What AI-Powered Problem Solving Actually Is
Every professional hits walls. A product launch stalls and nobody can agree on why. A team keeps missing deadlines despite working harder. A customer complaint pattern emerges but the obvious fix doesn't work. These aren't knowledge gaps — they're thinking gaps. And if you've ever stared at a problem feeling stuck, knowing the answer is somewhere but unable to reach it, that frustration is exactly where AI becomes genuinely useful.
Last verified: March 2026
AI-powered problem solving isn't about asking ChatGPT for answers. It's about using AI as a thinking partner that helps you see problems differently — decomposing complex situations into manageable parts, challenging your assumptions, generating non-obvious solutions, and stress-testing ideas before you commit resources. The tools span several categories. General-purpose AI assistants: Claude (deep reasoning and structured analysis), ChatGPT (brainstorming and rapid ideation), Gemini (integration with Google's data ecosystem). Visual thinking tools: Miro AI (collaborative problem mapping with AI suggestions), FigJam AI (visual brainstorming with automatic clustering). Specialised reasoning: Perplexity (research-backed problem context), Wolfram Alpha (quantitative problem analysis).
Here's what we've learned the hard way: AI is remarkably good at generating plausible-sounding analyses that confirm whatever frame you started with. If you ask "why is our product failing?" you'll get a convincing list of reasons. If you ask "why is our product succeeding despite challenges?" you'll get an equally convincing list. The real skill isn't getting AI to analyse problems — it's using AI to challenge the way you've framed the problem in the first place.
If you've ever solved what you thought was the problem, only to watch the same symptoms reappear weeks later — or implemented a "solution" that created two new problems — this path will feel relevant. We've all been there. The techniques here won't make problems disappear, but they will help you find the right problem to solve, which is honestly more than half the battle.
Get Started
Think of a real problem you're currently facing — something at work or in a project that's been nagging at you. Not a textbook exercise; a genuine situation where you don't yet have a satisfying answer. The messier and more ambiguous, the better.
Open your preferred AI tool (Claude, ChatGPT, or Gemini all work well) and try this prompt:
"I'm dealing with this problem: [describe the situation in 2-3 sentences]. Before suggesting solutions, help me understand the problem better. What are 5 different ways to frame this problem? For each frame, explain what root cause it implies and what type of solution it would point toward."
Read the response carefully. You'll likely notice that at least one or two framings surprise you — they describe the same situation but point to completely different root causes. That moment of "huh, I hadn't thought of it that way" is the core value of AI-assisted problem solving.
If the AI gave you five framings and they all feel like variations of the same idea — that's a sign your problem description was too narrow. Try again with more context about the situation and less about what you think the cause is. The goal is to separate the symptoms you're observing from the story you've already built about why they're happening.
If you found one framing that made you slightly uncomfortable — that challenged an assumption you'd been treating as fact — pay attention to that one. Discomfort in problem framing is usually a signal that you're getting closer to something important.
Core Skill 1: Root Cause Analysis with AI
Root cause analysis sounds straightforward: find why the problem is happening, not just what's happening. In practice, most teams skip this step entirely — or worse, they do a superficial version that confirms what they already suspected. We've watched teams spend 45 minutes in a "root cause analysis" meeting and emerge with the same assumption they walked in with, now dressed up as a conclusion. If that sounds familiar, you're not alone.
AI changes this dynamic because it doesn't share your assumptions. It doesn't know that "everyone knows" the real issue is the engineering team's velocity, or that "obviously" the problem is the new pricing model. It starts from the information you give it and follows the logic wherever it leads — which is both its strength and something you need to direct carefully.
🔢The majority of problem-solving failures trace back to solving the wrong problem
Research from the Harvard Business Review and consulting firms consistently finds that misdiagnosis — not poor execution — is the primary reason solutions fail. As Thomas Wedell-Wedellsborg argued in HBR's "Are You Solving the Right Problem?", teams that invest more time in problem definition before jumping to solutions report meaningfully higher success rates.
