Reusable AI pattern templates

Generative AI · Enterprise

Draft internal communications or documents with mandatory human review before release.

Problem shape: Knowledge workers spend disproportionate time on first drafts of routine documents.

Generative AI · SME

Suggest reply drafts for common customer questions; agent always sends the final message.

Problem shape: Small teams lack capacity to answer repetitive inquiries consistently.

Generative AI · Government

Assist caseworkers drafting citizen correspondence with mandatory supervisor release.

Problem shape: Case volumes create backlog on routine correspondence.

Generative AI · Defense

Support staff officers drafting planning narratives offline from operational systems.

Problem shape: Planning staff spend time reformatting known inputs into narrative form.

Generative AI · Regulated industry

Help prepare first drafts of regulated disclosures; compliance owns release.

Problem shape: Disclosure cycles consume scarce compliance capacity on formatting and first pass.

Retrieval-augmented generation (RAG) · Enterprise

Answer employee questions only from a curated, versioned policy corpus with citations.

Problem shape: Employees cannot reliably find the current approved policy answer.

Retrieval-augmented generation (RAG) · SME

Retrieve answers from the company's product handbook with source links.

Problem shape: New hires and partners ask the same handbook questions repeatedly.

Retrieval-augmented generation (RAG) · Government

Retrieve published guidance for citizens with clear citations and 'not legal advice' notice.

Problem shape: Citizens struggle to navigate fragmented published guidance.

Retrieval-augmented generation (RAG) · Defense

Retrieve from an authorized doctrine library; refuse out-of-corpus questions.

Problem shape: Staff lose time locating the correct doctrine excerpt.

Retrieval-augmented generation (RAG) · Regulated industry

Return the current approved procedure step with version stamp.

Problem shape: Frontline staff risk using outdated printed procedures.

Predictive ML · Enterprise

Produce forecasts as decision support; planners own the plan.

Problem shape: Planners lack timely signal on demand shifts.

Predictive ML · SME

Flag SKUs likely to stock out; purchasing confirms orders.

Problem shape: Stockouts and overstock both hurt cash and service.

Predictive ML · Government

Forecast expected service volumes to inform staffing — not automated rostering.

Problem shape: Demand spikes create citizen wait times.

Predictive ML · Defense

Estimate consumable demand for logistics planning with human confirmation.

Problem shape: Logistics planners need earlier demand signals for consumables.

Predictive ML · Regulated industry

Surface a risk score to underwriters/analysts; humans make the decision.

Problem shape: Analysts need consistent risk signals without ceding authority.

Computer vision · Enterprise

Flag potential defects on a line; quality engineers confirm disposition.

Problem shape: Manual inspection misses intermittent defects under volume.

Computer vision · SME

Detect empty shelf facings from store cameras or handheld photos for staff follow-up.

Problem shape: Small retailers lack continuous shelf auditing capacity.

Computer vision · Government

Triage citizen/inspector photos of infrastructure issues for work-order queueing.

Problem shape: Inspection photo backlogs delay maintenance prioritization.

Computer vision · Defense

Assist maintainers reviewing equipment condition imagery; no targeting use.

Problem shape: Maintainers need consistent first-pass condition cues.

Computer vision · Regulated industry

Flag possible PPE non-compliance for supervisor follow-up — not automated discipline.

Problem shape: Safety teams cannot watch every zone continuously.

Automation / RPA+ · Enterprise

Extract invoice fields into AP queue; clerks confirm before posting.

Problem shape: AP clerks re-key high volumes of similar invoices.

Automation / RPA+ · SME

Push order status updates from system of record to customer channels with templates.

Problem shape: Owners manually copy status into email/chat.

Automation / RPA+ · Government

Validate required fields and attachments on intake forms before case creation.

Problem shape: Incomplete submissions create rework loops for citizens and staff.

Automation / RPA+ · Defense

Suggest ticket routing for IT/support desks; agents confirm assignment.

Problem shape: Misrouted tickets delay resolution.

Automation / RPA+ · Regulated industry

Check that required KYC documents are present before analyst review.

Problem shape: Analysts waste time on incomplete KYC packs.

Agentic workflow · Enterprise

Agent gathers internal sources into a brief; human approves any external send or action.

Problem shape: Analysts spend cycles assembling known internal sources.

Agentic workflow · SME

Propose meeting or job slots; human confirms bookings.

Problem shape: Owners lose time coordinating schedules across channels.

Agentic workflow · Government

Assemble case packs from approved systems; officer decides next step.

Problem shape: Officers spend time gathering the same case artifacts repeatedly.

Agentic workflow · Defense

Propose logistics coordination steps inside policy bounds; human executes.

Problem shape: Coordination overhead slows routine logistics tasks.

Agentic workflow · Regulated industry

Collect evidence artifacts for a control test; compliance reviewer closes the finding.

Problem shape: Evidence chase consumes audit cycles.