How Autistic and ADHD Professionals Can Use AI at Work (2026)

TL;DR. AI is most useful to autistic and ADHD professionals for the work around the work. It can draft and rehearse a hard message, give you the likely readings of an ambiguous email, cut long documents and meetings down to what's asked of you, break a big task into ordered steps, plan a day, and draft an accommodation request for you to edit. Two rules make it safe. Keep private things out of the chat (diagnoses, medical details, other people's information), and remember that a work account may be visible to your employer. And you stay in control: the AI drafts, you decide.

Plenty of professionals find that the job itself is fine, and what drains them is everything around it. That might be a reply they can't find the right tone for, or a 40-message thread with one buried question in it. It might be a project with no obvious first step, or a meeting that ended without anyone saying what happens next. General-purpose AI assistants like Claude, ChatGPT and Gemini are good at exactly this kind of language and organizing work. This guide covers the practical uses, the privacy rules that matter more than usual here, and how to ask your employer for adjustments if you need them.

A note on language: we use identity-first wording ("autistic professionals") because many autistic people prefer it. Some people prefer person-first wording. Use whatever fits you.

This guide is general information, not medical, legal or employment advice, and AI is not therapy or treatment.

Where AI helps, at a glance

If this part of work takes a lot out of you	Ask the AI to	You still decide
A message you keep putting off	Draft three versions (direct, warm, brief) and rehearse the reply	Which version, and what to change
An email whose tone you can't read	List the possible readings and what in the text points to each	Whether to ask the sender directly
A long document, thread or meeting	Pull out decisions, deadlines and anything asked of you	What matters, and what to double-check
A big task with no clear start	Break it into ordered steps with a first step under 15 minutes	The order and the scope
A day with too many things in it	Turn your list into a realistic plan with breaks	What gets dropped
Asking for adjustments	Draft a short, specific request you can edit	What to disclose, and whether to send it

Draft and rehearse the hard messages

AI is very good at producing a first draft of the message you've been staring at. Give it the situation and what you want to happen, not your feelings about it, and ask for options:

I need to tell my manager I can't take on the new report this sprint because I'm already at capacity with X and Y. I want to sound cooperative, not defensive. Write three versions: one very direct, one warmer, one under 50 words. Don't apologize more than once.

Then rehearse. Ask it to play the other person: "Reply as my manager might if they push back. Then help me answer." Running the conversation two or three times in text can make the real one less of an unknown. It also helps with phone calls and meetings: ask for a short script with the three points you need to make, in order.

Two cautions. The AI doesn't know your manager or your workplace's norms, so treat its "how they might react" as one guess, not a prediction. And edit the draft until it sounds like you. People notice when a message doesn't match how someone usually writes.

Check the tone of an email

If working out what someone meant takes a lot of your energy, an AI can lay out the possibilities. The prompt that works best asks for evidence, not a verdict:

Here's an email from a colleague. List the two or three most likely ways to read its tone, and for each one, quote the words that point to it. Then tell me whether anything in it actually asks me to do something.

Compare that with "Is this email rude?", which invites a confident one-word answer the model can't really back up. AI reads tone the same way it reads everything: by pattern. It has no knowledge of the sender's habits, your history with them, or the office culture. When the stakes are real, use the AI to narrow the options, then check with a trusted colleague or simply ask the sender: "Just to make sure I've understood: do you need this by Friday, or is next week OK?"

Turn long documents and meetings into what's asked of you

Summarizing is one of the things these tools do most reliably. Ask for a structure, not just "a summary":

Summarize this document in three parts: (1) decisions already made, (2) deadlines with dates, (3) anything that asks me, [your role], to do something. Quote the exact sentence for each item in part 3.

Asking for the exact sentence lets you check the AI didn't invent or drop a request. That matters because AI can make things up with total confidence. For meetings, many workplaces already turn on transcription in Teams, Zoom or Google Meet. If yours does, paste the transcript in and ask for the same three parts. Don't record or transcribe a meeting yourself without checking your company's policy and the other attendees' consent.

Break big tasks into ordered steps

A large, vague task ("prepare the Q4 review") is hard to start for a lot of people. AI is good at turning it into steps:

Break "prepare the Q4 review" into ordered steps. Make the first step something I can finish in under 15 minutes. Mark which steps need information from someone else, and from whom.