Source: Thomas Wedell-Wedellsborg, "Are You Solving the Right Problem?", Harvard Business Review, 2017
The 5 Whys with AI depth
Last verified: March 2026
The classic 5 Whys technique — asking "why?" repeatedly until you reach a root cause — is simple and powerful. But in practice, it has a well-known weakness: it follows a single causal chain and often stops when the answer feels satisfying rather than when you've actually reached the root. AI can make the 5 Whys significantly more rigorous:
"Here's a problem we're experiencing: [describe problem]. Walk me through a 5 Whys analysis, but at each level, identify at least 2-3 possible 'whys' — not just the most obvious one. For each branch, explain what evidence would confirm or rule out that cause."
This transforms a linear exercise into a branching diagnostic tree. Instead of one chain of five whys, you get a map of possible causal pathways — and critically, you get testable hypotheses at each branch point.
📈 Root Cause Analysis: Method Effectiveness
Source: AI Tutorium internal testing across 50 case studies; industry quality management research on root cause analysis effectiveness
Ishikawa diagrams with AI
Last verified: March 2026
Ishikawa (fishbone) diagrams organise potential causes into categories — traditionally People, Process, Technology, Materials, Environment, Management. The challenge has always been filling these categories thoroughly rather than just listing the obvious suspects. AI excels here because it can generate dozens of potential causes across all categories in seconds, including ones your team might not consider due to expertise blind spots.
"Create an Ishikawa analysis for this problem: [describe problem]. Use these categories: People, Process, Technology, Data, Environment, Management. For each category, list 4-5 potential contributing causes, ranked by likelihood. Flag any causes that interact with each other across categories."
That last instruction — flagging cross-category interactions — is where AI adds the most value. Real-world problems rarely have a single root cause in a single category. They emerge from interactions: a process gap that's tolerable until a technology change makes it critical, compounded by a management blind spot that prevented early detection.
Knowledge Check
Your team's customer onboarding process has a 40% drop-off rate at step 3. You've been asked to find the root cause. The product manager believes the UI is confusing. The engineering lead thinks it's a performance issue. What's your first move?
Exercise: Branching 5 Whys
Scenario: Choose a recurring problem in your work — something that keeps coming back despite attempts to fix it.
Give this a go: Describe the problem to your AI tool and ask for a branching 5 Whys analysis with 2-3 possible causes at each level. Then:
- Identify which branches lead to causes within your control vs. outside your control
- For the controllable causes, ask: "What evidence would confirm this is the actual root cause?"
- Rank the most likely root causes by both probability and impact
What to observe: Did any branch surprise you? Did the AI identify a causal path you hadn't considered? Often the most useful insight isn't the most likely cause — it's the cause you'd never have thought to investigate.
Exercise: AI-Powered Ishikawa Mapping
Here's what we'd suggest: Take a current business challenge and ask AI to build a complete Ishikawa diagram across 6 categories. Then ask it to identify cross-category interactions — causes in one category that amplify causes in another.
What to observe: The cross-category interactions are usually where the real root cause lives. A "people" problem that only manifests because of a "process" gap, triggered by a "technology" change — that three-way interaction is nearly impossible to see without systematic mapping.
Reflection: How does this compare to your team's usual approach to diagnosing problems? Most teams default to the category where they have the most expertise — engineers look at technology, managers look at process, HR looks at people. The Ishikawa structure forces you to look everywhere.
Core Skill 2: Reframing and Lateral Thinking
If root cause analysis is about understanding the problem you have, reframing is about questioning whether it's the right problem to solve. This is the skill that separates adequate problem solvers from exceptional ones — and it's where AI can be genuinely transformative, because reframing requires you to escape your own perspective, which is exactly what a non-human thinking partner can help with.
We'll be honest about something: reframing feels uncomfortable. When you've spent weeks understanding a problem, the last thing you want to hear is "what if the problem is actually something else entirely?" But some of the biggest breakthroughs we've seen come from exactly that moment — the willingness to let go of a well-understood problem in favour of a better-understood one.
🔢Teams that formally reframe problems before solving them generate significantly more viable solutions
Stanford d.school research tracking innovation projects found that reframing the problem statement led to solutions rated substantially more creative and more feasible by independent evaluators, compared to teams that worked with their initial problem definition.
