RA LAB / BUSINESS CHALLENGES
Choose a challenge. Find a guide, an example and materials for a test.
RA LAB / BUSINESS PROBLEM SOLVER
Describe your situation or choose a challenge below. We will show an action plan and the intended output.
Find a verifiable delay between enquiry and payment and test a faster response.
A response-time map, AI-assisted loss hypotheses and a one-week test plan with conversion and quality checks.
Find a testable reason for churn and an appropriate next-contact moment.
A repeat-purchase map, an AI-assisted segmentation hypothesis and a retention test plan.
Find a useful follow-on offer and test its contribution without eroding trust.
A product-combination map, an offer hypothesis and a test of order value, frequency and margin.
Choose a partner hypothesis and test the path from a joint offer to actual profit.
A partner map, joint offer, pilot terms and a verifiable economics scenario.
Find a proven moment of value and test a convenient referral path.
A referral-trigger map, invitation flow and test of new lead quality.
Identify a verifiable gap between sales, collections and expenses.
A cash-timing map, deviation causes and an owner-led change plan.
Find a testable cause of funnel losses and choose one experiment.
A funnel map with conversion rates, a loss hypothesis and a measurable test plan.
Test whether a specific segment has a meaningful unmet need.
A customer job map, an evidence log and a next-test decision.
Connect segment, value, channel and sales in a testable plan.
A one-page GTM plan: segment, offer, channel, budget, metrics and owners.
Understand contribution per sale and test acquisition economics.
A revenue, variable cost, CAC and scenario worksheet with explicit assumptions.
Choose one process and test AI value against the current way of working.
An AI pilot brief: baseline, quality metric, cost, risks, human oversight and decision.
Find a workflow barrier and test a change with the team.
A barrier map, a human-control agreement and an adoption and quality test plan.
Find evidence of growth constraints and choose a manageable first step.
An evidence-based constraint map and a 90-day plan with actions, deliverables, accountable roles and review dates.
Find a cross-channel gap and verify that customers can continue without repeating their request.
A journey and context-handoff map, a gap list and an end-to-end test record for one scenario.
Matching works without saving your query. Practice answers remain in this tab until you enable saving. Do not enter names, contacts, client secrets or sensitive information.
Server saving: answers and steps are linked to a random key in a secure browser cookie for up to 90 days. This is pseudonymous, not anonymous data. It does not sync across devices.
Server saving is not enabled yet: the data controller and hosting configuration are being confirmed. You can work in this tab and download your results.
RA LAB / LEADERSHIP DECISIONS
Why did Nokia lose ground, Microsoft transform, and Kaspi and Decathlon connect customer tasks into growth systems? Explore leadership decisions and move to your own testable hypothesis.
Retail · B2B · Service
AIStart360 / Working library
18 AI prompts, 12 research papers, 12 international cases, 12 books and 6 expert quotes. Choose a challenge to explore the collection and original sources.
Prompts and templates are Bootcamp working adaptations. Research and historical cases do not promise the same outcome in your company.
02 / SOURCE CATALOG
How AI can support sales, retention, order value, partnerships, referrals and cash flow. Choose a lever to see relevant materials and tests.
B2B customer-base monitoring: missed purchase cycles, gaps in contact and a next action for repeat sales. A resource for growth lever 02 — retention.
A practical sequence for pilots, capability building, training and AI strategy.
A rigorous global view of AI capability, adoption, investment, policy and public impact.
Field evidence on productivity and heterogeneous effects across worker experience.
Shows how performance changes when knowledge work sits inside or outside the model’s capability frontier.
Explains AI through the economics of prediction and redesign of decision systems.
A systematic guide from item generation to scale development and evaluation.
A foundation for construct definition, item design, reliability and validity.
Frames demand around the progress customers hire a solution to make.
A compact method for learning from past behavior without collecting polite false positives.
Connects desired outcomes, opportunities, solutions and assumption tests in a weekly cadence.
