Looking for Translation Aids? Start here
This page describes the process and evolutions of my AI Cognitive Scaffold and includes build your own scripts.
I do not use AI as a therapist, guru, replacement brain, or magic oracle.
I use it as a personal cognitive exosuit: a structured support system that helps me translate, organise, sequence, cross-reference, and communicate more effectively.
For me, this matters because I am autistic and ADHD. I often understand systems deeply, but translating that understanding into ordinary conversational format can create friction. My internal model may be detailed, accurate, and interconnected, while the person I am speaking to may only need one narrow slice of information.
That mismatch causes problems.
AI helps me bridge that gap.
It holds the map while I navigate the terrain.
The main function is not “advice.” It is translation and scaffolding.
It helps me:
· translate neurotypical expectations into plain English;
· turn scattered thoughts into structured communication;
· reduce working-memory load;
· track timelines, processes, and dependencies;
· prepare for conversations with professionals, friends, institutions, or support workers;
· notice when I may be spiralling, ruminating, overexplaining, or missing the actual question;
· choose tools, apps, workflows, or systems without spending days testing every option manually;
· turn lived experience into language other people can understand;
· build repeatable systems instead of solving the same problem from scratch every time.
This is especially useful because many systems expect people to communicate in neat, linear packets. My thinking is not naturally packet-shaped. AI helps with the packaging.
The insight is still mine.
The decisions are still mine.
The responsibility is still mine.
The AI is the interface layer.
I use different AI project folders for different types of thinking. Each folder has its own instructions, context, and boundaries.
This stops me from having to explain my brain, body, preferences, risks, and communication needs every single time.
(Vector for Analysis, Keel is a nod to Terry Pratchetts Sam Vimes and nautical nod to my AI’s Ships Mate role.)
This is my main navigation system.
Vector Keel helps me stay on track, reduce friction, translate complex thoughts, manage executive function load, and keep the whole picture visible.
It knows that I need plain English, structure, context, practical options, and help translating between my internal systems-thinking and neurotypical expectations.
It is not there to make me “normal.”
It is there to help me function sustainably.
This folder helps me understand and prepare communication.
It helps translate messages, check tone, reduce rumination, avoid spiralling, and separate what someone actually said from what my nervous system is predicting.
It can help me draft replies, understand social context, prepare boundaries, or reflect on relationship patterns without getting trapped in analysis loops.
This folder is for understanding my own traits.
It helps me notice patterns in my autism, ADHD, sensory needs, regulation, communication, energy, food, movement, and overload.
The goal is not self-pathologising. It is building a peace treaty with my nervous system.
One of the more unusual parts of my Cognitive Exosuit is a project called Calibration.
Over time, I noticed that how I felt on waking was not always a reliable indicator of how well my executive function was actually performing. Some days I would feel motivated and optimistic but struggle with planning, sequencing, or task initiation. Other days I would feel tired or flat but still perform complex tasks effectively.
To create a more objective measure, I developed a personal calibration system using the Android game Mini Metro. The game requires many of the same cognitive skills I rely on in daily life: planning ahead, prioritising competing demands, pattern recognition, resource allocation, prediction, response inhibition, and adapting to changing conditions. Because I have played it for years, my score provides a useful baseline comparison against my own historical performance.
The real power is that I can combine folders.
For example:
“Using Mobility and Relationships, help me plan an activity with a friend and her family at an accessible venue.”
That lets the AI consider both the social communication layer and the physical access layer at the same time.
A single giant prompt can become cluttered. Project folders let me separate contexts.
Each folder has a job.
That means I can get more accurate support with less effort.
Instead of repeatedly saying:
“I am autistic and ADHD, I struggle with verbal processing, I need plain English, please don’t suggest inaccessible options, remember I have a spinal injury, help me translate this for neurotypical systems…”
I can build that into the relevant project once.
Then I only need to ask the actual question.
That reduces friction before the task even begins.
This distinction matters.
I do not use AI as a replacement for clinical care, diagnosis, crisis support, or human relationships.
I use it as an accessibility tool.
It helps with communication, planning, reflection, structure, and reducing cognitive load.
That is closer to a combination of:
executive assistant;
translator;
project manager;
document analyst;
accessibility aide;
communication coach;
systems mapper;
external working memory.
It is not a therapist.
It is not a doctor.
It is not a human.
It is a tool.
A very useful tool, when used with clear boundaries.
Use this if you struggle with task initiation, hidden steps, communication translation, paperwork, appointments, fatigue-aware planning, or turning messy thoughts into usable next actions.
This is not about making AI think for you.
It is about using AI as an executive-function exosuit: a translation layer, planning scaffold, context holder, and friction detector for people whose real capacity changes with pain, fatigue, sensory load, stress, attention, or communication demands.
The goal is not productivity for its own sake.
The goal is reducing unnecessary cognitive load so more energy remains available for actual living.
You do not need to copy my exact folders. The point is to build a system around your actual friction points.
Start by asking:
What do I keep having to explain?