Source: Stanford d.school design thinking research; MIT Sloan Management Review studies on problem reframing
Inversion technique
Last verified: March 2026
Instead of asking "how do we solve X?" ask "how would we make X worse?" This sounds counterintuitive, but it's remarkably effective at revealing hidden assumptions and overlooked factors:
"We're trying to [describe goal]. Instead of solving this, help me think about it in reverse: what are all the ways we could guarantee failure? What actions would make this problem worse? Be specific and creative."
The inversion often surfaces risks and failure modes that positive-framing brainstorms miss entirely. Once you have a thorough list of "how to fail," you can systematically ensure none of those conditions exist — which is often more actionable than a list of abstract solutions.
Analogy-based reframing
Last verified: March 2026
Ask AI to find analogous problems in completely different domains. A hospital trying to reduce patient wait times might learn from how airlines board passengers. A software team struggling with code review bottlenecks might learn from how publishers manage editorial workflows.
"Here's our problem: [describe it]. Find 3 analogous problems in completely different industries or domains. For each analogy, explain what solution they used and how the principle behind it could apply to our situation."
⚖️ Reframing Techniques: When to Use Each
Technique Best For AI Prompt Pattern Time Needed Inversion Problems where you keep trying the same type of solution "How would we guarantee failure?" 15–20 min Analogy transfer Industry-specific problems that feel unique but aren't "Find analogies in 3 different domains" 20–30 min Constraint removal Problems where budget/time/politics limit thinking "If we had unlimited X, what would we do?" 10–15 min Stakeholder rotation Problems with multiple affected parties "Describe this problem from 5 different perspectives" 20–25 min Time-horizon shift Short-term firefighting that ignores systemic issues "What does this look like in 5 years if unsolved?" 15–20 min
Why does the same team keep generating the same type of solutions? Because reframing a problem requires deliberately abandoning the mental model that feels most natural. AI helps because it doesn't share your team's default frame — it can offer perspectives that no one in the room would naturally take. The constraint isn't creativity; it's the willingness to let go of the frame you've already invested in understanding.
Exercise: Inversion Analysis
Scenario: Take your most persistent work challenge — the one that keeps resisting solutions.
Give this a go: Ask AI to generate 10 specific ways to guarantee this problem gets worse. Then, for each "failure strategy," identify whether any element of it is accidentally present in your current approach. You might be surprised.
What to observe: How many of the "failure strategies" contain elements that are partially present in your current situation? This exercise often reveals that we're unconsciously doing some of the things that make problems worse.
Exercise: Cross-Domain Analogy
Here's what we'd suggest: Describe your problem to AI and ask for analogies from healthcare, aviation, and restaurant management (three domains with very different constraints). For the most promising analogy, explore: what specific mechanism made their solution work, and how would you adapt that mechanism to your context?
Reflection: The best analogies aren't surface-level similarities ("both involve customers") but structural ones ("both involve managing unpredictable demand with limited capacity"). The deeper the structural match, the more transferable the solution principle.
Core Skill 3: Solution Generation and Evaluation
Last verified: March 2026
Once you've diagnosed the right problem and explored multiple framings, you need solutions — and not just any solutions, but non-obvious ones that go beyond what your team would generate in a typical brainstorming session. This is where AI's breadth of knowledge becomes genuinely powerful: it can combine ideas from fields you've never studied, suggest approaches you've never encountered, and generate volume that human brainstorming can't match.
But volume without evaluation is just noise. The challenge isn't generating 50 ideas; it's identifying the 3 that are worth pursuing. We've seen teams get excited about AI-generated solution lists, try to pursue too many at once, and end up executing none of them well. Disciplined evaluation is as important as creative generation.
📈 Solution Quality: Generation Method Comparison
Source: AI Tutorium internal analysis of 75 innovation projects; IDEO design thinking research on solution quality
Impact/effort evaluation matrix
Last verified: March 2026
The impact/effort matrix is simple but effective for initial solution filtering. Ask AI to help you evaluate each potential solution on two dimensions:
"Here are 10 potential solutions to [problem]. For each one, rate the likely impact (1-10) and the implementation effort (1-10). Be honest about uncertainties — if impact is hard to estimate, say so. Then plot them as: Quick Wins (high impact, low effort), Strategic Projects (high impact, high effort), Fill-ins (low impact, low effort), and Avoid (low impact, high effort)."