Uses customer purpose as the endpoint that organizes journeys and measurement.
A playbook for self-serve value, time-to-value and product-led acquisition.
A practical view of freemium as a deliberate acquisition and monetization system.
Strategy as diagnosis, guiding policy and coherent action—not a list of ambitions.
A use-case-aware operating model organized around Govern, Map, Measure and Manage.
A cross-sector companion for managing generative-AI risks across the lifecycle.
International principles for trustworthy, human-centred and accountable AI.
The official risk-based regulatory framework and implementation entry point.
18 interaction guidelines covering expectations, explanations, correction, feedback and control.
Actionable patterns for user needs, mental models, explainability, feedback and graceful failure.
A model for combining high automation with high levels of meaningful human control.
A human-validated approach to accelerating intent-taxonomy construction with LLMs.
A practical framework for task-specific evals, datasets, graders and continuous evaluation.
Implementation patterns for traces, datasets, experiments and online/offline evaluation.
Design product, packaging and pricing around willingness to pay and value metrics.
A rigorous foundation for experiment design, metrics, statistics and organizational culture.
A controlled study combining objective task outcomes with subjective developer experience.
How to make disciplined experimentation a repeatable organizational capability.
Research on the organizational challenges between AI pilots and enterprise value creation.
A commerce case spanning conversational discovery, recommendations and product knowledge quality.
A reference model for AI operationalization, rigorous impact assessment and industry collaboration.
Human-driven AI research spanning decision support, interaction and responsible collaboration.
Seven implementation lessons connecting evals, domain experts, product workflows and iterative deployment.
A field guide for finding, prioritizing and expanding AI use cases from tasks into end-to-end workflows.
Patterns for deciding when an agent is appropriate, designing orchestration and applying layered guardrails.
Five principles for balancing agent autonomy with human control, transparency, privacy and security.
Empirical patterns of how AI supports tasks, occupations and human–AI division of work.
The international management-system standard for governing the responsible development and use of AI.
Guidance for documenting foreseeable effects of AI systems on individuals, groups and society.
The official entry point to Kazakhstan’s first national AI standards under the 2024–2029 AI Development Concept.
A development-focused framework for compute, connectivity, context and competency as foundations for inclusive AI value.
A practical introduction to AI that prioritizes human needs, values, capabilities and meaningful control.
An accessible map of usability, trust, ethics and feedback in products where people collaborate with AI.
A cross-functional framework for aligning product, marketing, sales and customer success around one market motion.
A capability map spanning market research, segmentation, positioning, alignment and post-launch measurement.
A step-by-step method for positioning around competitive alternatives, differentiated value and the customers who care most.
A concise positioning workflow that starts with what customers would do if your solution did not exist.
A systematic library of experiments for reducing desirability, feasibility and viability risk.
Forty-four discovery and validation experiments compared by cost, setup time, duration and evidence strength.
Extends human-centered design toward societal systems, sustainability and long-horizon responsibility.
A decision-first lens that separates objectives, actions, uncertainty and evidence before choosing models.
A strategic-narrative approach that frames market change, stakes, promised land and the evidence that makes a commercial story credible.
A systems view of growth across market, product, channel and model fit—useful for testing whether a GTM engine can compound.
Explains why market-product fit precedes scalable channel and monetization choices.
Separates growth loops from one-off tactics and balances acquisition, retention and monetization around a durable engine.
Clarifies discovery as evidence gathering across value, usability, feasibility and business viability before delivery.
A practical operating model for empowered product teams discovering solutions customers value and the business can sustain.
A detailed operating guide to CAC, LTV, churn, payback and the metrics that explain recurring-revenue growth.
A warning against false precision in early-stage unit economics and a guide to the data maturity needed for credible ratios.
Twelve metrics for diagnosing the health, efficiency and scalability of a SaaS growth engine.
A shared discovery language across Situation, Pain, Impact, Critical Event and Decision for customer-centered revenue conversations.
