AI for Data & analytics teams
Claude for data analysis: Claude for Excel, text-to-SQL, notebooks with Claude Code, checking AI analyses, and the analytics tools that hold up.
Start here
Read these in order — each one assumes the last.
- 01GuideClaude for Data Analysis: The Complete 2026 GuideClaude for data analysis in 2026: chat, Excel, Claude Code, or Cowork, what the code execution sandbox is, its limits, failure modes, and how to verify results.
- 02GuideClaude for Excel: What It Does, Where It Fails, How to Use ItHow Claude for Excel works in 2026: install, supported builds, plan gating, cell-level citations, model debugging, what it cannot do, and a review checklist.
- 03GuideClaude Code for Data Analysts: Notebooks, SQL, and CSV Work Without a Data EngineerSet up Claude Code for analysis: an analysis repo with CLAUDE.md, CSV profiling, a pandas loop, read-only warehouse queries, and deny rules that hold.
- 04GuideThe Best AI Tools for Data Analysts in 202615 AI tools for data analysts in 2026: assistants with code execution, spreadsheet add-ins, notebooks, text-to-SQL, and terminal agents, with a verdict each.
- 05GuideWhich Claude Plan (and Model) Should a Data Analyst Pay For?Free, Pro, Max, or Team for data analysis: which plan unlocks Excel, Claude Code, and Cowork, how usage limits bite, and which model fits which analysis task.
Guides34
Tutorials and deep-dives
What Are Claude Skills? The Complete Guide
Claude Skills explained: what a SKILL.md is, how progressive disclosure keeps skills cheap, where they run, and how to install or write your own.
LLM API Pricing in 2026: Every Major Model Compared
Per-million-token prices for Claude, GPT, Gemini, DeepSeek, Mistral, and Grok — plus caching and batch discounts — verified against vendor pricing pages.
How RAG Actually Works: Ingestion, Chunking, Retrieval & Reranking
A clear, practical walkthrough of the retrieval-augmented generation pipeline — what each stage does, where it fails, and how the pieces fit together.
AI Literature Review Workflow: Screen, Extract, and Verify
Use AI for a literature review with explicit eligibility criteria, study-level evidence tables, source locations, and checks for missing or conflicting data.
AI Meeting Notes: Turn Transcripts into Decisions and Action Items
Use AI to turn a meeting transcript into clear minutes, with evidence for decisions, confirmed owners, unresolved questions, and a reviewable action list.
Analyze Customer Feedback with AI: Themes You Can Check
Analyze customer feedback with AI using a source register, a reviewed codebook, distinct customer counts, and examples that preserve context.
ChatGPT Projects vs Claude Projects for Teams
Compare shared AI project context in ChatGPT and Claude, then build a team brief with clear instructions, source ownership, and a review process.
Create an AI Presentation from a Report: A Reviewable Workflow
Turn a business report into an AI-assisted presentation with a claim-to-slide map, source checks, speaker notes, and a finished-deck review.
Create an AI Project Handoff Another Teammate Can Use
Create a project handoff with AI that preserves source files, current decisions, open questions, owners, and an acceptance check for the next teammate.
How to Research Documents with AI and Verify Citations
Turn a set of documents into a defensible AI research brief with a source register, conflict table, citation checks, and clear unanswered questions.
How to Summarize an Email Thread with AI and Draft a Reply
Use AI to reconstruct a long email thread, distinguish current decisions from old requests, identify open questions, and draft a reply for review.
Microsoft 365 Copilot vs ChatGPT for Work
Compare Microsoft 365 Copilot and ChatGPT for workplace tasks by source access, app context, deliverables, and a practical team pilot.
Adjudicate AI-Assisted Text Labels with an Evidence Ledger
Resolve disputed text labels by preserving each response, source excerpt and task-rule version rather than accepting the model or majority automatically.
Build a Source-Grounded AI Mind Map
Turn a bounded document set into a reviewed concept hierarchy with evidence for each branch and visible unresolved relationships.
Link Records Across Datasets with AI-Assisted Review
Build a cross-source entity crosswalk that preserves both record keys, unresolved candidates and cardinality decisions without silently joining or merging data.
The Best Document Parsers for RAG in 2026: LlamaParse vs Docling vs Unstructured vs Reducto
LlamaParse, Docling, Unstructured, Reducto, Marker and Mistral OCR compared on deployment, license, OCR, tables and output, with a verdict for each.
Anthropic's Data Plugin for Claude: Every Skill Explained
Every skill in Anthropic's open-source data plugin for Claude Cowork and Claude Code, the warehouse connectors it expects, the install commands, and its gaps.
