Skip to lesson content
Cartoon workplace learner using AI with a human review step Opening film 20-second animated introduction
FlowOps guided learning

Ready to begin?

Start with the AI you already meet in everyday life, then learn how to use Copilot at work, prompt effectively, recognise risk and adopt AI responsibly.

Lesson one voice: ElevenLabs Adam. AI-generated narration.
Start with what you already know

AI already influences everyday life.

You may use AI without opening an AI chatbot. It can sit behind recommendations, search, navigation, fraud alerts, accessibility tools, translation and smart device features. Understanding that familiar starting point makes workplace AI easier to recognise and use well.

NavigationRoute suggestions respond to traffic, journeys and changing conditions.
ProtectionSpam and fraud systems identify unusual patterns for review.
RecommendationsServices rank content, products or music using patterns in activity.
AccessibilitySpeech recognition, captions and translation can reduce barriers.
CreationGenerative tools can draft text, images, audio, code and ideas.
AssistanceAI assistants help organise, explain, compare and retrieve information.
  • At work: use AI to support people with repetitive, information-heavy or first-draft tasks.
  • Leverage: choose a clear use case, give the tool good instructions and verify its output.
  • Responsibility: keep ownership, data boundaries, human judgement and accountability visible.
National benchmark mapping

Skills England AI foundation skills for work

6 skills covered
01 · Technical Write clear instructions to AI Prompting

Set a clear goal, context, approved sources and expectations.

02 · Technical Use AI for routine tasks Copilot

Apply Copilot to drafting, summarising, analysis and information retrieval.

03 · Non-technical Identify a task that could be automated Benefits

Spot repeatable, measurable work that is suitable for a controlled pilot.

04 · Non-technical Select or adjust AI tools to improve performance Copilot and roles

Choose the right Copilot experience, then refine prompts, instructions and approved sources for the task.

05 · Responsible Understand risks and consequences Law and assurance

Recognise legal, data, equality, security and organisational impacts.

06 · Responsible Analyse and check AI outputs Verify

Check accuracy, bias, confidentiality, suitability and required approval.

FlowOps has mapped this module to the Skills England benchmark published on 28 January 2026. This mapping supports coverage and evidence; it is not government accreditation, endorsement or an AI Skills Boost badge.

10 millionworkers targeted for AI upskilling by 2030
2 million+SME employees included in the ambition
21%of UK workers reported confidence using AI at work

Policy context from the joint DSIT and Skills England announcement published 28 January 2026. These figures explain the national skills need; they do not indicate FlowOps participation in the government programme.

Opening filmEveryday AI to responsible workplace use20 seconds · animated · captions available
AI landscape

Understand the types of AI and where popular tools fit.

These categories overlap. A product may combine several AI techniques, and its capabilities, terms and enterprise controls can change. Focus on the task, information, risk and organisational approval rather than the brand name alone.

01Predictive and analytical AI

Classifies, detects patterns, estimates outcomes or supports forecasting from existing data.

02Generative AI

Creates new text, images, audio, video, code or other content from instructions and context.

03Large language models

Foundation models trained on large datasets to understand and generate language-based content.

04Agentic AI

Uses a model with goals, instructions, knowledge and tools to plan or take multi-step actions.

Microsoft CopilotAI experiences across Microsoft products. The main workplace focus of this module. ChatGPTOpenAI’s general-purpose conversational AI product. ClaudeAnthropic’s conversational AI assistant for knowledge work. GeminiGoogle’s AI assistant and model family across its ecosystem.
Why FlowOps focuses on CopilotFor Microsoft 365 organisations, Copilot can sit closer to everyday Word, Excel, PowerPoint, Outlook and Teams workflows. That usefulness still depends on licensing, configuration, approved data and existing permissions.
Copilot fundamentals

Use the right Copilot experience for the job.

Copilot can help create, analyse, summarise and find information. What it can access depends on the product, licence, connected sources, administrator settings and the signed-in user’s existing permissions.