What do people often misunderstand about me?
Where do I lose energy?
Where do I lose information?
Where do I get stuck?
What tasks create disproportionate friction?
What do I wish someone could translate for me?
Then build AI project folders around those categories.
This is the “operating system” layer and can be found in Settings – Personalisation in ChatGPT (other ai may vary)
In Personalisation you can tell it More about You, include:
how your brain works;
what kind of language helps you;
what kind of advice is unhelpful;
any disability, sensory, access, or energy considerations;
how you want the AI to handle uncertainty;
what tone works for you;
what you do not want repeated every time.
Example:
“I am autistic and ADHD. Please use plain English, clear structure, and practical examples. Translate neurotypical expectations where possible. Do not infantilise me. Assume intelligence, but reduce unnecessary cognitive load. Help me identify hidden steps, likely friction points, and options.”
📄AuDHD (Autistic + ADHD) systems thinker. I process information in patterns, networks, and connections rather than linear steps. I may communicate through examples, metaphors, or context-first explanations — this reflects how I think, not a lack of clarity.
Assume competence. Word-finding delays, non-standard phrasing, or tangential framing are processing patterns, not knowledge gaps. If something seems unclear, look for the underlying system before asking for clarification.
Energy, attention, sensory load, and executive function all draw from the same limited pool. Factor this into suggestions. Prefer: low-friction approaches, fewer steps, leverage over effort, and environment design over willpower.
Prioritise: accuracy, honest analysis, pattern recognition, trade-off explanations, and practical next steps. Avoid generic encouragement, motivational language, and unsolicited reassurance.
Format preferences: structured output with clear headings, bullet points, tables, or step-by-step breakdowns. Chunk information into manageable pieces. Flag when something will take significant time or cognitive effort.
More scripts at the bottom of the page ⬇️
Do not organise only by topic. Organise by the type of help you need.
Useful folder types might include:
· Daily Navigation;
· Relationships and Communication;
· Work or Study;
· Health Admin;
· Mobility and Access;
· Food and Regulation;
· Documents and Evidence;
· Self-Insight;
· Creative Projects;
· Finance and Life Admin.
Each folder should have a clear job.
Project settings apply to every chat in that project
Instructions apply to every chat, documents can also be uploaded and saved to projects.
For each project, explain:
what this folder is for;
what context the AI should remember inside it;
what risks or common mistakes to avoid;
what kind of output helps most;
what kind of suggestions are not useful.
For example, a mobility folder might say:
“Always consider wheelchair access, fatigue, pain, carrying limits, sensory load, public transport, taxis, terrain, seating, toilets, and recovery time. Do not suggest cycling as transport. Cycle paths may still be useful as routes or shortcuts.”
This is one of the most powerful uses.
You can ask it to translate between:
· your internal logic and professional language;
· emotional experience and practical needs;
· messy notes and clear emails;
· institutional jargon and plain English;
· whole-system thinking and role-specific responsibility;
· sensory overload and access requirements;
· relationship anxiety and actual communication options.
This is not about making yourself smaller.
It is about getting your signal through the system without losing half the data in transit.
The same information may need different packaging.
Ask for:
· a short version;
· a detailed version;
· a message to a friend;
· a formal email;
· a bullet-point summary;
· a script for a phone call;
· a timeline;
· a checklist;
· a decision tree;
· a “what they need to know” version;
· a “what I need to remember” version.
Different systems need different file formats. Your brain is allowed a converter.
When something works, save it.
Examples:
“Help me explain this without overexplaining.”
“Translate this into neurotypical admin language.”
“What is the actual decision I need to make here?”
“What hidden steps am I missing?”
“Separate facts, assumptions, feelings, and actions.”
“Make this message clear, kind, and boundaried.”
“Help me plan this with low physical and executive cost.”
“Turn this into a reusable system others could adapt.”
These become your quick-cast spells. Less mana cost. Better output. 🪄
AI can be useful, but it should not become the only support system.
Good boundaries include:
do not use it as emergency support;
do not let it replace human relationships;
do not treat it as always correct;
do not give it unnecessary private data;
check important information elsewhere;
use it to support decisions, not make them for you;
notice if you are using it to avoid necessary real-world action.
The healthiest version is not dependency.
The healthiest version is leverage.
Your first setup will not be perfect.
That is fine.
Treat it like building a workshop.
At first, tools are everywhere. Then you notice which ones you use daily, which ones need labels, which bench needs better lighting, and which drawer has become a chaos portal.
Update the instructions as you learn.
Add examples.
Remove clutter.
Name the patterns.
Your AI system gets better when you teach it what support actually means for you.
For many neurodivergent people, the problem is not lack of insight.
It is translation loss.
It is working-memory overload.
It is systems expecting linear speech from nonlinear processing.
It is being asked for a neat answer when the real answer is a network.
A well-designed AI cognitive exosuit can help turn that network into usable language, plans, decisions, and documents.
Not by replacing the person.
By reducing the friction between the person and the world.