ChatGPT vs Claude for Data Analysis
ChatGPT vs Claude for data analysis, compared on sandbox behavior, file limits, Excel add-ins, chart output, citations, warehouse access, and verification.
Claude Skills for Data Analysts: 5 to Upload Today
Five portable skills that make Claude behave like a careful analyst: first look, chart choice, memo writing, SQL explanation, and formula auditing.
How to Check an AI Data Analysis Before You Trust It
Eight ways an AI-generated analysis goes wrong, the specific check that catches each one, and a copyable checklist to run before a number ships.
Plan AI Survey Branching Before Building the Form
Turn approved survey questions and routing rules into an explicit branch table, with stable IDs and unresolved paths before an AI form builder runs.
Text-to-SQL with Claude: A Safe Read-Only Postgres Setup
Set up text-to-SQL with Claude on Postgres safely: a read-only role, an MCP server in read-only mode, schema and metric context, and a two-way check.
The Best Text-to-SQL Tools in 2026
Text-to-SQL tools compared on how they ground the model in your schema, what accuracy really means, read-only safety, deployment, licensing, and pricing model.
Draft AI Chart Long Descriptions from Verified Data
Review a chart-description draft against the frozen source table, actual visual observations and its implemented page placement.
Create Standard Operating Procedures with AI from a Real Workflow
Create an SOP with AI from captured steps, explicit prerequisites, exception paths, and a teammate walkthrough that checks the procedure.
Test AI Survey Logic with Known Response Paths
Create complete response-path cases for an AI-built survey, check a local route model, then preview the actual form before collection.
Review AI Duplicate-Record Candidates Before Merging
Separate duplicate-record candidate generation from human match decisions, retaining source keys and field evidence before any merge.
Review OCR Text Against the Original Scan
Review OCR text beside source scans, preserve exact wording and document uncertain characters in a correction ledger before producing a corrected copy.
Prelabel a Text Dataset with AI Without Replacing Review
Prepare a small supervised text-labeling batch with separate model suggestions and human responses, using stable record IDs and an approved task specification.
Claude's Document Skills: Excel, PowerPoint, Word, and PDF
How Anthropic's pre-built document skills let Claude produce real .xlsx, .pptx, .docx, and PDF files — on claude.ai, the API, and in Claude Code.
Tools26
The AI tooling directory
Dovetail
Dovetail organizes customer research and feedback with AI transcription, project summaries, searchable evidence, and channels for recurring signals.
Elicit
Elicit supports academic research with a Research Agent, research reports, and a structured workflow for screening papers and extracting evidence.
Glean
Glean is an enterprise AI search and assistant platform that retrieves company knowledge through configured connectors and source access controls.
Be My Eyes
Evaluate Be My Eyes for a personal picture question, recording useful clarification, remaining uncertainty and independent evidence for consequential details.
Scribe
Scribe captures on-screen workflows into editable visual guides and adds AI drafting features for titles, descriptions, and process documentation.
Miro
Evaluate Miro as a shared review surface for an approved concept hierarchy and its relationship ledger.
Typeform
Typeform builds online forms with AI-assisted questions, endings and branching rules that owners can preview and revise.
Argilla
Evaluate Argilla for a supervised text task with separately preserved suggestions, human responses and decision evidence.
Tamr
Evaluate Tamr with a bounded mastering pilot that preserves source keys, exposes uncertain candidates and names the owner of each decision.
Mistral OCR
Mistral AI's hosted OCR and document-understanding API: PDFs, Office files and images in, per-page Markdown and structured JSON out, billed per page.
Databricks Genie
Databricks' conversational analytics layer: Genie Agents answer natural-language questions over Unity Catalog data using curated instructions and SQL.
Deepnote
A collaborative data notebook whose AI agent edits and runs your blocks, with data apps, a semantic layer, and integrations across warehouses and BI tools.
Hex
A collaborative notebook and data-app platform whose AI agents write SQL and Python, answer questions in Threads, and run on curated workspace context.
Julius
A chat-first AI data analyst: upload a spreadsheet or connect a warehouse, ask in plain English, and get charts, code, and shareable exports back.
Pandasai
A Python library that adds a chat method to your dataframes: it generates and runs pandas code to answer questions, with an optional Docker sandbox.
Thoughtspot Spotter
ThoughtSpot's agentic analyst: it resolves questions into search tokens against a governed semantic model rather than raw SQL, then acts on the answer.