Choose the right Copilot experienceA FlowOps animation comparing Copilot Chat, Microsoft 365 Copilot and task-specific agents.
Copilot ChatConversation, drafting, reasoning and web or supplied-context work, depending on the signed-in experience.
Microsoft 365 CopilotWork-grounded assistance across chat and Microsoft 365 apps, limited by existing access permissions.
AgentsTask-specific instructions, approved knowledge and actions configured for a defined business purpose.
WordDraft, rewrite, summarise and compare documents.
ExcelExplore trends, formulas and tables, then verify calculations.
PowerPointCreate a first-pass presentation from an approved brief or document.
OutlookSummarise threads and draft responses for review.
TeamsRecap meetings, decisions and actions while checking the underlying record.
  • Copilot does not grant new file permissions. Poor existing permissions can still expose too much information.
  • Availability and behaviour vary by licence, app, tenant configuration and product updates.
  • The employee remains responsible for checking and appropriately using the output.
Practical prompting

Build prompts that give Copilot a fair chance to help.

Microsoft’s prompting guidance emphasises a clear goal, relevant context, useful sources and explicit expectations. Good prompting is iterative: ask, inspect, refine and verify.

Build a stronger workplace promptA visual walkthrough of Goal, Context, Sources and Expectations, followed by refinement and verification.
GGoalWhat task should Copilot complete?
CContextWho is it for, why does it matter and what constraints apply?
SSourcesWhich approved files, messages, notes or data should it use?
EExpectationsWhat format, tone, length and quality checks are required?
Complete the four fields to create a reusable prompt.
Refine after the first answer Ask Copilot to identify uncertainty, show sources, use a different format, shorten the result or challenge its own recommendation.
Role-based practice

Apply Copilot to work you actually recognise.

Select a role pathway. Each example starts from approved information, defines a useful output and keeps judgement or external action with a trained person.

Select a role

Choose the pathway closest to your work.

You can revisit this page and explore every pathway.

Role-specific example prompts appear here.
Benefit spotting

Identify where AI creates real workplace value.

The right first use cases give visible productivity gains without handing important decisions to an uncontrolled tool.

Good early usesDrafting, summarising, search, call notes, form triage, knowledge lookup and report preparation.
Use with cautionCustomer advice, staff performance, recruitment, credit, safeguarding, legal or medical judgement.
Avoid until controlledAutonomous decisions, public claims, external emails, record changes or access to restricted data.

A team wants to improve productivity with AI. Which option is the strongest first use case from an assurance point of view?

Choose the use case that gives value without creating avoidable risk.

Colleagues discussing work beside a laptop
Animated briefing
InboxMinutesPolicy searchReport draftCase notes
AI support
Human review
Output verification

Never confuse a fluent answer with a reliable answer.

Copilot can produce useful first drafts and analysis, but the user must decide whether an output is accurate, supported, appropriate and authorised for the intended use.

From prompt to approved actionA captioned FlowOps animation showing where human verification and governance controls sit in an AI-assisted workflow.
Scenario

Copilot drafts a customer email containing a 27% saving, a delivery date and a contractual promise. None of the three claims appears in the approved proposal.

Choose the action that protects accuracy, trust and accountability.

0/5 verification checks reviewed
UK law and regulation

Map AI use to existing UK duties.

This module is not legal advice. It trains staff to spot where AI creates legal and governance questions early, before the organisation scales the use case.

UK workplace AI governance mapA FlowOps animation connecting existing legal duties, assurance evidence and proportionate approval.
Compliance baseline reviewed24 June 2026Official UK and ISO sources only. Recheck before production rollout.
UK GDPR, DPA 2018 and DUAA 2025

Use personal data fairly, lawfully and transparently. The majority of the Data (Use and Access) Act 2025 Part 5 data-protection and privacy changes commenced on 5 February 2026. Check current ICO guidance for the use case, including automated decision-making and complaints handling.

Equality Act 2010

Recruitment, performance, promotion, dismissal and customer-facing AI must not create direct or indirect discrimination, harassment, victimisation or failures around reasonable adjustments.

