The 2026 Skills Index.

What it is: A working understanding of what AI systems can and cannot do, how they produce output, and where their limits and risks sit.

Why now: Understanding what AI can and cannot do is now baseline professional fluency, the way digital literacy was a decade ago. 

Build it: Take Elements of AI (University of Helsinki, free) or Google's AI Essentials, then run one real weekly task through an AI tool.

Level 3 looks like: You use AI tools routinely, explain their strengths and failure modes in plain language, and judge when output is safe to use without help.

What it is: The craft of writing instructions, examples and context so an AI system produces accurate, usable output first time.

Why now: Named on LinkedIn's 2026 rising list. The quality of AI output tracks the quality of the instructions and context you give it. 

Build it: Study the free prompting guides from Anthropic, OpenAI and Google, and keep a personal library of prompts that worked.

Level 3 looks like: You write prompts that reliably produce usable output, reuse a tested personal library, and fix a weak result by changing the instruction rather than the tool.

What it is: Confident, habitual use of the AI assistants built into the office tools you already work in, such as copilots in documents, email and spreadsheets.

Why now: Copilots now sit inside Office, Google Workspace and most CRMs, and the gap between fluent and hesitant users widens monthly. 

Build it: Pick the assistant inside tools you already use and commit one daily task to it for a month.

Level 3 looks like: You complete routine drafting, summarising and analysis with built in assistants daily, unaided, and choose sensibly when not to use them.

What it is: Designing multi step automations in which AI agents plan, execute and hand off work under defined rules and human checkpoints.

Why now: The 2026 shift is from single prompts to agents that plan and execute multi step work, and someone must design and supervise them. 

Build it: Build one small automation in Zapier, Make or n8n with an AI step, then add error handling and a human checkpoint.

Level 3 looks like: You can build a working multi step automation with an AI step, error handling and a human checkpoint, and it runs without you nursing it.

What it is: A conceptual grasp of how models are trained, evaluated and degraded, covering training data, accuracy measures and drift.

Why now: You need not train models to benefit, but knowing how training data, evaluation and drift work helps you judge vendors and spot weak claims. 

Build it: Take DeepLearning.AI's AI for Everyone on Coursera, or Google's free Machine Learning Crash Course.

Level 3 looks like: You can explain training, evaluation and drift to a colleague, question a vendor's accuracy claims sensibly, and spot when a model is being oversold.

What it is: Connecting language models to your own data and systems through APIs, retrieval and integration patterns to create useful tools.

Why now: Connecting models to your own data through APIs and retrieval is becoming the standard pattern for genuinely useful internal tools. 

Build it: Work through the official Anthropic or OpenAI API quickstarts, then build one retrieval tool over your own documents.

Level 3 looks like: You have shipped a small working LLM tool over real data that colleagues actually use, and you can explain its retrieval pattern and its limits.

What it is: The policies and practices that keep AI use lawful, fair and safe, covering data protection, transparency, oversight and accountability.

Why now: Data governance and responsible AI are named on LinkedIn's 2026 list, as the EU AI Act and UK guidance raise the bar for safe deployment. 

Build it: Read the ICO's AI and data protection guidance, then draft a one page AI use policy for your own team.

Level 3 looks like: You apply your organisation's AI policy correctly without prompting, flag risky use cases early, and can justify each tool's data handling if asked.

What it is: Systematically checking AI generated work for accuracy, bias and fitness for purpose before it is used or published.

Why now: Models make confident mistakes, so verification, testing and review are exactly where human value now concentrates. 

Build it: Define what good looks like before you prompt, then check sources, numbers and edge cases after. Make it a habit, not an exception.

Level 3 looks like: You verify AI output against sources as a matter of habit, catch confident errors before they ship, and keep a checking standard others can follow.

What it is: Structuring content so AI assistants and answer engines can find, parse and cite it, the successor discipline to classic SEO.

