How Enterprises Can Create a High-Quality AI Use Case on the GenAI Fund Platform

A practical guide to defining AI projects that unlock revenue, new business value, and operational impact.

Across Southeast Asia, enterprises are rapidly embracing AI — yet most still struggle with the first and most important step: defining a clear, actionable AI use case.

A strong use case helps your organisation:

  • Align business and technical stakeholders

  • Communicate needs clearly on the GenAI Fund (GAF) Platform

  • Attract the right AI startup partners

  • Shorten the timeline from idea → PoC → deployment

This guide walks you step-by-step through each section of the Use Case Submission Form, offering copyable formulas, key component explanations, and revenue-focused examples.

Section Formula Key Component Explanations  Example 
1. Problem Statement [Business context] + [Who is affected] + [Pain point] + [Revenue/value impact] Business context: industry, scale, operating environment.

Who is affected: teams, roles, customers impacted.

Pain point: the specific workflow bottleneck.

Revenue/value impact: how the issue hurts revenue, activation, conversion, or cross-sell.

 We are a regional retail bank serving over 3 million customers. Our digital sales and onboarding teams rely on manual lead qualification and static recommendation rules, limiting relevance and slowing follow-up. As a result, high-intent customers often fail to complete onboarding or miss suitable product offers. This leads to lower activation rates and revenue leakage across credit card, loan, and wealth products.
2. Expected Outcomes & Success Metrics “We aim to achieve [business outcome] by improving [process] by [X%], resulting in [value impact].” Business outcome: revenue increase, conversion uplift.

Process: the operational lever (e.g., personalisation accuracy, lead scoring).

Target metric: % improvement expected.

Value impact: revenue growth, new business, margin gain.

 We aim to increase digital onboarding and product cross-sell conversion by improving recommendation and lead qualification accuracy by 30%. More personalised offers will encourage customers to complete onboarding and apply for relevant financial products. This uplift will drive increases in revenue from new-to-bank customers. The initiative is expected to materially improve quarterly retail banking performance.
3. Current Solution “Today, we handle [process] using [tools]. This results in [limitations], which leads to [impact].” Process today: manual, fragmented, or outdated workflow.

Tools used: spreadsheets, email, legacy CRM.

Limitations: delays, variability, low insight.

Impact: revenue loss, poor experience, low conversion.

 Today, we manage digital banking leads manually using spreadsheets, email, and inconsistent CRM updates. This causes slow follow-up, poor lead visibility, and lost opportunities to engage high-intent prospects. Many customers disengage before completing onboarding or applying for products. This contributes to lower conversion rates across cards, loans, and wealth offerings.
4. Target Users & Impacted Teams “This solution will be used by [team] and impact [beneficiaries]. It covers [X] users across [locations].” Users: who will use the AI tool.

Beneficiaries: who gains value (e.g., customers, managers).

Scale: # of users or branches.

Context: daily operational fit.

 This solution will be used by our 25-person Digital Sales Operations team and will directly impact frontline RMs across 80 branches. Wealth advisors and lending specialists will also benefit from more qualified leads and higher conversion potential. Users across web and branch channels will interact with personalised recommendations. This is expected to significantly increase customer lifetime value.
5. Data Availability & Readiness “We have [data type] in [format], with [volume]. Data is [quality] and stored in [system].” Data type: transactions, onboarding, product applications.

Format: structured tables, logs, PDFs.

Volume: quantity available for modelling.

Quality/storage: consistency + where it lives.

 We have over 10 million customer interactions, including onboarding data, transaction histories, and application records, stored in structured tables within our enterprise data warehouse. The data is consistently formatted with timestamps, customer IDs, and product attributes. Quality is high and refreshed daily. This dataset is ideal for revenue-focused models such as propensity scoring and personalised cross-sell.
6. Integration, Deployment & Infrastructure Requirements “The solution must integrate with [systems], support [security], be deployed in [environment], and enable [workflow needs].” Integrations: CRM, onboarding system, product engine.