If the list is too long, ask it to group the steps into three stages and show only the first stage. If a step is still too big, paste that step back and ask it to break it down again. The Job Accommodation Network lists checklists, task flow charts and written instructions among accommodation ideas, and you can make all of those yourself in a few minutes.

Plan a day you can actually do

Give the AI your list, your fixed meetings and your real constraints, and ask for a plan rather than a schedule packed to the minute:

Here are my meetings today and eight things on my list. I have about four hours of focus time. Pick the three that matter most given these deadlines, put them in my best focus hours (mornings), add breaks, and list what should move to tomorrow.

Ask it to explain why it chose the order. If the reasoning is wrong (it doesn't know a deadline moved, or that one task needs a quiet room), correct it and ask again. The plan is only as good as what you tell it.

Draft a workplace accommodation request

If changes at work would help (written instructions after meetings, a quieter desk, noise-cancelling headphones, regular check-ins, notice of changes), you can ask for them. AI can help you write the request. The rules about how to ask come from your country's law, not the AI, so here is what the official sources say.

In the US, the EEOC's guidance says that to request an accommodation, a person "may use 'plain English' and need not mention the ADA or use the phrase 'reasonable accommodation,'" and that requests "do not need to be in writing." The Job Accommodation Network (JAN) suggests putting it in writing anyway, so you have a record. Your employer is allowed to ask for documentation when the disability or the need isn't obvious. The EEOC says that should be limited to what's needed to show you have a disability and need the accommodation, and that in most situations an employer can't ask for your complete medical records. Cost is often smaller than people expect: in JAN's survey of employers who contacted it between 2019 and 2024, 61% of those who reported costs said their accommodations cost nothing to put in place. JAN gives free, confidential one-on-one guidance to employees as well as employers.

In the UK, ACAS says employers "must make reasonable adjustments by law." You can ask in writing or in a meeting, and ACAS publishes a free request letter template. ACAS also says neurodivergence "will often amount to a disability under the Equality Act 2010," and that "a worker does not need a diagnosis to be considered disabled." Separately, the government's Access to Work scheme can help pay for things like assistive software or a job coach. Its eligibility page names autism and ADHD and says "you do not need to be diagnosed with a condition to apply." It covers England, Scotland and Wales; Northern Ireland has a different system.

A prompt that keeps your private details out of the chat:

Draft a short, polite email to my manager asking for workplace adjustments. Don't mention any diagnosis. Say that I work best with: written follow-ups after meetings that assign me tasks, a few days' notice before priorities change, and permission to use noise-cancelling headphones at my desk. Ask to meet for 15 minutes to discuss it. Keep it under 150 words.

You choose what to disclose and when. The AI doesn't need your diagnosis to write a clear request, and leaving it out of the prompt is the simplest way to keep it private. Before sending, read it as your manager would: is each request specific, and is it clear what you're asking them to do?

Keep private things private

This matters more than usual here, because the things you might want help with (a diagnosis, a disagreement with a manager, an accommodation request) are personal.

What not to paste into any AI chat:

• Diagnoses, medication, or other medical details
• Other people's personal information (a colleague's name alongside a complaint, a client's details)
• Confidential company material your employer hasn't approved for that tool
• Anything you'd be upset to see stored, reviewed, or read by someone else

Personal accounts. On the consumer plans of Claude, ChatGPT and Gemini, your conversations can be used to train models unless you switch that off. Our guide Does AI train on your data? gives the exact setting for each app. In Claude it's Settings → Privacy → "Help Improve our AI models". For a one-off sensitive chat, Claude's Incognito chats, ChatGPT's Temporary Chat and Gemini's Temporary chat are not used for training.

Work accounts. Business plans generally don't train on your data by default. Anthropic says it won't use inputs from Claude for Work for training, and OpenAI says the same for business data. But "not used for training" is not the same as "private from your employer." Anthropic says a Team or Enterprise plan's Primary Owner can request data exports "which may contain your conversations with Claude," and that incognito chats on those plans are included in those exports. OpenAI says ChatGPT Business admins "can view, access, export, and delete end user conversations in the workspace." Google gives Workspace admins audit logs of Gemini activity and control over how long conversations are kept.

So a simple rule: draft anything about your own health or accommodations in a personal account with training turned off, or leave the personal details out entirely, and use your work account for work content your employer has approved. Not sure what a specific plan does? The training and retention lookup checks it in one step.

You stay in control

AI is a drafting and organizing tool. It doesn't know you, your job, or your workplace, and it can be confidently wrong.