Why now: As buyers ask AI assistants instead of search engines, brands need content machines can find, parse and cite accurately. 

Build it: Learn schema markup and entity based SEO, then test how AI assistants currently describe your organisation.

Level 3 looks like: You can audit a page for machine readability, apply schema and answer first structure correctly, and show that AI assistants cite your content.

What it is: Combining rules based automation with AI so processes that needed human judgement at every step can run with judgement only where it matters.

Why now: Process optimisation features on LinkedIn's 2026 list, and pairing RPA with AI now automates work that scripts alone never could. 

Build it: Map one repetitive process end to end, then automate it with Power Automate, UiPath Community Edition or Make.

Level 3 looks like: You have automated at least one real end to end process combining rules and AI, documented it, and reduced its manual effort measurably.

What it is: Reading, questioning and drawing sound conclusions from charts, tables and dashboards, including knowing what they hide.

Why now: Every function now produces dashboards, and misreading them is more dangerous than ignoring them. 

Build it: Take a free data literacy short course on FutureLearn or OpenLearn, then narrate one real chart aloud each week: what it says, and what it hides.

Level 3 looks like: You interpret standard business dashboards correctly without help, question dubious axes, samples and averages, and brief others on what the data does and does not say.

What it is: Building robust, auditable spreadsheets using tables, modern functions and structured references rather than fragile manual workings.

Why now: Still the working language of business data, and modern functions like XLOOKUP, LET and dynamic arrays have quietly made it far more powerful. 

Build it: Microsoft Learn's Excel paths are free. Rebuild one manual report with tables, named ranges and zero hard coded numbers.

Level 3 looks like: You build models with tables, named ranges and functions like XLOOKUP and LET that others can audit, with no hard coded numbers hiding in formulas.

What it is: Writing queries that retrieve, join and aggregate data directly from databases, without waiting on an analyst.

Why now: The fastest route from question to answer in most organisations is still a well written query, with no analyst queue required. 

Build it: Work through the free SQL course on Kaggle Learn, then ask for read access to one real database at work and use it weekly.

Level 3 looks like: You write your own multi table joins and aggregations on live data, sanity check the results, and answer routine data questions the same day without support.

What it is: Turning data into clear visual displays in tools such as Power BI, Tableau or Looker that drive decisions rather than decorate them.

Why now: Decision makers act on what they can see. Power BI, Tableau and Looker skills convert analysis into action. 

Build it: Follow Microsoft Learn's free PL-300 path or Tableau's free training videos, then replace one recurring slide deck with a live dashboard.

Level 3 looks like: You build and maintain a live dashboard others rely on, choose chart types that fit the question, and your visuals need no verbal explanation to land.

What it is: Applying core statistical ideas such as variance, significance, sampling and base rates to judge what numbers really show.

Why now: Averages, variance, significance and base rates are the immune system against being fooled by data, including AI generated data. 

Build it: Khan Academy's statistics course is free and excellent. Pair it with The Art of Statistics by David Spiegelhalter.

Level 3 looks like: You routinely ask about sample size, baseline and significance before acting on a number, and you can spot a misleading average or cherry picked range unaided.

What it is: Structuring findings into a narrative with context, tension and a clear recommended decision, so insight turns into action.

Why now: Insight changes nothing until someone acts on it. Structure, sequencing and narrative move a finding into a decision. 

Build it: Read Storytelling with Data by Cole Nussbaumer Knaflic, then rework one past analysis into a three slide story: context, conflict, decision.

Level 3 looks like: Your analyses end with a decision, not a chart dump. You can take a finding to a mixed audience and have them leave knowing what to do next.

What it is: The basics of how data moves: pipelines, transformations, scheduling and storage that keep reports and models fed with clean inputs.

Why now: Clean, reliable pipelines are the unglamorous foundation under every dashboard and every AI initiative. 

