OrgReview — User Guide

AI Organisational Impact Analysis · SIA

New

Use case scoring redesigned — July 2026

August 2026: Task enrichment — ESCO (European) and O*NET (American) database enrichment now available on the Upload tab; appends standardised tasks to extracted roles for consultant review · Earlier updates: Session passwords — protect sessions with a password via New ID · Extract & Review flow — single button starts extraction and takes you straight to the Review tab · Quality Check removed — pipeline runs directly to AI scoring · FTE Savings option on Review tab

Sign In & Set Your Session

Enter the password when prompted on first load. Your session is saved in the browser — you will not need to re-enter it unless you clear your browser data.

Session ID: Each analysis run needs its own Session ID — the unique identifier for that specific run. Use a short memorable name like acme-baseline or acme-scenario-2 (letters, numbers, hyphens only). You can:

• Type a name — Type your chosen name into the Session ID field and press Enter or click Use this ID.
• New ID — Opens a modal where you can edit the generated ID to any name you prefer, then choose whether to add a password before creating the session.
• Copy / Share — Copy the Session ID and give it to a colleague so they can load the same session on their browser.
• Delete — Remove all server data for the current session (a warning appears for sessions older than 7 days).

⚠️ Set the Session ID before uploading files. Files are stored under whichever ID is active when you upload.

Session passwords: When clicking New ID, the modal lets you optionally protect the session with a password:

• The generated ID is shown in an editable field — change it to anything you like before creating
• Click Yes under "Set a password?" to reveal the password input, then type your chosen password
• Click No to skip — session is created without a password
• Once set, anyone who types or switches to that session ID will be prompted for the password before access is granted
• ⚠️ Passwords cannot be recovered — make sure to remember it or note it somewhere safe

Projects (multi-scenario): Use the two buttons at the right of the session strip to organise runs by client:

• 📁 Save to project — Click, type a project name (e.g. acme-corp), click Save. The current run is now filed under that project. You can save multiple runs to the same project name.
• 🔍 Open project — Click, type a project name (partial match works). All runs saved to matching projects are listed — click Switch next to any run to load it instantly.

Projects are stored in your browser only — client names are never shown on screen unless you open one of these popovers.

Upload Client Documents — Upload tab

Go to the Upload tab. Drag and drop or click to upload any combination of:

• Org charts — Excel, CSV, or PDF
• Job descriptions — Word, PDF, or plain text
• Headcount reports — any spreadsheet format
• Process documents — PDFs describing workflows
• Process diagrams — BPMN (.bpmn) or image exports (PNG/JPG) of swimlane / flowchart process maps.

Once your files are uploaded, optionally enable Task enrichment below the file list — choose ESCO (European database) or O*NET (American database) to append standardised tasks from an occupational database to your extracted roles. Then click ▶ Extract & Review →. This starts the extraction pipeline and takes you to the Review tab.

Review Extracted Roles — ✏️ Review tab

Once extraction is complete, the Review tab shows:

• Stats — total roles, departments, and tasks extracted
• Preview — the first 5 roles with their tasks (click any role to expand its task list)
• FTE Savings — choose Yes or No for headcount-weighted FTE savings calculations

Then choose one of two paths:

• ✏️ Extra Review — Download the prepared-roles.xlsx, edit it in Excel or Google Sheets (columns: source_file | role_title | department | headcount | task), and upload the corrected version before running analysis.
• ⚡ Run Analysis — Everything looks good — start the full analysis pipeline immediately. You are switched to the Analyse tab automatically and the pipeline begins.

Run the Analysis — Analyse tab

Go to the Analyse tab and click ▶ Run Analysis. The pipeline automatically processes all files you uploaded in your session.

The pipeline runs these stages automatically:

1. Ingest — reads and extracts text from all uploaded files
2. Extract — identifies roles, departments, and headcount from the text
3. Analyse — scores every role E0–E3 using AWS Bedrock (Claude Opus 4.7) with AEI calibration
4. Aggregate — computes department and enterprise-level summaries
5. Generate — writes all output reports (Excel, HTML, JSON)
6. Use Cases — identifies 5–12 AI use cases and scores them on 6 KPIs

The full pipeline typically takes 3–8 minutes depending on the size of the organisation. A live progress bar shows each stage in real-time.

Job History: Below the progress bar, you'll see a history of all pipeline runs for this session. You can:

• Reconnect — Resume watching a running job if you refresh the page
• Cancel — Stop a running job
• View Results — Jump directly to results for completed jobs

Session Lock: Once a pipeline run completes, the session is locked. To run a new analysis, generate a new Session ID.

📚 References for the E-level scoring model

The E0–E3 framework and the realistic time-saved weights (E1 = 0.5, E2 = 0.3, E3 = 0.2) are grounded in four papers, all embedded verbatim (with page citations) into the Bedrock system prompt at runtime:

