Interpretation and Decision Problems
| Worker Use Case | Procurement and Supplier Comparison |
| What it does | A procurement team receives multiple supplier proposals, tender responses, or product quotations. Rather than a procurement manager reading each document separately and building a comparison manually in a spreadsheet, an AI worker reads all the documents, extracts comparable fields including pricing, terms, delivery timelines, warranties, exclusions, and references, normalises them into a common structure, highlights trade-offs, flags compliance requirements the proposals do or do not meet, and produces a recommendation memo with the analysis clearly laid out. |
| What triggers it | A set of supplier proposals arriving following an RFQ or tender process, or a procurement manager initiating a product evaluation. |
| What it produces | A structured comparison document showing each supplier or product across a common set of criteria, with a recommendation and a summary of key trade-offs. |
| Where the human sits | The procurement or sourcing lead reviews the comparison, applies commercial judgment the AI cannot have including supplier relationship history, strategic fit, and risk appetite, and makes the final award decision. |
| Where it shows up | Procurement teams in mid-to-large organisations, IT sourcing teams evaluating software vendors, legal teams comparing outside counsel proposals, and operations teams assessing logistics or facilities contracts. |
| Why it matters | This card and Contract and Document Intelligence are variations of the same underlying operation: read multiple long-form documents, extract comparable content, surface differences. The distinction is that procurement comparison is about selecting between options, while contract intelligence is about reviewing a single document against standards. Same cognitive operation, different decision context. |
Examples
- Product comparison agent — reads supplier product specifications, pricing sheets, and customer reviews, normalises them into a common comparison structure, and produces a recommendation memo with trade-offs clearly highlighted
- Supplier proposal comparison worker — processes multiple tender responses, extracts comparable fields across each submission, and identifies which proposals meet mandatory criteria before the procurement manager reviews
- Compare competing options assistant — analyses two or more vendor proposals against a defined set of evaluation criteria and produces a structured scoring matrix with a recommended shortlist for procurement sign-off
- Purchase recommendation generator — takes a procurement requirement, retrieves the approved vendor list and catalogue, applies policy rules around preferred suppliers and spend thresholds, and recommends the most compliant purchasing option
- Vendor shortlist and trade-off memo assistant — condenses a full supplier evaluation into a concise decision memo suitable for presenting to a procurement committee, with risks and trade-offs explicitly named
- Contract obligation tracker — monitors active supplier contracts for upcoming renewal dates, performance obligations, and payment milestones, and alerts the procurement team before a deadline passes unnoticed
| Worker Use Case | Contract and Document Intelligence |
| What it does | A contract, legal filing, regulatory document, or complex policy document needs to be reviewed. Rather than a lawyer or compliance officer reading every line from scratch, an AI worker reads the document, extracts key clauses including term, termination rights, liability caps, indemnities, data processing obligations, and exclusivity, compares them against the organisation’s approved standard positions, flags deviations, suggests redline language for non-standard clauses, and produces a summary that tells the reviewer exactly where their attention is needed. |
| What triggers it | A contract received from a counterparty for review, a regulatory filing requiring analysis, or a document uploaded to a legal workflow platform. |
| What it produces | A redlined document with suggested changes, a clause extraction summary, a deviation report comparing the document to standard positions, and a risk flag summary for the reviewing lawyer or compliance officer. |
| Where the human sits | A lawyer or compliance officer reviews the flagged deviations, applies legal judgment to proposed redlines, makes negotiation decisions, and approves the final document. The AI does not execute changes. It proposes them. |
| Where it shows up | In-house legal teams reviewing commercial contracts, compliance teams reviewing regulatory filings, procurement teams reviewing supplier agreements, and real estate teams reviewing lease documents. |
| Why it matters | This is Document Intelligence in its most direct form. The cognitive work, which covers extracting content, comparing against standards, flagging deviations, and proposing corrections, is identical whether the document is a commercial contract, an employment agreement, a construction tender, or a data processing agreement. Only the clause library and review criteria differ. |
Examples
- Contract redlining assistant — reads a counterparty’s contract draft, compares each clause against the organisation’s approved standard positions, and proposes redline language for non-standard terms for the reviewing lawyer to accept or modify
- Contract clause extraction agent — processes large volumes of contracts and extracts key data points such as term, notice period, liability cap, and renewal date into a structured repository for ongoing management
- Discovery timeline analysis assistant — reads legal case materials, extracts events and dates, builds a chronological timeline, and surfaces patterns relevant to the legal team’s case strategy
- Regulatory filing summariser — reads lengthy regulatory documents and produces a structured summary identifying obligations, deadlines, and required actions for the compliance team
- Proposal and document discrepancy checker — compares two versions of a document or two related documents and flags inconsistencies in terms, figures, or commitments before either document is signed or submitted