Build it: Learn the basics of ETL and scheduling with the free dbt Fundamentals course, then document one data flow your team depends on.

Level 3 looks like: You can document and maintain a simple data flow end to end, diagnose why a feed broke, and explain the pipeline behind your team's main report.

What it is: Defining owners, definitions, standards and checks so the data everyone relies on stays accurate, consistent and trusted.

Why now: Named on LinkedIn's 2026 list, because AI amplifies whatever you feed it, including the errors. 

Build it: Define owners, definitions and quality checks for your five most used data fields, and write them down somewhere the whole team works.

Level 3 looks like: The fields you own have named definitions, owners and quality checks written down, and colleagues come to you to settle what a metric actually means.

What it is: Using a scripting language to clean, analyse and automate data work beyond what spreadsheets can comfortably handle.

Why now: When spreadsheets run out of road, a little code goes a long way, and AI assistants have flattened the learning curve dramatically. 

Build it: Start with the free Python for Everybody course, then automate one weekly spreadsheet job with AI assisted code.

Level 3 looks like: You write working scripts that automate a recurring data task, debug them yourself with or without AI help, and others can rerun your code from your notes.

What it is: Designing controlled tests with hypotheses, sample sizes and success measures agreed before the result is known.

Why now: Opinions are cheap. Controlled tests settle arguments and compound learning in marketing, product and operations alike. 

Build it: Run one honest A/B test on something you own, a subject line, a page, a process, and write up the result either way.

Level 3 looks like: You design tests with the hypothesis and success measure fixed up front, run them cleanly, and write up the result honestly whichever way it lands.

What it is: The everyday behaviours and basic concepts that protect accounts, devices and data: authentication, phishing awareness, updates and least privilege.

Why now: Phishing, ransomware and supply chain attacks now reach every team, so baseline security behaviour is everyone's job, not just IT's. 

Build it: Work through the NCSC's free staff training resources, then switch on a password manager and multi factor authentication everywhere.

Level 3 looks like: You use a password manager and multi factor authentication everywhere, spot and report phishing attempts reliably, and never need chasing on security basics.

What it is: Working knowledge of the major cloud providers, their core services, and the cost and architecture choices they involve.

Why now: Most new systems are born in AWS, Azure or Google Cloud, and cloud cost and architecture choices shape what teams can build. 

Build it: Each provider runs free fundamentals training and an entry certification: AWS Cloud Practitioner, Azure Fundamentals or Google Cloud Digital Leader.

Level 3 looks like: You hold or could pass an entry cloud certification, navigate your organisation's cloud console confidently, and can explain a workload's cost drivers.

What it is: Building and shipping software using current languages, frameworks and engineering practice, from version control to testing.

Why now: Software remains the highest leverage way to scale a process, and Coursera's 2026 report puts software and product development among its three in demand career areas. 

Build it: freeCodeCamp and The Odin Project are free and project based. Ship one small real tool, not ten tutorials.

Level 3 looks like: You ship small working tools end to end: coded, tested, versioned and documented well enough that another developer could pick them up cold.

What it is: Using AI pair programmers to write code faster while keeping human review, testing and judgement firmly in charge.

Why now: Tools like GitHub Copilot, Cursor and Claude have changed how code gets written. The skill is steering, reviewing and testing what they produce. 

Build it: Use an AI pair programmer on a real project for a fortnight, and review every suggestion before accepting it. The discipline is the skill.

Level 3 looks like: You use an AI assistant on real projects daily, review every suggestion before accepting, and your merged code holds the same quality bar as hand written work.

What it is: Automating how software is built, tested and released, using pipelines, containers and infrastructure as code.

Why now: Shipping safely and often is a competitive weapon, and automating build, test and deploy is what makes it routine. 

Build it: Learn Git, CI/CD and containers through the free GitHub Skills courses and Docker's getting started guide, then automate one deployment.