A six-model GTM operating system connecting revenue, customer lifecycle data, math, execution and growth stages.
A simple model connecting annual contract value, required customer count and the resulting sales and marketing motion.
Field-tested guidance on leadership, executive hiring, organization design and operating systems after product-market fit.
Using [enquiry table], [field dictionary] and [workflow rules] for [period], check duplicates, gaps, stages and payments. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [customer request and ID], [catalogue/version] and [approved rules], draft a response. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Check [original request], [draft], [verified sources] and [experiment log]. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
In an experiment with 453 professionals, access to ChatGPT reduced average writing-task time by 40% and raised assessed quality by 18%. The tasks did not reproduce the full commercial context, payment outcomes or long-term customer relationships. Include fact-checking time.
In its February 2024 launch report, Klarna described 2.3 million conversations and two-thirds of support chats handled in the first month. Historical company self-report, not independent evidence of sales growth or a statement about current service.
Ingka’s February 2024 announcement described a home-product assistant using needs, space, style and budget. A launch example, not a published controlled conversion experiment or a verified sales uplift.
Use structured questions about the customer’s situation, problem, implications and value. Draft questions from an approved brief and remove leading or invented assumptions. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Consider selling as helping someone decide. Draft a clear explanation of value without pressure or hidden terms. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Using [purchase history], [usage cycles] and [contact rules], propose transparent segments and one repeat-purchase cohort. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [confirmed purchase], [customer need] and [approved offer], propose a relevant contact time and message. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [initial cohort], [orders], [payments], [returns] and [costs], calculate repeat buyers within [window]. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Ascarza studies why the highest-risk customers are not necessarily the best targets for a retention offer. Responsiveness to intervention matters. A risk prediction does not show that a message or discount will change behaviour. Context and offer matter.
Findings. The authors studied the endowed progress effect: when an eight-step task was presented as ten steps with two already completed, people completed it more often and faster. Perceived progress, the explanation for the bonus and how it is recorded matter. [R04] This is not proof that every loyalty programme is profitable. Rewards have a cost and the mechanism must not mislead customers.
Microsoft’s May 2019 account describes reinforcement learning using purchase history, availability, time and weather. Machine-learning personalisation, not a generative chatbot or a controlled repeat-purchase effect.
Tesco’s April 2024 announcement and supplier material describe personalised challenges with time-limited rewards. Launch and mechanism evidence; the supplier has a commercial interest. This is not an independent retention or profit estimate.
Audit observed customer behaviour and cohort economics before proposing a repeat-purchase action. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Connect customer lifetime-value thinking to an explicit, testable customer strategy. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Using [customer task], [catalogue], [compatibility rules] and [costs], propose useful complements. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [approved options], [stock] and [terms], draft a recommendation explaining how it solves the task. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [fixed cohort], [orders], [returns] and [variable costs], calculate contribution per eligible customer. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Iyengar and Lepper compare limited and extensive choice sets in three studies. Smaller sets sometimes supported purchase or task completion. The findings do not establish a universal number of products or price tiers. Context and the choice task matter.
A meta-analysis of 50 experiments, 63 conditions and 5,036 participants found a mean effect near zero with substantial variation. One striking study cannot show that reducing options always helps; a near-zero mean does not rule out a local effect.
Amazon Science’s 2019 retrospective describes item-to-item collaborative filtering for a large catalogue. A technical history, not evidence for a fixed percentage of revenue attributable to recommendations.
A 2015 developer paper describes algorithms, offline evaluation and A/B tests aimed at retention and engagement. A subscription content service, not a shop-cart experiment or proof of higher order value.
