--- "title": "What AI skills should analysts have in 2026?" "description": "The practical AI skills analysts at a firm need in 2026, from giving AI the right context and checking its work to analyzing data and knowing when not to use it, plus how to build those skills across a team." "section": "AI Training" "sidebarTitle": "AI skills for analysts" "published": "2026-10-23T13:08:59.594Z" "updated": "2026-10-23T13:08:59.594Z" --- ## What AI skills should analysts have in 2026? It depends on the work your analysts do. If their days are mostly reading, summarizing and drafting, the core skills are giving AI good context, structuring requests and checking outputs against sources. If they work heavily with numbers and models, they also need to use AI for data analysis and spreadsheet work while verifying every figure. In every case, the most valuable skill is judgment: knowing which tasks to hand to AI, which to keep, and how to spot a confident answer that is wrong. ## Who this is for Heads of people, team leads and managers who hire, train or supervise analysts and associates, and analysts who want a clear picture of which AI skills are worth building. ## The short version - The basics are context, clear requests and careful review, not clever prompt tricks. - Analysts should be able to use AI on documents, data and drafts, and verify all three. - Judgment about when not to use AI matters as much as speed. ## Why analysts are the group to get right Analysts and associates do much of the work that general AI assistants handle well today: reading long documents, pulling out key points, building first drafts, cleaning up data and preparing materials for senior people. That makes them the group where good habits pay off fastest, and where bad habits spread fastest too. If a junior person learns to paste in a question and forward the answer, the firm inherits that habit for years. The goal is not to make analysts AI specialists. It is to make them better analysts who use AI well. ## The core skills ### 1. Giving AI the right context Most weak AI output comes from weak input. Analysts should know how to give the assistant the background it needs: who the work is for, what the deadline and standard are, and which documents matter. They should attach or reference the actual source material instead of asking the AI to rely on general knowledge, as long as your data rules allow it. Our guide to [creating prompting standards](/docs/create-prompting-standards) covers a simple structure teams can share. ### 2. Breaking work into steps Asking for a finished memo in one request rarely works well. Strong users break the job down: summarize the documents, list the open questions, propose an outline, then draft section by section. Each step is easier to check, and mistakes get caught earlier. ### 3. Working with documents Analysts should be comfortable using AI to summarize long documents, compare versions, extract specific terms or figures into a table and find where something is mentioned. They should also know the limits. Long documents can be summarized unevenly, and an assistant may miss a detail buried in an appendix. Asking the AI to cite the page or section it used makes checking faster. ### 4. Working with data and spreadsheets Many business AI assistants can analyze uploaded spreadsheets, write formulas and produce charts, as of October 2026. Features vary by product and plan, so check what your tools include. Analysts should be able to ask for an analysis, read how the AI did it and recheck key numbers by hand or in the spreadsheet itself. A useful rule: any number that goes in front of a client or decision maker is verified by a person against the source. ### 5. Checking and verifying output This is the most important skill and the one most often skipped. Analysts should habitually: - Compare claims and figures against the source documents. - Ask the AI to list its assumptions and anything it was unsure about. - Watch for invented details, such as names, dates or citations that do not exist. - Be extra careful with anything legal, financial or client-facing. ### 6. Knowing when not to use AI Some tasks are faster done directly. Some involve data that should not go into an AI tool at all. Some need a judgment call the analyst should own. Good analysts can explain why they used AI on a task, or why they did not. This connects directly to your [AI acceptable use policy](/docs/write-an-ai-acceptable-use-policy). ### 7. Writing and editing with AI Analysts should use AI to get past a blank page and tighten drafts, then edit the result into the firm's voice. A draft that reads like generic AI text reflects poorly on the analyst and the firm. ## How to build these skills across a team 1. **Train on real work.** Use the documents and tasks your analysts handle every week, not generic examples. See [how to train non-technical employees to use AI](/docs/train-non-technical-teams-on-ai). 2. **Set clear review rules.** Make it explicit which outputs need checking and by whom. 3. **Have managers review the process, not just the result.** Ask analysts how they used AI on a piece of work. It surfaces both good habits and risky ones. 4. **Give it time.** Skills build with practice over weeks, not in one session. See [how long it takes to train a team on AI](/docs/how-long-to-train-a-team-on-ai). 5. **Measure it.** Track whether people use AI on the agreed tasks and whether quality holds. Our guide on [measuring AI training results](/docs/measure-ai-training-results) explains how. ## Common mistakes - **Assuming younger staff already know how.** Familiarity with chat tools is not the same as using them well for professional work. - **Banning AI for junior people.** They will use it anyway, without guidance. Teach safe, effective use instead. - **Rewarding speed alone.** Faster work with more errors is a step backward. ## When you do not need us If your team already has a strong AI user who can run regular sessions on real work, and your review rules are clear, you can build these skills in-house with the steps above. Outside help is more useful when you want a structured program across several teams, or role-specific training tied to your firm's actual workflows. ## How Ruby helps - **AI training:** hands-on sessions for analysts and associates using your firm's real documents, data and deliverables. - **AI strategy:** setting review rules, usage policies and simple measures so skills turn into reliable results. - **AI engineering:** building assistants and agents that handle repeat analyst tasks and connect to the systems your team already uses. If you want your analysts using AI well rather than just often, [book a call with Ruby](https://rubyadvisory.com/book-a-call). ## Related pages - [How to create prompting standards for your company](/docs/create-prompting-standards) - [How to train non-technical employees to use AI](/docs/train-non-technical-teams-on-ai) - [What should an AI training program for a small firm include?](/docs/ai-training-program-for-a-small-firm) - [How to measure whether AI training worked](/docs/measure-ai-training-results)