Extract a powerful, clear visual from spreadsheet chaos.

What the AI will do

• Inspect the workbook and related records to establish what the numbers mean and what the audience needs.
• Create alternative charts and check the chosen visual for clarity and numerical accuracy.

What you can provide

• The workbook or another readable version of the data, even if it is untidy.
• Available definitions, related company data, emails, chats, project documents, and meeting context.

Where you'll take part

• Clarify consequential definitions, confirm the evidence-supported takeaway, and choose a chart to refine.

What you'll get

You have a usable chart whose message, values, scales, and labels have been checked against the source—or an explicit partial result if necessary data or rendering access is missing.

Activity context

Find the comparison that makes the numbers matter.

For the user

Share the spreadsheet in its current state. The AI will investigate what its fields and figures mean, help find a worthwhile point, and create visual alternatives. You will settle important ambiguities and choose the one that communicates best.

Guidance for the AI

These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.

Start with the activity and the user's context

Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.

Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.

Check the capabilities needed for this work

Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.

When useful access is missing:

• Explain what you cannot reach and how that material would help.
• Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
• If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
• Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.

Ask focused questions and respect the answers

Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.

Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.

Keep evidence and independent contributions distinct

Distinguish evidence, inference, and missing information. Check your work against the source material.

Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:

• Give each agent a bounded brief and the context appropriate to its role.
• Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
• Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.

Produce, check, and revise the result

Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.

Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.

Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.

Your AI prompt

Help me extract a powerful, clear visual from spreadsheet chaos. Find a meaningful point in the data and create a truthful chart for the audience who needs it.

1. Understand the numbers and their use

Inspect the workbook with available tools, including relevant sheets, formulas, filters, hidden rows, totals, and notes. Search related emails, chats, data dictionaries, project documents, underlying company datasets, and calendar invitations or meeting records to establish definitions, reporting periods, audience, and decisions the chart could inform. Use source references to distinguish actuals, forecasts, targets, and preliminary figures.

Build a traceable working account of:

• Measures, units, dates, denominators, and the level each row represents.
• Missing values, duplicates, transformations, and totals that need reconciliation.
• Comparisons the data supports and the questions the audience has actually expressed.

Try relevant access before asking me to gather material. Offer help obtaining a usable export if necessary. Ask about unresolved definitions or conflicting versions before aggregating on those assumptions. If the data is only an image, verify uncertain cells with me; a missing number is not zero.

2. Find and confirm the story the evidence supports

Examine appropriate distributions, trends, differences, and concentrations rather than immediately choosing a chart type. Check potentially striking findings against denominators, reporting changes, outliers, missingness, and the relevant baseline. Trace calculations back to source cells or rows. Do not interpret a correlation as a causal finding.

Propose a concise takeaway suited to the audience. If two worthwhile points compete, explain their difference and ask which purpose matters here. If my preferred story conflicts with the figures, show the conflict and help identify an honest point. Confirm the consequential interpretation before constructing visuals around it.

Divide the work without losing the evidence

Use subagents where available. An analysis agent first identifies the core point and supporting evidence; after I confirm the consequential interpretation, brief separate visual agents to explore contrasting forms. Give each the same verified source material or a traceable subset it can actually inspect, definitions, audience evidence, agreed point, viewing format, and its particular design direction. Ask for an actual rendered candidate, not just a design description. Let their first designs finish independently before comparing them.

Assign a fresh reviewer the rendered candidates and evidence-supported audience background, initially without the designers' explanation or intended takeaway. Ask what each image communicates, what they notice first, and what remains confusing. Then provide the evidence and agreed point for a separate fidelity check. The lead reconciles findings, fixes material problems, and owns the final artifact.

Keep source reading and production within each agent's capacity. Split large inputs into named portions, keep source and coverage records, and use waves if concurrency is limited. Without subagents, use separate design passes and a fresh-session review with a compact evidence handoff where possible; accurately identify review that shares the drafting context. Preserve the actual design and checking work in the fallback.

Visual directions to draw on

These practical directions adapt existing visual work; choose forms for this evidence and audience, rather than imitating a designer or using every approach.

• Quiet comparison: aligned small charts, direct labels, and meaningful baselines can reveal the exception. Edward Tufte's sparkline examples.
• Change unfolding: trajectories or successive frames can make an important transition visible. Gapminder Tools guide.
• Human detail: show individual cases and supplied qualitative context within a larger pattern. Giorgia Lupi's Data Humanism.
• Meaningful scale: comparable quantities shown together can make an abstract difference comprehensible. Information is Beautiful's Billion Dollar-o-Gram.
• A bold visual argument: purposeful geometry, contrast, and direct labels can carry a consequential comparison. Du Bois and the Atlanta University team's original charts.
• Relationships revealed: ordered networks, flows, or hierarchies can expose actual dependencies and hubs. Nadieh Bremer's alis_ OS.
• A meaningful metaphor: an unexpected form can encode real attributes consistently, rather than merely decorate a page. Shirley Wu's account of visual flowers and trees.

The descriptions above are enough to begin if a reference page cannot be opened. Inspect original examples when accessible and useful. Use only relationships, quantities, and comparisons supported by this project's evidence.

3. Render and test contrasting charts

Create two or three genuinely different presentations of the agreed evidence, not cosmetic variants. Choose the simplest readable encodings that convey the point. A quiet aligned comparison may work better than an elaborate image; an unusual form is valuable when it reveals something the ordinary chart hides. Explain necessary units, periods, and uncertainty in concise labels or notes.

Check every candidate's calculations and encoded values against the working data. Check axis ranges and baselines, area or size proportionality, labels, categories, and comparability of periods. If aggregation hides a material exception, expose it or qualify the takeaway. Have the fresh reviewer first read the chart itself, then verify what it implies against the data. Correct errors before asking me to choose.

Show the actual rendered candidates. Ask which makes the point clearest, refine it, and inspect the revised output at the size it will be used. Deliver a usable chart file or artifact and the supporting data or calculations needed to reproduce it. When tools cannot render or required figures remain unavailable, investigate an alternative and identify the partial result honestly. Stop with the checked chart rather than changing the underlying business records.

Done when

You have a usable chart whose message, values, scales, and labels have been checked against the source—or an explicit partial result if necessary data or rendering access is missing.

These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.