Vanna
An MIT-licensed Python framework for text-to-SQL: a user-aware agent that learns from successful queries and streams tables, charts, and summaries back.
ABBYY FineReader PDF
ABBYY FineReader PDF applies AI-based OCR to scans and PDFs, with a Windows OCR Editor for checking recognized text against the source image.
Label Studio
Label Studio is an open-source annotation interface that can present imported or model-generated predictions for people to review and correct.
Dedupe
Evaluate Dedupe through an owner-reviewed matching pilot with known pairs, preserved source keys, candidate coverage checks and unresolved cases.
Surveymonkey
SurveyMonkey combines AI survey drafting with editable survey questions and plan-dependent skip logic for respondent paths.
Docling
Open-source Python library that parses PDFs, DOCX, PPTX, HTML, and images into structured Markdown and JSON with layout, tables, and reading order for RAG.
LlamaParse
Hosted document-parsing API from LlamaIndex that turns complex PDFs — tables, charts, figures, handwriting — into clean, LLM-ready Markdown for RAG.
Postgres MCP Pro
The maintained Postgres MCP server — safe SQL execution, EXPLAIN with hypothetical indexes, workload-driven index tuning, and database health checks.
Supabase MCP
Supabase's official MCP server — run SQL and migrations, read logs and advisors, generate types, and deploy Edge Functions, with read-only and project scoping.
LM Studio
A desktop app for discovering, downloading, and running open-weight LLMs locally with a GUI and a local OpenAI-compatible server.
Glossary20
AI terms, defined precisely
Data Lineage
Data lineage records where data comes from, how it changes, and where it is used, helping teams trace an AI output back through its inputs.
Evidence Synthesis
Evidence synthesis brings findings from multiple studies together to answer a research question while retaining differences, limitations, and uncertainty.
Thematic Analysis
Thematic analysis interprets patterns of meaning in qualitative data, such as interview transcripts, open survey answers, and customer feedback.
Record Linkage
Understand the reviewable crosswalk needed before joining sources with unrelated keys, ambiguous candidates and explicit cardinality policies.
Process Capture
Process capture records how a task is actually performed, preserving observed actions and context as input to a reviewed procedure or training guide.
Entity Resolution
Understand the identity question behind similar records, including false merges, missed matches and the source owner’s unresolved decisions.
AI Data Analyst
An AI data analyst is a tool that takes a data question, writes and runs the code or SQL to answer it, and returns a chart or summary you still have to check.
Code Execution (Code Interpreter)
Code execution lets an AI assistant write and run real code in a sandbox, so numbers, files, and charts come from a computation rather than an estimate.
Conversational Analytics
Conversational analytics is asking questions of governed business data in plain language and getting a chart or number back, without opening a dashboard.
Semantic Layer
A semantic layer defines business metrics, dimensions, and joins once, so every query and every AI agent computes the same number the same way.
Text-to-SQL
Text-to-SQL is turning a plain-language question into a SQL query a database can run, using a model grounded in your schema, documentation, and past queries.
Blocking in Record Matching
Understand how a candidate-selection rule can miss known pairs and why missing fields need coverage checks before identity review.
Skip Logic
Skip logic routes a respondent to different survey questions or endings based on answers or other stated conditions.
Optical character recognition (OCR)
Optical character recognition converts text pictured in a scan or image into machine-readable characters; recognition output still needs source checks.
Grounding
Grounding ties a model's output to verifiable sources — retrieved documents, tool results, citations — instead of training-data memory.
Context Window
The context window is the maximum text — measured in tokens — an LLM can consider at once: prompt, conversation, documents, and its own output combined.
Embedding
An embedding is a vector of numbers representing text's meaning, placed so similar texts land close together — the foundation of semantic search and RAG.
Hallucination
A hallucination is fluent, confident output that is factually wrong or fabricated — plausible text unsupported by any source, the signature LLM failure mode.
RAG (Retrieval-Augmented Generation)
RAG retrieves relevant documents from your own data and injects them into an LLM's prompt at query time, grounding answers in facts the model wasn't trained on.
Structured Output
Structured output makes an LLM return data in a guaranteed shape — JSON matching your schema — so code can consume model responses without parsing prose.
Agents11
Specialized subagents for focused work
Entity Match Reviewer
Review an entity-match candidate ledger against supplied records and owner criteria; flag unsupported match decisions, conflicting fields and transitive contradictions without merging or identifying people.
Literature Screening Reviewer
Use this agent to audit a literature screening log against a supplied protocol and local abstracts or paper extracts: trace each eligibility judgment to criteria and source text, flag unsupported exclusions, identify unresolved eligibility, and record protocol drift. Review the log without searching the web or changing inclusion decisions.
