no longer the author struggling to type; I am the editor of my own thoughts.

This is a critical distinction. It allows me to maintain the high standards required by Ofsted and the college while preserving my energy. It reduces work-related stress, which we know is a leading cause of teacher attrition in the UK. I can simply speak my truth, and let the machine handle the syntax.

The Human in the Loop: A Scientific Warning

However, let us channel the caution of Bill McGuire regarding climate systems: technology is a tool, not a saviour. We must keep the human in the loop.

AI is not perfect. It can misinterpret pronunciation—especially with regional accents! It can struggle if two students share a first name.

 * Quality Control: You cannot automate empathy. If you mumble "good job" into the microphone, the AI will produce a generic "good job" report. To get quality out, you must put quality in. You must detail why the work was good.

 * Verification: You are the professional. You must check the output. The AI is the admin assistant; you are the manager. It is an editor, not the author.

Ripple Effects: Empowering the Learners

Perhaps the most "magical" result of this experiment hasn't been my own workload reduction, but seeing the impact on my students.

I work with learners who have significant literacy barriers. Asking them to write an article for the college newsletter is often a demoralising task. It highlights their deficits and erodes their self-esteem. If I hand them a blank piece of paper, I am often handing them anxiety.

I recently applied this same workflow to them. I allowed them to speak their news stories. We captured their raw thoughts—their excitement, their unique voice—and used the AI to structure it into a newsletter article.

The change was electric. They weren't limited by their ability to spell or hold a pen. They were judged on their ideas. For a student who has spent their life feeling "behind," seeing their words polished and published is a game-changer. It builds confidence in a way a spelling test never could.

Note: We must be ethical. We cannot use this for accredited assessments where writing is the assessed skill. That would be cheating. But for expression, reflection, and inclusion? It is vital.

Conclusion: The Endurance to Adapt

Ernest Shackleton once said, “Difficulties are just things to overcome, after all.”

In the landscape of Special Education, the difficulties are often bureaucratic. We are drowning in paperwork that takes us away from our primary purpose: the students.

By adopting a scientific mindset—viewing the classroom as a system and admin as data—we can use tools like contextualised speech AI to hack the system. We can reduce our workload, improve the quality of our feedback, and most importantly, conserve our energy for the human interactions that actually change lives.

If you are a teacher, a job coach, or a support worker, I implore you: stop typing and start speaking. Your cognitive energy is too valuable to waste on the admin beast.

"This text was conceived and directed by a human, using Voice-to-Text and AI assistance to overcome a dyslexia induced literacy barrier."