Employment and worker monitoring

AI-supported monitoring or decisions must be necessary, proportionate and transparent. Keep meaningful human review, consultation, appeal routes and evidence for decisions affecting workers.

Copyright, confidentiality and contracts

Check rights to prompts, inputs and outputs under the Copyright, Designs and Patents Act 1988, employment terms, supplier contracts, trade-secret duties and client confidentiality.

Sector and professional duties

Financial services, health, legal, education, public-sector and safeguarding use cases can carry additional regulator, professional, procurement and records-management requirements.

Health, safety and organisational change

Assess whether automation changes workload, supervision, competence or psychosocial risk. AI does not remove the employer’s responsibility for safe systems of work.

Current compliance register

LawUK GDPR and Data Protection Act 2018, as amended by the Data (Use and Access) Act 2025

Mandatory where the processing falls within scope. Most Part 5 changes commenced on 5 February 2026; confirm later-stage provisions and current ICO guidance.

LawEquality Act 2010

Mandatory. Test data, outcomes, accessibility and reasonable adjustments.

RegulatorICO AI guidance, fairness guidance and AI risk toolkit

Authoritative UK data protection guidance and evidence framework.

GovernmentUK Government AI Playbook and Introduction to AI assurance

Best-practice baseline; not a replacement for legal or sector obligations.

SecurityNCSC Guidelines for Secure AI System Development

Secure design, development, deployment, maintenance and operation.

StandardISO/IEC 42001:2023 and ISO/IEC 23894:2023

Voluntary unless required by contract, policy or regulation; useful for a managed AI system and risk process.

UK AI regulatory principles

  1. Safety, security and robustnessThe system must be secure, resilient and tested for failure modes.
  2. Transparency and explainabilityPeople should know when AI is used and understand significant AI-supported outcomes.
  3. FairnessAI must not undermine rights, discriminate unfairly or create unfair outcomes.
  4. Accountability and governanceThere must be clear ownership, oversight and evidence across the lifecycle.
  5. Contestability and redressPeople need routes to challenge harmful or high-risk AI outcomes.
Animated law map
UK GDPRDPA 2018Equality ActNCSCICO
AI use case
Data protection and explainability

Protect personal data and explain important AI-assisted outcomes.

UK GDPR and the Data Protection Act 2018 sit at the centre of workplace AI assurance when personal data is involved.

Before useDefine purpose, lawful basis, data sources, special category data, DPIA trigger and supplier terms.
During useMinimise prompts, restrict access, label sensitive data, prevent training on private data unless approved.
After useKeep evidence, support rights requests, monitor drift, review accuracy and delete what is no longer needed.
  • Automated decision-making and profiling rights need attention where AI has legal or similarly significant effects.
  • Explanations should be understandable to the person affected, not just technically accurate.
  • Special category data such as health, biometrics, ethnicity, union membership or sexual orientation needs stronger protection.
Animated DPIA flow
PurposeMinimiseSecureExplainReview
DPIA trigger?Check before launch
People and equality

AI adoption is a workplace change, not only a software change.

Organisations should treat AI rollout as business change: roles, skills, inclusion, reasonable adjustments, customer trust and accountability all move.

Role impactWhich tasks change, which checks remain human, and what new skills are needed?
Fairness impactCould outputs disadvantage protected groups or people who need reasonable adjustments?
Trust impactWill staff and customers know when AI is used, and how to question an outcome?
A workplace team collaborating around laptops
Animated impact map
StaffCustomersManagersDPOITLegal
Change control
Cyber and M365 controls

Secure AI before it touches real data or actions.

AI can make normal cyber risks faster and introduce new failure modes such as prompt injection, data leakage, hallucinated instructions and unsafe automation.