Security: SSO, VPC, access control.

Deployment: cloud/on-prem/VPC.

Workflow needs: dashboards, approvals, API outputs.

 The solution must integrate with our digital onboarding system, CRM, and product recommendation engine to deliver real-time personalised offers. It must support SSO, comply with banking data residency rules, and run in our secure VPC. Digital sales leaders require dashboards to monitor uplift and adjust strategies. Output must be API-based to populate web, mobile, and branch interfaces.
7. Budget Input a number or TBD (recommended to enter a number) Why it matters: signals commitment, attracts better startups.

Typical PoC: USD 10k–50k.

Guideline: even rough estimates help.

 Budget: USD 40,000. A clear budget ensures startups provide accurate, revenue-aligned proposals and allocate senior resources to drive measurable uplift in banking product sales.
8. Tags (Optional) Choose 2–3 tags (industry / process / workflow) Examples: Banking, Digital Sales, Personalisation.

Purpose: improves search + matching accuracy.

 Banking, Digital Sales, Personalisation.
9. Attachments (Optional) Upload workflow samples, screenshots, documents Purpose: improves clarity + startup proposal accuracy.

Recommended: anonymised forms, screenshots, journeys.

 Attachments such as anonymised onboarding forms, CRM screenshots, or customer journey maps help startups design precise revenue-generating solutions.

 

1. How to Write a Strong Problem Statement

A high-quality problem statement sets up the entire success of your AI project.
It must clearly outline what the problem is, who is affected, and why it matters for revenue or enterprise value.

A strong problem statement always includes four components:

Key Component 1: Business & Industry Context

What it is: A short description of the business environment where the problem exists.
Why it matters: AI startups can only design relevant solutions when they understand your model, scale, and industry constraints.
Simple examples:

  • “We are a regional e-commerce platform serving 10M monthly shoppers.”

  • “We operate 150 F&B outlets across major Southeast Asian cities.”

Key Component 2: Who Is Affected

What it is: The teams, roles, or customers involved.
Why it matters: Different users require different workflows and integrations.
Examples:

  • “Our online sales optimisation team of 40 specialists…”

  • “Branch sales advisors and onboarding staff…”

Key Component 3: Current Pain Point

What it is: The operational bottleneck today.
Why it matters: AI cannot solve vague issues — only specific friction points.
Examples:

  • “Personalisation rules are manually maintained and outdated.”

  • “Lead qualification is slow and inconsistent.”

Key Component 4: Business Impact (Revenue or Value Loss)

What it is: Clear consequences tied to revenue, margin, customer experience, or productivity.
Why it matters: Business impact helps prioritise and justifies investment.
Examples:

  • “We miss upsell opportunities.”

  • “Slow follow-up reduces high-value conversions.”

Copyable Formula

[Business context] + [Who is affected] + [Pain point] + [Revenue/value impact]

Comprehensive Example 1 — E-Commerce Revenue Growth

We are a regional e-commerce marketplace serving over 10 million monthly shoppers. Our merchandising and product discovery teams struggle with outdated manual tagging and rule-based recommendations, limiting product relevance and cross-sell opportunities. As a result, customers frequently abandon browsing sessions due to irrelevant suggestions. This leads to an estimated 18% GMV loss driven by poor personalisation accuracy.

Comprehensive Example 2 — F&B New Revenue Stream Creation

We operate 150 F&B outlets with a growing digital ordering channel. Our marketing team manually builds promotions without any data-driven segmentation, resulting in low repeat rates and missed upsell opportunities. Customers do not receive relevant recommendations for bundles or meal add-ons. This prevents us from launching subscription products or personalised bundles that could unlock a significant new revenue stream.

2. How to Choose the Right AI Capabilities

You don’t need to be a technical expert — just select the capabilities related to your problem. Startups will refine the final approach.

A quick guide:

  • Generative AI — automate content creation (emails, reports, marketing assets).