Level 3 looks like: You maintain a CI/CD pipeline that deploys without drama, can containerise an application, and recover a failed release calmly using the runbook.

What it is: Connecting systems through their interfaces so data flows automatically instead of through copy and paste.

Why now: Modern stacks are assembled, not built from scratch, and the people who connect systems quietly become indispensable. 

Build it: Learn REST basics with Postman's free academy, then connect two tools your team already pays for and remove one manual handoff.

Level 3 looks like: You can read API documentation cold, build a working integration between two real systems, and have removed at least one manual handoff this year.

What it is: Assembling working internal tools, apps and automations on platforms such as Power Apps or Airtable without conventional coding.

Why now: Power Apps, Airtable and similar platforms let operations people ship internal tools in days, easing the developer bottleneck. 

Build it: Rebuild one spreadsheet driven process as a simple app with Power Apps or Airtable, and add one automation to it.

Level 3 looks like: You have replaced a spreadsheet driven process with a working app that others use daily, including at least one automation, and you maintain it yourself.

What it is: Using Git based workflows of branches, commits and reviews to change shared work safely and traceably.

Why now: Git underpins how modern teams change anything safely, increasingly including documents, data and infrastructure, not just code. 

Build it: GitHub's free Skills courses cover branching, pull requests and review. Practise by versioning something you actually maintain.

Level 3 looks like: You branch, commit, raise and review pull requests correctly without guidance, and your commit history tells a story a stranger could follow.

What it is: Applying UK GDPR principles in daily work: lawful basis, minimisation, retention and individual rights.

Why now: UK GDPR obligations and customer expectations keep rising, and privacy by design is far cheaper than retrofitting it after a breach. 

Build it: Study the ICO's free guidance and checklists, then run a data audit on one process: what you hold, why, and for how long.

Level 3 looks like: You can run a basic data audit on a process you own, answer what is held, why and for how long, and you spot a privacy risk before anyone has to point it out.

What it is: How devices, networks and servers connect and fail: IP, DNS, wifi, VPNs and the basics of keeping things up.

Why now: Hybrid work, cloud and connected devices all stand on networking, which is half of the WEF's second fastest rising skill area. 

Build it: Cisco's free Networking Essentials or CompTIA Network+ materials cover the foundations. Then map your own team's setup and where it could fail.

Level 3 looks like: You can map your team's network setup, diagnose common connectivity faults methodically, and explain where the single points of failure sit.

What it is: Planning, sequencing and steering work to deliver agreed outcomes on time, with risks and owners visible throughout.

Why now: Still the backbone skill for turning intent into delivery, and named in LinkedIn's 2026 UK leadership and delivery skill areas. 

Build it: Google's Project Management Certificate or APM's resources cover the craft. Then run one real project with a one page plan, owners and dates.

Level 3 looks like: Your projects run from a written plan with owners and dates, status is visible without being asked for, and slippage gets flagged early rather than explained late.

What it is: Deciding what to build and why, by joining customer insight, commercial judgement and delivery prioritisation.

Why now: Organisations increasingly organise around products, and the role blends customer insight, prioritisation and commercial judgement. 

Build it: Read Inspired by Marty Cagan, then write a one page brief for something you own: problem, users, success measure.

Level 3 looks like: You can write a one page brief covering problem, users and success measure, prioritise a backlog defensibly, and say no with reasons that hold.

What it is: Translating between business need and technical delivery: mapping processes, gathering requirements and defining what good looks like.

Why now: Someone has to translate between what the business wants and what gets built, and AI projects have made that translation more valuable. 

Build it: BCS and IIBA publish entry level pathways. Practise by mapping one process as is and to be before any tool is chosen.

Level 3 looks like: You produce as is and to be process maps colleagues recognise as accurate, and your requirements rarely bounce back as misunderstood.

What it is: Reading and building budgets, forecasts and the core statements well enough to argue with the numbers, not just receive them.