• Anthropic — Massenkoff, M. & McCrory, P. (2026). Labor market impacts of AI: A new measure and early evidence. Published 5 March 2026. (With companion Appendix.) The primary empirical source. Introduces the "observed exposure" measure that underlies reference/job_exposure.csv. Verbatim methodology from p. 6: "fully automated implementations receive full weight, while augmentative use receives half weight." This is the direct justification for our E1 = 0.5 weight. Also supplies the top-10 exposure table (Figure 3, p. 7) used as empirical anchors in calibration. The companion Appendix provides the full per-occupation SOC-level exposure table used to build job_exposure.csv, plus the methodological detail behind the automation/augmentation split.
• OpenAI — Patwardhan, T. et al. (2025). GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks. Capability benchmark covering 1,320 tasks across 44 occupations and the 9 sectors contributing most to U.S. GDP. Headline result (§3.1, p. 5-6): "47.6% of deliverables by Claude Opus 4.1 were graded as better than (wins) or as good as (ties) the human deliverable." Directly supports E1 = 0.5 — parity on roughly half of real knowledge-work tasks. §3.2 on "try n times, if unsatisfactory, fix it yourself" grounds E2 = 0.3 (human-in-the-loop review cycles).
• McKinsey Global Institute (2023). The economic potential of generative AI: The next productivity frontier. Chui, M., Hazan, E., Roberts, R., et al. June 2023. Sets the theoretical ceiling we deliberately discount. Key insight #4, p. 3: "Current generative AI and other technologies have the potential to automate work activities that absorb 60 to 70 percent of employees' time today." Our weights (E1=0.5, E2=0.3, E3=0.2) explicitly sit below this theoretical ceiling to reflect realistic, today-deployable savings. Customer-service case (p. 15) — AI can decrease productivity for highly skilled agents — is the concrete evidence that E1 ≠ 1.0.
• OECD (2022). OECD Framework for the Classification of AI Systems. OECD Digital Economy Papers No. 323, February 2022. Source of the four-level autonomy typology (p. 53) that maps onto E0/E1/E2/E3. Defines human-support, human-in-the-loop, human-on-the-loop, and human-out-of-the-loop autonomy. Page 53 also establishes that high-autonomy AI is policy-gated in critical or rights-affecting contexts — directly justifies E3 = 0.2 (deployment-limited today even when capability is high).

Weight choice rationale: E1 = 0.5 (Anthropic — Massenkoff & McCrory half-weight rule for augmentation), E2 = 0.3 (human review keeps ~70% of task time), E3 = 0.2 (high theoretical capability but deployment-limited today per OECD autonomy framework).

Full calibration formula: Calibrated E-score = 0.7 × task-weighted score + 0.3 × Anthropic observed_exposure (Massenkoff & McCrory). The 70/30 blend keeps the model sensitive to role-specific task detail while anchoring to empirical data.

Review Results — Results tab

Go to the Results tab. You will see:

• Enterprise Summary card — overall exposure band, average E-score, role and headcount totals, with a headline stat showing percentage of tasks AI-impactable
• E-level stacked bar chart — visual breakdown of E0/E1/E2/E3 distribution across the organization
• Department Breakdown table — E-score and exposure band per department with E-level bars. Click any department row to expand and see individual roles
• Per-role drill-down — expanded department view shows each role with its own E-level stacked bar
• AI Use Case Assessment table — ranked use cases scored on 6 dimensions (V1–V3 value + F1–F3 feasibility, 1–3 scale) with separate value, feasibility, and combined scores
• Excel download — full task-level analysis with a Department Summary sheet, ready for client delivery
• HTML reports — heatmap, implementation matrix, FTE savings, ROI, impact assessment, QA report — all previewable in the browser

Click 👁 Preview on any HTML file to open it in a new tab. Click ⬇ Download to save any file locally. All file links include authentication tokens automatically.

AEI Calibration: E-scores are calibrated against the Anthropic Economic Index (756 SOC occupations) using a 70/30 blend of task-level analysis and empirical benchmark data. Confidence levels (high/medium/low) are reported in the FTE savings output.

📚 References for AI use case identification & the 6 KPIs

Use cases are generated from the E1/E2/E3 task list and scored on six dimensions (1–3 scale). Scores are recomputed server-side — the LLM cannot alter the arithmetic.

• V1 Breadth — how many roles and tasks the use case touches (1 = one role/few tasks → 3 = three or more roles/high task count)
• V2 Task nature — what the person is freed to do (1 = admin freed → 3 = client-facing, revenue-adjacent, or judgment-critical time freed)
• V3 Reusability — how directly the use case reuses across roles (1 = role-specific → 3 = reusable across three or more roles with minimal adaptation)
• F1 Readiness — E1 share of underlying tasks (1 = mostly E2/E3 → 3 = mostly E1, deployable now)
• F2 Data and tooling — data access and approved tooling (1 = significant dependency/blocker → 3 = ready or near-ready)
• F3 Change complexity — change management overhead (1 = senior roles/cultural shift → 3 = task-level/junior-mid/low complexity)

📚 References for the FTE savings calculation

The FTE savings headline uses: FTE saved = weighted_e_score × headcount, with an "incremental" variant that subtracts already-realised baseline adoption.

The potential vs. realised distinction motivating incremental_e_score = weighted × (1 − baseline_adoption) comes from Anthropic — Massenkoff & McCrory (2026), p. 5: "AI is far from reaching its theoretical capability: actual coverage remains a fraction of what's feasible." We don't double-count savings the client has already captured through existing tools.

Baseline-adoption detection — internal design choice, not externally cited. The keyword list used to detect current AI tool usage (ChatGPT, Copilot, Claude, Cursor, Jasper, Gemini, etc.) and the 0.0–1.0 adoption scale in analyze.py are our own heuristics. They are not drawn from a published adoption survey.

Framing note: FTE savings figures are presented as reallocatable capacity (time that can be redirected to higher-value work), not as headcount reductions. This framing is a consulting choice, not a citation-backed claim.

Ask Questions — Chat tab

Go to the Chat tab to ask natural-language questions about the analysis. Examples:

• "Which department has the highest AI exposure?"
• "What are the top 3 quick-win use cases?"
• "Explain the E-score for the Finance team"
• "Which roles should we prioritise for reskilling?"

The chat automatically uses the analysis context when results have been loaded.