  • Conversational AI — automate customer or employee interactions.

  • Predictive Analytics — forecast demand, churn, behaviour.

  • Computer Vision — interpret photos/videos for inspections or verification.

  • Recommendation Systems — deliver personalised offers, bundles, next-best actions.

  • Document Intelligence — extract structured information from PDFs or scanned files.

  • Voice AI — transcribe calls, automate hotline interactions.

  • Anomaly Detection — detect unusual patterns (fraud, safety risks).

  • Search & Knowledge Retrieval — improve SOP/policy searchability.

  • RPA — automate repetitive tasks.

  • Other — multi-step agent workflows or optimisation engines.

Your goal is simply to indicate the direction of the solution. Startups will handle the technical decisions.

3. How to Define Expected Outcomes & Success Metrics

The strongest AI outcomes are always tied to revenue uplift, new business creation, or productivity linked to cost or margin impact.

A strong outcomes section includes four components:

Key Component 1: Business Outcome

What it is: The high-level goal (preferably revenue-related).
Examples:

  • “Increase conversion rate.”

  • “Grow repeat purchases.”

Key Component 2: Process to Improve

Why it matters: Shows how AI will influence the outcome.
Examples:

  • Recommendation accuracy

  • Lead qualification

  • Offer optimisation

Key Component 3: Measurable Target

Why it matters: Helps startups design measurable PoCs.
Examples:

  • “Improve by 20%”

  • “Reduce by 40%”

Key Component 4: Value Realised

Why it matters: The business case for leadership.
Examples:

  • Revenue gain

  • Margin improvement

  • New recurring revenue

  • Reduction in churn-related losses

Copyable Formula

“We aim to achieve [business outcome] by improving [process] by [X%], resulting in [revenue uplift / new business / cost savings].”

Comprehensive Example 1 — Revenue Uplift

We aim to increase sales conversion for high-intent shoppers by improving personalisation accuracy by at least 25%. More relevant product suggestions will reduce browsing drop-off and increase average order value. This directly contributes to GMV growth. The solution will also support more efficient cross-sell and upsell strategies.

Comprehensive Example 2 — New Business Stream

We aim to launch an AI-powered meal subscription offering by predicting customer preferences and optimising personalised bundles. Improving bundle accuracy and pricing optimisation by 30% will enable us to scale subscription adoption. This creates a recurring revenue stream and increases customer lifetime value. The new business line is expected to contribute a material share of digital revenue within 12 months.

4. How to Describe Your Current Solution Clearly

Startups need an accurate baseline to design a feasible AI solution.
A clear description includes three components:

Key Component 1: How the Process Works Today

Examples:

  • “Manual review through spreadsheets.”

  • “Handled through email and messaging apps.”

Key Component 2: Tools & Systems Used

Examples:

  • Excel, Google Drive

  • WhatsApp

  • SAP, Salesforce

  • Internal portals

Key Component 3: Limitations & Impact

Examples:

  • Slow processing → revenue loss

  • Poor visibility → missed opportunities

  • Inconsistent quality → customer dissatisfaction

Copyable Formula

“Today, we handle [process] using [tools]. This results in [limitations], which leads to [delay, cost, or revenue loss].”

Comprehensive Example (Revenue Impact)

Today, we manage inbound inquiries manually through email and WhatsApp, and updates to CRM occur inconsistently. This leads to delayed follow-ups and missed opportunities to engage high-intent prospects. Many potential buyers drop off before any salesperson contacts them. This directly reduces conversion rates and limits our overall revenue growth.

5. How to Identify Target Users & Impacted Teams

Knowing who will use the solution helps startups design better workflows.

Include three components:

Key Component 1: Primary Users

Examples:

  • Marketing operations

  • CRM teams

  • Sales advisors

  • Customer service agents

Key Component 2: Beneficiaries

Examples:

  • High-value customers

  • Category managers

  • Business unit leaders

Key Component 3: Scale

Examples:

  • “15 analysts”

  • “50 advisors across 12 branches”

Copyable Formula

“This solution will be used by [team] and impact [beneficiaries]. It covers approximately [X] users across [locations].”