Why now: Financial operations and reporting is one of LinkedIn's eight fastest rising categories for 2026. Numbers fluency earns a seat at decisions. 

Build it: Take a free finance for non finance managers course on OpenLearn, then rebuild your team's budget so you can defend every line.

Level 3 looks like: You build and defend your own budget line by line, read a profit and loss statement without help, and can explain variance to finance in their language.

What it is: Making products and content findable and persuasive across search engines, social platforms and AI assistants.

Why now: Discovery is splitting across search engines, social platforms and AI assistants, and being findable now spans all three. 

Build it: Google Skillshop and HubSpot Academy are free. Pick one page or product and improve its visibility measurably over a quarter.

Level 3 looks like: You can run a basic keyword and intent audit, improve a page's visibility measurably over a quarter, and report the result in traffic and conversions, not activity.

What it is: Producing written, visual and video content that holds attention and serves a measurable purpose.

Why now: Video and short form content now do the heavy lifting in marketing, sales and internal communication alike. 

Build it: Learn one editing tool properly, CapCut or DaVinci Resolve are free, and publish weekly for a quarter. Volume teaches faster than theory.

Level 3 looks like: You publish consistently in at least one format, edit your own video or graphics competently, and can point to content that measurably moved a number.

What it is: Running a disciplined pipeline: qualifying, advancing and closing opportunities with clean CRM data behind every stage.

Why now: Account development and go to market strategy sit inside LinkedIn's 2026 revenue growth category. Disciplined pipeline work is a skill, not a personality. 

Build it: Salesforce Trailhead and HubSpot Academy offer free structured paths. Then keep one pipeline scrupulously clean for ninety days.

Level 3 looks like: Your pipeline is current and honest, next steps exist on every open deal, and your forecast is taken at face value because it keeps coming true.

What it is: Mapping and improving the end to end journey customers actually experience, not the one the org chart imagines.

Why now: Service orientation sits in the WEF's core skills top ten, and journey level thinking separates good products from loved ones. 

Build it: Map one customer journey end to end, find the single worst moment, and fix it. Repeat quarterly.

Level 3 looks like: You can map a journey end to end with its moments of truth, identify the worst friction point from evidence, and have fixed at least one this year.

What it is: Running online retail well: storefronts, merchandising, conversion, fulfilment and returns as one connected system.

Why now: Online retail keeps absorbing share, and running storefronts, logistics and conversion well is a transferable operations craft. 

Build it: Shopify's free learning paths cover the full stack. Then own one metric, conversion, returns or delivery time, and move it.

Level 3 looks like: You own at least one trading metric such as conversion or returns, know its drivers, and have moved it in the right direction with deliberate changes.

What it is: Using spend, supplier and logistics data to buy better and keep goods and services flowing under pressure.

Why now: Supply shocks made resilience a board topic, and LinkedIn's 2026 list includes logistics coordination and process optimisation. 

Build it: CIPS publishes entry resources, and supply chain courses are free on OpenLearn. Map your top five supplier risks.

Level 3 looks like: You can produce a spend and risk view of your top suppliers, run a structured negotiation or tender, and know which supplier failure would hurt first.

What it is: Measuring and reporting environmental and social performance against recognised frameworks and tightening disclosure rules.

Why now: Environmental stewardship is in the WEF's top ten rising skills, and UK disclosure expectations keep tightening for suppliers as well as listed firms. 

Build it: IEMA's foundation courses and the GHG Protocol's free guidance are the standard starting points. Draft a simple ESG snapshot for your team.

Level 3 looks like: You can assemble a credible ESG snapshot for your area using recognised definitions, with every figure traceable to a source and nothing you could not defend.

What it is: Understanding emissions scopes, baselines and reduction levers well enough to plan and track credible cuts.

Why now: Net zero commitments now flow down supply chains as contract clauses, so carbon numeracy is becoming a commercial skill. 