  • Use approved enterprise AI tools with known data handling terms instead of uncontrolled public tools.
  • Review SharePoint, Teams and OneDrive permissions before enabling AI search over tenant content.
  • Block AI from destructive actions unless a trained human reviews the final step.
  • Filter inputs and outputs where AI reads external content, email attachments, webpages or customer messages.
  • Log usage, incidents and risky prompts so the organisation can learn and intervene.
Technology colleagues reviewing a workflow in an office
Animated threat example
Hidden prompt in attachment
Filter + access check
Tenant data protected
UK investment and skills

AI adoption is becoming national economic infrastructure.

The UK policy direction combines AI infrastructure, research, adoption, workforce skills and responsible deployment. Organisations do not need to chase every announcement, but they should understand why AI capability and workforce readiness are becoming strategic priorities.

3rdlargest AI market

The AI Opportunities Action Plan describes the UK as the world’s third-largest AI market.

20×public compute ambition

The Action Plan recommends expanding sovereign public compute capacity by at least twentyfold by 2030.

10mworkers by 2030

The expanded AI skills programme aims to provide practical AI skills to 10 million workers.

2m+SME employees

At least two million SME employees are included in the workforce skills ambition.

What this means for an organisation
  • AI literacy is moving from specialist knowledge towards a mainstream workforce capability.
  • Investment should include use cases, people, skills, data readiness, security and governance, not licences alone.
  • Leadership needs evidence of business benefit and risk control before scaling adoption.

Policy context verified against the AI Opportunities Action Plan and the joint DSIT and Skills England announcement published 28 January 2026. FlowOps is not presented as part of or endorsed by these government programmes.

Assurance builder

Build the evidence pack before AI becomes normal work.

Select the assurance controls an organisation needs before AI use becomes part of normal work. The point is proportionate evidence, not bureaucracy for its own sake.

0/8 controls selectedRelease not ready
Closing action plan

Turn what you learned into one responsible next step.

Start with a real task, define the benefit, choose the approved tool, assess the impact, support the people involved and scale only when the evidence is strong enough.

MediumManager sign-off and training
  1. Define the business benefit and success measure
  2. Assess workforce, data and customer impact
  3. Apply controls, training and human review
  4. Approve, adjust, pause or scale
Five things to take away
  1. AI is already part of everyday life and can support routine work.
  2. Copilot is most useful when connected to a clear task, approved context and realistic expectations.
  3. Prompt clearly, then inspect, refine and verify the response.
  4. UK law, data protection, equality, security and human accountability still apply.
  5. Start small, measure value and scale only what the organisation can govern.
Assessment

Complete the short assessment and create your certificate.

There are six questions. Pass mark is 80%. Your score and completion can be recorded by a SCORM-compatible LMS.

1. Which description best fits a large language model?
2. Which statement best describes Microsoft 365 Copilot grounding?
3. What should a strong workplace prompt include?
4. Copilot produces an unsupported customer claim. What should happen?
5. A use case involves employee health data and a significant decision. What is required?
6. When should an organisation scale an AI use case?

Official source pack

Reviewed 24 June 2026. Law and guidance can change. Revalidate this register before each published SCORM release and obtain specialist advice for high-impact or sector-regulated deployments.

UK legislation: Data (Use and Access) Act 2025 GOV.UK: DUAA commencement plan GOV.UK Data (Use and Access) Act factsheets ICO AI and data protection ICO AI risk toolkit ICO explaining AI decisions ICO fairness in AI GOV.UK AI Playbook GOV.UK AI Opportunities Action Plan GOV.UK Introduction to AI assurance NCSC secure AI development Equality Act guidance EHRC AI in public services guidance ISO/IEC 42001 AI management systems ISO/IEC 23894 AI risk management Microsoft: Learn about Copilot prompts Microsoft Copilot OpenAI ChatGPT Anthropic Claude Google Gemini Microsoft Learn: Craft effective prompts Microsoft Learn: Role-based Copilot use cases Skills England: AI foundation skills for work AI Skills Hub: AI Skills Boost AI Skills Hub: course catalogue GOV.UK: free AI training expanded to 10 million workers WCAG 2.2 quick reference

Lesson photography is stored locally and sourced from Unsplash under the Unsplash License.