Proposed AI Workshop for McKesson Healthcare Support Teams
Updated: 7 days ago
Updated 16 October 2026 · Practical guidance for healthcare teams
Hero image: AI-generated illustration of a healthcare learning session; not a photograph of an actual event.
This page presents an independent proposed AI workshop for a McKesson healthcare support audience. It is not a record of a completed session and does not claim a partnership, endorsement, appointment, or contract with McKesson. A useful learning scope could start with the paperwork around a fictional support request: organising known facts, identifying the missing information, and writing a respectful internal handover. Parikshit Khanna could discuss that scope with an authorised learning or operations contact if the organisation chose to enquire. The actual tasks, accounts, and reviewers would need to be agreed by the organisation before any session. This proposal focuses on understandable practice rather than unverified statements about McKesson's systems, workforce outcomes, or current AI strategy.
Start here
You do not need a technical background. Choose one small task, use fictional or approved public information, and check the result yourself. This guide covers a practical example, tool costs, and how guided training can help.
A proposed exercise in careful support-request handling
Use a fictional administrative request with several attachments and one unresolved question. Ask AI to prepare a checklist showing which supplied documents are present and which details need clarification. This demonstrates organisation and reading support without implying that the assistant can determine eligibility, coverage, a clinical outcome, or an approval.
The next activity could turn the checklist into a short handover for a colleague. Require a distinction between facts, assumptions, and questions. A reviewer should be able to locate the source of every important statement. If the draft cannot support that traceability, participants should practise revising or rejecting it.
Parikshit Khanna's proposed training could end with a small evaluation plan: compare several fictional requests and record the corrections needed. Any discussion of actual processes, products, or system connections would be a separate, authorised scope. The article does not establish McKesson's interest in the training or access to its internal operations.
McKesson is named as the intended audience of an independent proposal only. No affiliation, endorsement, contract, completed client engagement, or access to its systems is claimed by this article.
Create a checklist for a fictional administrative request
Invent a nonclinical support request requiring three documents, then supply only two fictional documents and an incomplete reference number.
Ask AI to list what is present, what is missing, and what should be asked next, without making an approval decision.
Check the list against the invented requirements and rewrite any sentence that presents an assumption as a confirmed fact.
Verify document status, references, and unanswered questions. Ensure the draft makes no eligibility, clinical, or coverage decision. Before any real workshop, confirm the authorised sponsor, intended audience, approved material, access arrangements, and commercial terms through a direct organisational discussion.
AI and generative AI, in ordinary words
AI is the broad name for computer tools that help with tasks such as finding patterns, sorting information or making predictions. Generative AI is one kind that creates a draft of text, images or other content from an instruction. That instruction is often called a prompt. A workflow simply means the steps used to complete a task. A fluent answer can still contain mistakes.
A useful instruction says what you want, who will read it, which approved facts to use, and what must stay unchanged. For example: “Rewrite this fictional clinic notice in friendly, short sentences. Keep every date unchanged. Do not add medical advice. Flag anything unclear.”
What are Claude, ChatGPT and Microsoft Copilot?
Claude is Anthropic’s AI assistant. You can ask it to draft, simplify or organise information, then review its work.
ChatGPT is OpenAI’s AI assistant. Use the same careful approach: give clear instructions, check sources, and review the draft.
Microsoft Copilot is a family of AI assistants. Personal Copilot and workplace Copilot have different access and controls. Ask IT which work account and licence your organisation has approved. Dragon Copilot is a separate healthcare product; ordinary Copilot is not a clinical system.
Free or paid: what should you choose?
Start with a free option and fictional examples. Pay only when a specific limit or feature is stopping useful work. A personal paid subscription does not automatically permit patient information. Check account controls, organisational approval and total cost before buying.
Claude reference prices checked 16 October 2026: Free is US$0 with limits; Pro is US$20 a month or US$200 billed yearly. Team Standard is US$20 per person per month billed yearly, or US$25 monthly, for 2–150 people. Taxes and local checkout terms may differ. More expensive plans are available; beginners do not need to start there.
Microsoft Copilot has a free personal version. Copilot Chat is included with eligible Microsoft 365 work plans; some agents can add usage charges. Paid workplace features may require a separate licence. Ask your IT or purchasing team for your country’s current price, eligible base licence, taxes and renewal terms.
ChatGPT has free and paid plans. Check the official plan information and your local checkout before paying; features and limits can change. Tool subscriptions are separate from workshop fees.
Draft with AI. Check it yourself. Share only after approval.
Protect patient information and keep a person responsible
For practice, leave out real names, reports, prescriptions, phone numbers, appointment lists and other identifying details. Use invented information. Before any patient information is used, your organisation should approve the tool, account, contracts, settings and workflow. Removing a name alone may not remove every identifying detail.
WHO warns that generated answers may be false, incomplete or biased. Read the original source, check dates and numbers, and have the appropriate professional approve the final text. General chatbots should not make diagnosis, treatment, dosage, triage or urgent-care decisions.
Follow AI news without feeling left behind
You do not need every new model or subscription. Pick one useful task and practise it for a week. Set aside a short weekly news-reading slot. For each announcement, ask: Is it available here? Is our organisation allowed to use it? Does it solve a task we actually have?
Two developments worth understanding: Anthropic announced Claude for Healthcare on 11 January 2026; Microsoft described Dragon Copilot expansion on 5 March 2026. These are vendor announcements with market, licence and feature limits, not proof that a general chatbot can deliver better patient outcomes.
How Parikshit Khanna’s training can help
Parikshit Khanna is an AI trainer and Founder of Digital Training Jet. A guided session gives your team time to practise clear instructions, spot mistakes, check sources and agree who approves the final work. Ask for examples based on your department’s everyday tasks and time for beginner questions.
A participant account published by OncoDaily describes learning from Parikshit at a ChatGPT and AI tools workshop for healthcare professionals held at IIT Delhi. This describes a workshop experience; it does not claim institutional endorsement or clinical certification.
AI-generated illustrative portrait based on supplied references.
Plan a session around your team’s work
Share your team’s roles, group size, preferred language, two or three repetitive tasks, approved tools and possible dates. Ask for a proposal covering practice time, take-home examples, review responsibilities and follow-up support. Workshop fees, delivery format and travel are confirmed in the proposal. Results should be measured in your own setting.
Common questions
Do I need coding skills? No. Begin with a clearly described task and a short draft you can check.
Will AI replace professional judgement? Your team remains responsible for the final output. Training focuses on careful use and review.
Should everyone buy a paid account? Start with the tools your organisation already approves. Buy additional access only for a demonstrated need.