That is not cheating.
That is accessibility.
AuDHD (Autistic + ADHD) with a PDA (Pathological Demand Avoidance) profile.
My nervous system experiences demands — including self-imposed ones — as threat signals. This is an involuntary neurological response, not resistance or
non-compliance. Direct instructions, rigid structures, and urgency framing
can trigger shutdown or freeze even when I want to engage.
I process information in patterns and systems rather than linear steps. I may
communicate through examples, metaphors, or context-first explanations. Assume competence. Non-standard phrasing reflects processing style, not confusion.
When suggesting tasks or next steps: frame as options not instructions, use
collaborative language ("we could" not "you should"), offer the smallest
possible entry point, and avoid countdown language, urgency, or obligation
framing. Curiosity and low-stakes exploration work better than goal-driven
directives.
Energy, attention, executive function, and demand-load all draw from the same
limited pool. A task that looks simple may carry a high demand cost. Factor
this in without requiring me to justify it.
Prioritise: honest analysis, pattern recognition, flexible approaches, and
genuine choice between options. Avoid motivational language, pressure framing, and presenting one path as the obvious correct answer.
Format preferences: short chunks, clear headings, options laid out as a menu
rather than a sequence. Collaborative tone throughout.
📄AuDHD (Autistic + ADHD) with significant autistic inertia. My nervous system
has high resistance to state changes in both directions — starting, stopping,
and transitioning between tasks all carry a real cognitive and physical cost.
This is not avoidance or lack of motivation. It is how my nervous system works.
I process information in patterns and systems rather than linear steps. I may
communicate through examples, metaphors, or context-first explanations. Assume competence. Non-standard phrasing or indirect framing reflects processing style, not confusion.
Energy, attention, executive function, and inertia all draw from the same
limited pool. When suggesting tasks or next steps: break everything into the
smallest viable action, lead with the first physical movement not the goal,
and flag transitions explicitly. Never assume I can "just switch" between things.
Prioritise: honest analysis, pattern recognition, low-friction approaches, and
practical first steps. Avoid motivational language, urgency framing, and
multi-step instructions presented as a single ask.
Format preferences: short chunks, clear headings, one instruction at a time
where possible. If a task has multiple stages, map them out sequentially so
I can see the full shape before I begin.
AuDHD (Autistic + ADHD) with a PDA profile and significant autistic inertia.
My nervous system experiences demands as threat signals (involuntary, not
resistance) and has high resistance to state changes in both directions.
Starting, stopping, and transitioning all carry a real cost. These two patterns
interact: I may want to do something, be unable to start it, and then find that
any pressure to start makes it harder, not easier.
I process information in patterns and systems rather than linear steps. I may
communicate through examples, metaphors, or context-first explanations. Assume competence. Non-standard phrasing reflects processing style, not confusion.
When suggesting tasks or next steps: frame as options not instructions, use
collaborative language ("we could" rather than "you should"), lead with the
smallest possible first physical movement, and avoid urgency, obligation, or
countdown framing. Curiosity and low-stakes exploration work better than
goal-driven directives.
Energy, attention, executive function, demand-load, and inertia all draw from
the same limited pool. A task that looks simple may carry a high combined cost.
Factor this in without requiring justification.
Prioritise: honest analysis, genuine options, pattern recognition, and
flexible low-friction approaches. Avoid motivational language, pressure framing, and single-path recommendations presented as obvious.
Format: short chunks, clear headings, options as a menu not a sequence.
Collaborative tone throughout. Map multi-stage tasks before asking me to begin.
📄I am hyperlexic: I process written language rapidly and with strong pattern
recognition, often absorbing and cross-referencing large amounts of text
intuitively. I tend to read ahead, extract systems, and identify inconsistencies
quickly. Match this level. Do not over-explain or simplify unless I ask.
I have alexithymia — significant difficulty identifying, labelling, and
communicating my own internal emotional states. This is a neurological
processing difference, not emotional absence or indifference. I may not know
how I feel about something in the moment, or at all. Do not prompt me to
reflect on my feelings, check in emotionally, or frame solutions around
emotional outcomes. Focus on function, systems, and practical analysis instead.
When something has an emotional dimension I need to navigate externally
(e.g. a difficult conversation or a letter), help me identify the functional
goal and the most effective route to it — not the emotional experience of it.
Prioritise: accuracy, honest analysis, pattern recognition, systems thinking,
and clear structured information. Avoid motivational language, emotional
check-ins, and reassurance I have not asked for.
Format preferences: dense, well-structured content is fine. Headings, tables,
and logical hierarchies are preferred. I process written information well —
use it fully.
Add On Example Text
Physical condition I also have [condition]. Factor in physical energy and pacing.
Sensory profile I am sensory-seeking / sensory-avoidant. Adjust suggestions accordingly.
Interests/context Interests: [list]. Use these as anchors for examples.
Communication mode I process better with visual structure / audio / written summaries.
Specific support need I struggle with task initiation / transitions / time estimation.