Examine how choice is structured. Test clarity rather than assuming fewer options always sell better. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Using [verified partner profiles and sources], [customer segment] and [value proposition], compare complementary contributions and one joint scenario. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [agreed capabilities], draft a joint offer, handover rules and negotiation questions. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [attributed deals], [payments], [commissions], [returns] and [costs], calculate confirmed channel contribution. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Dyer and Singh’s relational view describes joint value through specific assets, knowledge exchange, complementary resources and governance. This is a theoretical framework, not a trial proving a fixed partnership profit. A contract or prominent partner is not evidence of a working business model.
Cao and Zhang study supply-chain collaboration, collaborative advantage and firm performance using US manufacturing survey data. Survey associations and structural modelling do not establish causality or transfer automatically to agency sales.
The November 2023 partnership announcement combines product and data integration with procurement through AWS Marketplace. An announced scope, not an independent estimate of additional partner or customer profit.
The May 2023 announcement combines enterprise workflow expertise with AI technology. An announcement of direction, not proof of completed implementation for every customer or universal payback.
Map ecosystem dependencies and the contributions required to deliver a customer value proposition. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Design partner channels around responsibilities, customer experience and transaction economics. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Using [confirmed value events], [feedback] and [permission], identify an appropriate moment to invite a referral. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [approved offer] and [referral rules], draft an invitation and an optional short forwarding message. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [referral IDs], [new customers], [payments], [returns], [rewards] and [costs], reconcile first purchases. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
A study followed roughly 10,000 German bank customers for almost three years; referred customers had at least 16% higher average value than matched customers, with segment differences. This is a context-specific comparison, not random channel assignment or a universal 16% improvement. Referrals are not necessarily free.
Berger and Milkman study New York Times content and experiments, linking sharing to activation and other content features. Sharing does not establish payment, trust or high-quality demand. Emotional pressure is not proposed as a practice.
Airbnb Engineering’s March 2014 account describes redesigning referrals for web and mobile, with rewards after a referred guest’s first trip. Historical company self-report about that programme, not all Airbnb bookings or generative AI.
Dropbox Help describes storage rewards for referrals, with plan-specific limits and qualifying actions. The help page establishes mechanics, not a causal contribution to business growth.
Understand why people share information. Prepare useful messages without fabricating experience or emotion. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Use customer feedback to improve service. Willingness to recommend and NPS are not actual referral purchases. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Using [reconciled invoices], [due dates], [partial payments] and [disputes], classify delay reasons and missing evidence. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [verified balance], [contract terms] and [last action], draft a staff next step or reminder. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Using [initial invoice cohort], [receipts], [due dates] and [costs], calculate timely full payment and remaining overdue balances. Use permitted or anonymised data. A person verifies facts and approves consequential actions.
Cash is King describes cash flows of US small businesses in a bank-client sample. The historical median buffer covered about 27 days of normal outflows. This is a historical sample description, not a 2026 global benchmark or a reserve rule for your company.
Barrot and Nanda study the US Federal Quickpay reform, finding recipient employment effects alongside displacement of other firms. This is a payment-reform study, not an AI-tool trial or universal collection return. Company and economy-wide effects differ.
A Danone representative describes cash-application automation in a HighRadius-hosted webinar, reporting 95% automatic matching and 75% lower process costs. Participant and supplier-platform report, without an independent audit. Matching received money is not faster customer payment.
A 2023 Microsoft presentation describes collection worklists and reports a 10% improvement in collection efficiency. A company slide without the detailed metric formula. It cannot be translated into 10% more cash, lower DSO or higher profit.
Understand financial statements, assumptions and the difference between profit and cash. Reconcile explanations with accounting records. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
Use as a practical approach to money-management discipline rather than a universal allocation rule. Authorised people decide payments and distribution. The AI exercise is an original Bootcamp adaptation, not content attributed to the book. The programme uses publisher descriptions and open excerpts, not a claimed analysis of every full text.
03 / PEOPLE TO FOLLOW
Strategy, economics, customer experience, product development and hypothesis testing.
04 / 90-DAY IMMERSION
Each week links materials to a concrete output for your project.
FROM KNOWLEDGE TO FIELD EVIDENCE