Build it: The Carbon Literacy Project runs accredited day courses. Then baseline one activity you control and find its biggest lever.

Level 3 looks like: You can baseline the emissions of an activity you control using standard factors, name its biggest reduction lever, and track progress against it.

What it is: Designing products, packaging and processes to keep materials in use: reduce, repair, reuse and recover.

Why now: Waste is unpriced inefficiency, and circular design thinking turns disposal costs into recovered value. 

Build it: The Ellen MacArthur Foundation publishes free courses and case studies. Redesign one product, package or process to lose less.

Level 3 looks like: You have redesigned at least one product, package or process to lose less, with the saving quantified in material, cost or carbon.

What it is: Identifying what could go wrong, judging likelihood and impact, and preparing responses before they are needed.

Why now: Resilience, flexibility and agility rank near the top of the WEF's rising skills, and risk thinking is the organisational counterpart. 

Build it: Learn the ISO 31000 vocabulary, then write a one page risk register for your team: top five risks, owners, early warnings.

Level 3 looks like: Your area has a live one page risk register with owners and early warnings, reviewed on a rhythm, and your last incident followed the plan rather than improvisation.

What it is: Knowing the rules that govern your work and keeping evidence that you follow them, ready for inspection at any time.

Why now: Policy compliance and safety monitoring appear on LinkedIn's 2026 rising list as AI regulation, data law and sector rules all thicken. 

Build it: Pick the regulation that most affects your work and read the regulator's own guidance, not a summary of it. Keep a simple evidence file.

Level 3 looks like: You can pass a spot check on your area without a scramble: obligations mapped, evidence filed, and gaps already known and scheduled.

What it is: Meeting the duty of care for physical and psychological safety at work, from risk assessments to early support.

Why now: Wellbeing has moved from poster to duty of care, and hybrid work created risks that older policies never imagined. 

Build it: IOSH Managing Safely or a NEBOSH award covers the fundamentals, and MHFA England trains mental health first aiders.

Level 3 looks like: Risk assessments for your area are current and acted on, near misses get reported and learned from, and people raise wellbeing concerns to you early.

What it is: Reducing waste, variation and rework through structured improvement methods such as Lean and Six Sigma.

Why now: Operational efficiency is one of LinkedIn's eight fastest rising categories for 2026, and Lean thinking is its most proven toolkit. 

Build it: A free Lean or Six Sigma yellow belt introduction plus one real improvement project beats any amount of theory. Map a process, remove a step.

Level 3 looks like: You can map a process, measure its baseline, remove a step or a defect cause, and prove the improvement held three months later.

What it is: Using data on skills, performance and attrition to plan hiring, development and succession ahead of need.

Why now: Talent management sits in the WEF's rising top ten, and skills data is becoming the planning currency for hiring and development. 

Build it: CIPD's people analytics resources are a solid base. Start by analysing one dataset you already have: your team's skills matrix.

Level 3 looks like: You maintain a current skills matrix for your team, can show where the gaps and single points of failure sit, and your hiring case is built on that evidence.

What it is: Capturing expertise and structuring it so others learn it quickly: guides, checklists, courses and searchable knowledge.

Why now: With 39 percent of key skills changing by 2030, teams that capture and transfer knowledge quickly compound the advantage. 

Build it: Read Make It Stick, then turn one expert's head knowledge into a checklist, guide or short video the whole team can use.

Level 3 looks like: You have turned at least one expert's know how into a guide or checklist others actually use, and new starters reach competence visibly faster for it.

What it is: Writing instructions, references and process documents that people can follow correctly first time without asking.

Why now: Clear documentation lets processes survive staff changes, audits and AI ingestion alike. It is quiet infrastructure. 

Build it: Google's free Technical Writing courses are excellent. Then rewrite one confusing internal document and measure the questions it stops.

Level 3 looks like: Your documents survive contact with new readers: people follow them unaided, and the question volume on what you have documented measurably drops.