Comprehensive Example (Value Creation)

This solution will be used by our CRM and loyalty operations team of 12 analysts to drive personalised customer engagement. It will benefit more than 500,000 active loyalty members across our digital channels. Category managers will leverage insights to optimise campaign ROI and revenue contribution. The impact is expected to significantly increase customer lifetime value.

6. How to Explain Data Availability & Readiness

AI solutions depend heavily on data.
A good data section includes four components:

Key Component 1: Data Type

Examples:

  • Transactions

  • Customer interactions

  • Product catalogues

  • Documents

Key Component 2: Format

Examples:

  • SQL tables

  • CSV

  • PDFs

  • Images/videos

Key Component 3: Volume

Examples:

  • “20M customer transactions”

  • “50,000 invoices”

  • “18 months of chat logs”

Key Component 4: Quality + Storage

Examples:

  • “Mostly clean and consistent”

  • “Mixed quality across branches”

  • “Stored in Salesforce / SAP / Data Warehouse”

Copyable Formula

“We have [data type] in [format], with approximately [volume]. Data is [quality] and stored in [system].”

Comprehensive Example (Personalisation Model)

We have more than 20 million transaction records in structured tables within our cloud data warehouse. The dataset includes SKU-level detail, timestamps, and customer profiles that enable behavioural modelling. Quality is high and refreshed daily from our POS and e-commerce systems. This supports AI models that drive revenue growth through personalised recommendations and upsell strategies.

7. How to Describe Integration, Deployment & Technical Requirements

This helps startups design a solution compatible with your IT environment.
Include four components:

Key Component 1: Required Integrations

Examples:

  • Salesforce

  • Shopify

  • SAP

  • Internal data warehouse

Key Component 2: Security Requirements

Examples:

  • SSO

  • RBAC

  • Data residency

  • VPC deployment

Key Component 3: Deployment Preference

Examples:

  • Cloud

  • On-prem

  • Hybrid

  • Enterprise VPC

Key Component 4: Workflow or Access Requirements

Examples:

  • Manager approvals

  • Dashboard needs

  • API-based outputs

  • Mobile-first UI

Copyable Formula

“The solution must integrate with [systems], support [security], and be deployed in [environment]. Users require [workflow/access needs].”

Comprehensive Example (Driving Revenue Uplift)

The solution must integrate with our Shopify storefront and internal CRM to enable real-time personalised recommendations. It should support SSO and be deployed within our cloud VPC to meet security and data governance requirements. Category managers will require dashboards to monitor uplift performance and adjust recommendation rules. The system must expose APIs that feed directly into our merchandising engine to drive revenue impact.

8. How to Specify Your Budget

You can input a number or select TBD — but a clear budget number always leads to better startup engagement.

Typical PoC ranges: USD 10,000–50,000

Examples:

  • “30,000” (recommended)

  • “TBD” (only if truly necessary)

9. Tags (Optional)

Choose 2–3 tags describing:

  • Industry

  • Business process

  • Workflow type

Examples:

  • Retail, Personalisation, Marketing

  • Finance, Compliance, Document Processing

10. Attachments (Optional but Very Helpful)

Strong attachments help startups understand your workflow immediately.

Useful uploads include:

  • Sample documents

  • Screenshots

  • SOPs or process flows

  • Example outputs

Even one attachment improves accuracy of startup proposals.

Final Thoughts: Why This Matters

A high-quality use case leads to:

  • Better startup matching

  • Faster PoC design

  • More accurate budgeting and timelines

  • Higher revenue and business value outcomes

  • Faster movement from experimentation to production

When enterprises define their needs clearly, the GAF Platform can match them with the right AI partners — accelerating transformation across the entire region.