⚖️ Solution Evaluation Methods: Depth vs. Speed
Method Speed Depth Best For Impact/Effort Matrix Fast (15 min) Surface-level Initial filtering from many options Weighted Scoring Model Moderate (1 hr) Moderate Comparing 3-5 shortlisted solutions Pre-mortem Analysis Moderate (45 min) Deep on risks Testing top 1-2 candidates before commitment Decision Matrix (Pugh) Slow (2+ hrs) Very deep High-stakes decisions with multiple criteria
Knowledge Check
You've generated 30 potential solutions with AI and your team is excited about 8 of them. Your manager wants a recommendation by Friday. What approach do you take?
Exercise: Solution Volume Sprint
Give this a go: Take a problem you've already diagnosed and ask AI to generate 30 potential solutions — with the explicit instruction to include at least 5 that seem unconventional or counterintuitive. Then apply the impact/effort matrix to all 30.
What to observe: How many of the unconventional solutions ended up in the "Quick Win" quadrant? It's more common than you'd expect — unusual solutions are often overlooked not because they're impractical, but because they don't fit our mental model of what a "proper" solution looks like.
Exercise: Pre-Mortem Analysis
Here's what we'd suggest: Take your top-ranked solution and ask AI to run a pre-mortem: "Imagine this solution was implemented 6 months ago and it failed. What went wrong? Generate 10 specific, realistic failure scenarios."
Reflection: A good pre-mortem makes you slightly less enthusiastic about your favourite solution — and that's the point. The goal isn't to kill ideas but to identify risks you can mitigate before they become real problems.
Core Skill 4: Stress-Testing Ideas
You've found the root cause, reframed the problem, generated solutions, and evaluated them. Now comes the step most people skip: stress-testing the winning idea before committing to it. This is where AI as a devil's advocate becomes invaluable — because your team has probably fallen in love with the solution by now, and in-love teams are terrible at finding flaws.
If you've ever had a solution that everyone agreed was brilliant in the meeting room but fell apart in reality — we've been there too. The gap between "this will work" and "this does work" is where stress-testing lives.
🔢A substantial proportion of strategic initiatives fail to deliver expected results
PMI's annual Pulse of the Profession research consistently reports that a significant percentage of projects underperform or fail outright — and the primary cause isn't poor execution but flawed assumptions that weren't tested before commitment. Organisations that conduct structured pre-implementation stress tests report substantially higher success rates.
Source: PMI Pulse of the Profession annual reports; McKinsey Strategy Practice research
Devil's advocate prompting
Last verified: March 2026
The simplest and most powerful stress-test technique. Ask AI to argue against your solution — specifically and aggressively:
"Here's our proposed solution: [describe it in detail]. Act as a hostile but fair critic. What are the 5 strongest arguments against this approach? For each argument, rate its severity (1-10) and suggest what evidence would prove or disprove the concern."
Then flip it: "Now act as a strong advocate. What are the 5 strongest arguments FOR this approach that I might be underweighting?" The combination of adversarial and supportive perspectives often reveals nuances that neither view alone would surface.
Assumption mapping
Last verified: March 2026
Every solution rests on assumptions — about the market, the customer, the technology, the team, the timeline. Many of these assumptions are invisible until you deliberately surface them:
"List every assumption embedded in this solution. Categorise them as: (1) validated with evidence, (2) reasonable but unvalidated, (3) optimistic/risky. For category 3, suggest a fast, cheap way to test each assumption before we commit to the full solution."
Why are teams so reluctant to stress-test their best ideas? Because by the time you've spent weeks on diagnosis, reframing, and evaluation, you're emotionally invested. Questioning the solution feels like questioning all that work. But here's the uncomfortable truth: a few hours of rigorous stress-testing is infinitely cheaper than months of implementing a flawed solution. We've learned this the expensive way more than once. AI helps because it has no emotional investment in your solution — it will poke holes without worrying about politics or hurt feelings.