Project Handoff Auditor
Use this agent to review a prepared human-to-human project handoff against its local source files: find missing deliverables, unsupported completion claims, unclear ownership, stale or conflicting decisions, and questions the receiving teammate cannot resolve from the packet. Report evidence and fixes without editing the packet.
Survey Design Reviewer
Use this agent to review a draft customer survey before collection: check leading wording, double-barreled questions, missing response options, inconsistent scales, unsupported recall periods, and whether each question can answer the stated research objective. Return an annotated review and proposed rewrites without changing the survey.
Procedure Exception Reviewer
Review a drafted workplace SOP against captured walkthroughs and supplied exception cases: identify absent branches, unmet prerequisites, unclear stop conditions, and unsupported recovery steps without executing or editing the procedure.
Analysis Reviewer
Use this agent to review a finished analysis for methodological errors before it ships — checking grain and double counting, join fan-out, rows silently dropped by filters and inner joins, sampling and truncation, null handling, time zone and date boundaries, numbers in the prose that disagree with the code's output, charts that mislead, and causal language resting on correlational evidence. Examples — 'review this notebook before I send the deck', 'the query and the summary disagree somewhere, find it', 'does this analysis actually support the conclusion it draws?'.
OCR Fidelity Reviewer
Review supplied OCR text and readable source-page images for transcription fidelity. Flag changed characters, reading-order errors and unsupported proposed corrections with page/region evidence; do not interpret document meaning, guess unreadable text, certify authenticity or overwrite the source.
Annotation Evidence Reviewer
Review a supervised text-label adjudication packet for traceable excerpts, preserved independent responses and rule-supported resolutions; flag missing evidence without choosing final labels.
Image Description Reviewer
Review still-image text alternatives against page purpose, verified visual evidence and source data; flag missing meaning, invented detail and chart-description conflicts without claiming conformance.
SQL Pro
Use this agent for SQL itself — correct joins and window functions, indexing, EXPLAIN plans, schema design, and safe migrations on Postgres/MySQL. Examples — making a slow query fast, designing a normalized schema, writing a reversible migration.
Data Scientist
Use this agent for data analysis — exploration, statistics, SQL, and clear findings. Examples — analyzing a dataset, writing an analytical SQL query, summarizing experiment results.
Skills17
Reusable capabilities Claude loads on demand
Decision Log Reconciler
Reconcile an existing business-project decision log with supplied dated notes and change records: connect decisions to evidence, preserve explicit supersession links, separate accepted choices from proposals, and list contradictions awaiting an owner. Use when project records disagree about what was decided; do not write an architectural ADR or create meeting minutes.
Feedback Codebook Designer
Design a versioned coding codebook for recurring customer-feedback classification from a supplied pilot sample: define one code per job with inclusion rules, exclusions, verbatim examples, boundary cases, and an unresolved bucket. Use before labeling support tickets or survey comments, or when a feedback taxonomy has overlapping labels; do not synthesize interviews or rank product priorities.
Research Evidence Extractor
Extract a structured study evidence table from supplied paper text or readable research excerpts using a user-defined column schema: record study and source IDs, methods, population, outcomes, units, time points, source locations, and missing data. Use after papers have been selected and before synthesis; do not search for papers, decide eligibility, compute pooled effects, or draft conclusions.
Survey Path Test Builder
Build a survey-paths.json test packet from approved single-choice question IDs, complete answer-to-destination rules and named endings; preserve missing routes for owner resolution.
Source Concept Map Builder
Build a concept-node and relationship table from supplied document passages for one focal question, preserving source locations and contested edges before an AI mind map is drawn.
Annotation Task Spec Builder
Specify the record and annotation-interface contract for an owner-defined supervised text-labeling task: approved label IDs, input text, model suggestions, human response fields and abstain handling.
Process to SOP
Convert supplied workplace walkthrough notes or readable process captures into one draft SOP with sourced steps, explicit inputs and outputs, observed branches, stop conditions, and an owner review queue. Use after a real process has been captured; do not invent policy or write an incident runbook.
OCR Correction Ledger
Build a proposed OCR correction ledger from a fixed OCR text copy and readable source-page images or human-verified image notes, preserving original spans, Unicode offsets, replacements, locators and unresolved status. Use before a reviewer creates a corrected copy; do not extract business conclusions or alter original scans.