• You make the decisions. What to send, what to disclose, what to prioritize and whether to follow the AI's suggestion at all are always yours. If a suggestion doesn't fit, ignore it.
• Tell it how you like information. Most assistants let you save standing instructions or preferences in their settings, so you don't have to repeat them. Something like "Use short bullet points, put any action I need to take first, no small talk" is a good start.
• Check anything factual. Dates, names, numbers and quotes in a summary should be checked against the original before you act on them.
• Its read on people is a guess. Use it to widen your options, not to settle what someone "really" meant.
• Stop if it adds stress. If going back and forth with a chatbot is becoming its own chore, that's a sign to close it. It's meant to take work off your plate, not add to it.

What AI can't do

It can't replace support from people: a manager who agrees to clearer instructions, a colleague you trust, an occupational health service, a coach, or a clinician. It isn't therapy or treatment. It won't know about unwritten rules unless you tell it, and it can't make your employer agree to anything. What it can do is take a share of the drafting, decoding and organizing off you, so more of your energy goes to the work you're good at.

---

Sources

All sources checked live on October 19, 2026.

1. Job Accommodation Network (JAN) — Accommodation and Compliance: Autism Spectrum (accommodation ideas including written instructions, checklists, task flow charts, noise-canceling headsets): https://askjan.org/disabilities/Autism-Spectrum.cfm
2. JAN — Accommodation and Compliance: Attention Deficit/Hyperactivity Disorder (key accommodations including to-do lists, assistive technology, written instructions): https://askjan.org/disabilities/Attention-Deficit-Hyperactivity-Disorder-AD-HD.cfm
3. JAN — Costs and Benefits of Accommodation (2019–2024 employer survey; 61% of employers reporting costs said accommodations cost nothing): https://askjan.org/topics/costs.cfm
4. JAN — Employees' Practical Guide to Requesting and Negotiating Reasonable Accommodations: https://askjan.org/publications/individuals/employee-guide.cfm
5. JAN — About JAN (free, confidential, one-on-one guidance): https://askjan.org/about-us/index.cfm
6. US EEOC — Enforcement Guidance: Reasonable Accommodation and Undue Hardship under the ADA (plain-English requests, no writing requirement, documentation limits): https://www.eeoc.gov/laws/guidance/enforcement-guidance-reasonable-accommodation-and-undue-hardship-under-ada
7. ACAS — Reasonable adjustments at work: https://www.acas.org.uk/reasonable-adjustments
8. ACAS — Asking for reasonable adjustments: https://www.acas.org.uk/reasonable-adjustments/asking-for-reasonable-adjustments
9. ACAS — Reasonable adjustments for neurodiversity (no diagnosis needed): https://www.acas.org.uk/reasonable-adjustments/adjustments-for-neurodiversity
10. ACAS — Neurodiversity at work: https://www.acas.org.uk/neurodiversity-at-work
11. ACAS — Reasonable adjustment request template: https://www.acas.org.uk/reasonable-adjustment-request-template
12. GOV.UK — Access to Work: https://www.gov.uk/access-to-work and eligibility: https://www.gov.uk/access-to-work/eligibility
13. Anthropic Privacy Center — How do I change my model improvement privacy settings?: https://privacy.claude.com/en/articles/12109829-how-do-i-change-my-model-improvement-privacy-settings
14. Anthropic Privacy Center — Is my data used for model training? (commercial products): https://privacy.claude.com/en/articles/7996868-is-my-data-used-for-model-training
15. Anthropic Help Center — Who owns and manages the data of my Team?: https://support.claude.com/en/articles/9265372-who-owns-and-manages-the-data-of-my-team
16. Anthropic Help Center — Use incognito chats (included in Team/Enterprise exports): https://support.claude.com/en/articles/12260368-use-incognito-chats
17. OpenAI — Enterprise privacy (no training on business data by default; Business admin access to conversations): https://openai.com/enterprise-privacy/
18. Google Workspace — Generative AI in Google Workspace Privacy Hub (no training outside your domain without permission; admin audit logs and retention): https://knowledge.workspace.google.com/admin/generative-ai/generative-ai-in-google-workspace-privacy-hub
19. The AI Career Lab — Does AI train on your data? and Why does AI make things up?

Claude is a product of Anthropic, ChatGPT of OpenAI, and Gemini of Google. The AI Career Lab is not affiliated with any of them. This article is general information, not legal, medical or employment advice; for your situation, JAN (US) and ACAS (UK) give free guidance.