| Worker Use Case | Marketing and Campaign Content |
| What it does | Marketing teams need two related but distinct things: content that conforms to brand standards, and analysis that tells them whether that content is working. On the drafting side, an AI worker generates campaign copy, social content, email sequences, and promotional material using approved brand guidelines, tone of voice, and legal constraints. On the analysis side, it pulls campaign performance data from connected platforms, identifies what is working and what is not, and produces recommendations for optimisation. |
| What triggers it | A campaign brief, a content calendar entry, a performance report request, or a set of campaign results uploaded for analysis. |
| What it produces | Brand-aligned draft content ready for review, or a campaign performance analysis with optimisation recommendations. |
| Where the human sits | A marketing manager or brand lead reviews all draft content before publication, approves performance analysis interpretations, and makes final decisions about campaign changes. |
| Where it shows up | Marketing teams in any commercial organisation, communications teams in nonprofits and public sector organisations, and growth teams in startups managing multi-channel campaigns. |
| Why it matters | Marketing teams often procure two separate tools for content generation and campaign analysis, when both are expressions of capabilities the organisation should already have in shared infrastructure. This card makes that redundancy visible. Structured Drafting handles the generation layer. Interpretation and Normalisation handles the analysis layer. One shared infrastructure. Two applications. |
Examples
- Campaign performance analyst — pulls data from connected marketing platforms, identifies which channels and creatives are performing above and below benchmark, and produces an optimisation recommendation report
- Promotional content generator — produces on-brand campaign copy for email, social, and paid channels from a campaign brief, using approved tone guidelines and audience segmentation parameters
- Content adaptation worker — takes a piece of approved content and reformats it for different channels and audiences: a long-form article becomes a LinkedIn post, an email series, and a set of social captions without requiring a new brief each time
- Ultra-personalised campaign generator — uses customer segment data from the CRM to produce individualised email content at scale, with each version reflecting the recipient’s purchase history, preferences, and lifecycle stage
- Marketing campaign management assistant — tracks campaign milestones, content approvals, and publication schedules across multiple active campaigns, flagging delays and surfacing blockers for the marketing manager
- Brand asset drafting assistant — produces first-draft brand copy for product descriptions, landing pages, and promotional materials aligned to approved brand guidelines and legal sign-off requirements
| Worker Use Case | Data Interpretation, Forecasting, and Decision Support |
| What it does | Decision-makers need to understand what their data is telling them, not just what the numbers are, but what they mean and what should happen next. An AI worker pulls data from connected reporting systems, interprets trends and anomalies, compares current performance against targets or historical benchmarks, builds forward-looking models based on defined assumptions, and produces a decision memo that translates the numbers into a recommendation with a clear explanation of the reasoning and the confidence level. |
| What triggers it | A periodic reporting cycle such as a monthly close or quarterly review, an anomaly detected in a live dashboard, or a request from a decision-maker preparing for a board or leadership meeting. |
| What it produces | A performance interpretation summary, a forecast model with stated assumptions, or a decision memo recommending a course of action based on the data, with all source data cited so the recommendation can be challenged. |
| Where the human sits | The analyst, finance director, or business leader reviews the interpretation, stress-tests the assumptions in the forecast, and makes the final decision. The AI produces the analysis. It does not make the decision. |
| Where it shows up | Finance teams preparing management accounts and forecasts, sales operations teams analysing pipeline and revenue performance, marketing teams interpreting campaign analytics, and strategy teams building business cases. |
| Why it matters | This is Interpretation and Normalisation at its most visible, and its most consequential. The dangerous version of this card treats the AI’s interpretation as authoritative. The safe version keeps the AI in the role of analysis producer and the human firmly in the role of decision-maker. The human checkpoint here is not optional, it is the whole point of the pattern. |
Examples
- Analytics dashboard interpreter — connects to reporting platforms, reads current performance metrics, compares them to targets and prior periods, and produces a plain-language summary of what the data is saying for non-technical decision-makers
- Sales report generator — pulls pipeline and revenue data from the CRM, applies defined segmentation and period filters, and produces a formatted sales performance report for management review
- Fundraising performance analyst — retrieves donation data, campaign results, and donor behaviour patterns, and produces an analytical report identifying what is working and where fundraising performance needs attention
- Business finance organiser — categorises transactions, reconciles accounts, and produces a structured financial overview for small business owners or department budget holders without requiring accountancy expertise
- Demand forecast worker — applies defined forecasting models to historical sales data, seasonality patterns, and pipeline signals, and produces a forward-looking demand forecast with confidence intervals for the planning team
- Investment memo drafter — assembles financial model outputs, market data, risk factors, and deal rationale into a structured investment memo format for review by the investment committee