Analysis Memo Writer
Turn findings and figures you supply into a stakeholder writeup with a fixed spine: headline finding, what the number is and what it is not, method in three lines, a mandatory caveats and limitations section, what would change the conclusion, and one recommended next step. Never invents a figure and never upgrades a correlation into a cause. Use when the analysis is done and has to survive being read by someone who was not in the query.
Chart Chooser
Pick the chart for a stated question and data shape and defend the choice: the recommendation, the reasoning, the alternatives rejected and why each fails this question, encoding rules for axis baseline, sorting, color, and labels, and generated code in the plotting library you name. Use when you know what you want the chart to say and need the one form that says it, not a gallery.
Dataset First Look
Turn a pasted CSV header with sample rows, a pasted table, or an attached data file into a fixed plain-language profile: shape, column inventory with inferred types, null and blank patterns, cardinality, suspicious columns, duplicate-key risk, outliers worth a look, and the questions to settle before analyzing. Use when a dataset just landed and you need to know what you are holding before you write a query.
Spreadsheet Formula Auditor
Audit pasted spreadsheet formulas or a described model for the errors that survive review: hardcoded values buried inside formulas, ranges that drift or truncate, one cell in a row that does not match its neighbors, circular references, sign errors, IFERROR masking a real failure, and volatile functions, returned as a fix list ordered by how much money the error moves. Use before a model that carries a real decision leaves your hands.
SQL Explainer
Walk through a pasted SQL query in execution order in plain language: what each step does, the grain of the result, where a join can fan rows out, which filters silently drop rows, and a ranked list of what could be wrong. Comprehension only, not performance tuning. Use when you inherited a query, are reviewing one before trusting its numbers, or have to explain it to someone who does not read SQL.
Entity Match Candidate Ledger
Build a field-evidence ledger for proposed duplicate or cross-source record pairs, preserving source keys, original values and match/different/unresolved owner decisions without merging records.
Annotation Disagreement Ledger
Build an adjudication ledger from conflicting supervised text-label responses, preserving original reviewer labels, exact excerpts, task-rule versions and unresolved owner decisions.
Multimodal Document Extractor
Extract structured data from documents and images with a vision-language model — define the target schema, prompt the VLM to fill it from the page (invoices, forms, receipts, statements, IDs), and verify critical fields against the source. Use when you need reliable structured output from messy, varied, or scanned documents that defeat template-based OCR.
SQL Optimizer
Diagnose a slow SQL query from its execution plan and propose a verified optimization — finding the real bottleneck (sequential scan, missing or unused index, bad join order, app-side N+1) and measuring the fix before and after. Use when a query is slow and you need a fix backed by EXPLAIN ANALYZE, not a guess.
Commands10
Slash commands for Claude Code
Check Evidence Table
Check a study evidence CSV against a source registry with code: validate required columns and cells, row IDs, source references, duplicate study/outcome/time-point keys, and allowed status values, then report structural errors without judging research claims.
Count Feedback
Count a single-label feedback CSV with code: validate record IDs, collapse exact duplicate rows, reject conflicting rows for one ID, and report records, distinct known respondents, missing identities, and label totals without inventing prevalence.
Snapshot Handoff
Create a deterministic inventory of an approved handoff directory: record relative file paths, byte lengths, and SHA-256 hashes with code, list skipped symlinks and unreadable files, and write a snapshot without interpreting project status or changing source files.
Check Procedure Graph
Check explicitly recorded SOP step dependencies with Python: validate IDs and prerequisite references, compute deterministic execution layers for acyclic graphs, and report steps blocked by a cycle without executing the procedure.
Check Entity Match Ledger
Check entity-match-ledger.json for source-linked evidence values, duplicate pairs, reviewer presence and contradictory match groups without merging records.
Define Metric
Write or refine a metric definition — name, plain-language meaning, grain, filters, source tables and columns, edge cases, and owner — into analysis/metrics/<slug>.md, after searching the repo for a definition that already exists.
First Look
Read the first rows and the true row count of a CSV or spreadsheet, profile it against a fixed checklist, and write the result to analysis/profiles/<name>.md so the dataset's condition is on record before anyone queries it.
Check Annotation Batch
Check annotation-batch.json for approved labels, distinct response reviewers, exact text evidence and unresolved disagreements while preserving the input.
Check OCR Correction Ledger
Check ocr-correction-ledger.json for page/span references, exact original text slices, overlapping proposed edits and reviewer evidence on verified corrections, without applying changes.
Check Survey Paths
Check survey-paths.json for complete single-choice routes, cycles, reachable nodes and expected test paths without editing the file or contacting a form service.