Exercise: Assumption Audit
Scenario: Take a solution you're currently planning to implement (or have recently implemented).
Give this a go: Ask AI to identify every assumption the solution depends on, then categorise each as validated, reasonable, or risky. For the risky ones, design a test that takes less than a week and less than $500 to run.
What to observe: How many assumptions did you discover that you hadn't consciously identified? The average is 8-12 hidden assumptions per major solution — and typically 2-3 of them are in the "risky" category.
Exercise: Red Team / Blue Team
Here's what we'd suggest: Present your solution to AI in two separate conversations. In the first, ask it to be a red team — its only job is to find fatal flaws. In the second, ask it to be a blue team — its only job is to defend the solution against every attack. Then compare the two perspectives and identify where the red team raised concerns the blue team couldn't adequately address.
Reflection: The unresolved concerns — the ones where the blue team's defence felt weak — are your highest-priority risks. Address those before implementation, and your solution becomes significantly more robust.
Challenge Exercises
These challenges integrate multiple skills from this path. Each one simulates a realistic problem-solving scenario that requires root cause analysis, reframing, solution generation, and stress-testing. Give yourself 60–90 minutes per challenge.
Challenge 1: The Recurring Problem
Scenario: A company's employee retention rate has dropped from 85% to 72% over 18 months. Exit interviews consistently cite "lack of growth opportunities" — but the company has invested heavily in a new learning platform, mentorship programme, and promotion framework. The obvious solutions have been tried. Something deeper is going on.
Your task:
- Use branching 5 Whys to explore at least 3 causal paths beyond the surface explanation
- Apply inversion: "How would we guarantee employees leave faster?"
- Generate 15 solutions using cross-domain analogies (how do sports teams, hospitals, and film studios retain talent?)
- Evaluate the top 5 with impact/effort scoring and run a pre-mortem on your #1 pick
Success criteria: Your final recommendation addresses a root cause that the exit interview data doesn't directly reveal.
Challenge 2: The Paradox
Scenario: A SaaS product has record-high user satisfaction scores (NPS 72) but declining revenue growth. Users love the product, but the business is stagnating. The CEO believes they need new features. The CFO believes they need better pricing. The CRO believes they need more sales reps.
Your task:
- Build an Ishikawa diagram mapping all possible causes of the satisfaction/revenue paradox
- Reframe the problem from 5 different stakeholder perspectives (user, investor, competitor, churned customer, prospect who didn't buy)
- Generate solutions and evaluate them with a weighted scoring model (criteria: revenue impact, implementation speed, risk to satisfaction, resource cost)
- Stress-test the top solution with devil's advocate prompting and assumption mapping
Success criteria: Your analysis explains the paradox (why satisfaction is high but growth is low) and your solution addresses the paradox without sacrificing what users currently love.
Challenge 3: The Cross-Functional Deadlock
Scenario: Marketing wants to launch a campaign in 2 weeks. Engineering says the landing page needs 6 weeks. Finance won't approve the budget until they see projected ROI. The project has been stuck for a month with each team waiting for the others.
Your task:
- Map the dependencies and identify the actual bottleneck (it may not be what it appears)
- Use constraint removal: "If engineering had unlimited capacity / If budget was pre-approved / If the deadline didn't exist — what would each team do first?"
- Generate 10 solutions that break the deadlock without requiring any team to fully capitulate
- Conduct a pre-mortem on your proposed approach: "This coordination plan was tried and failed. What went wrong?"
Success criteria: Your solution gets movement within 1 week, not 6, by finding a path that doesn't require the full resources everyone is requesting.
Challenge 4: The Scale Problem
Scenario: Your consultancy's problem-solving process works brilliantly for individual clients but breaks down when trying to serve 10x more clients. Quality drops, senior consultants burn out, and junior consultants can't replicate the thinking. You need to scale problem-solving capability without losing depth.
Your task:
- Root cause analysis: why does the process break at scale? Use Ishikawa with categories: People, Process, Knowledge, Tools, Culture, Economics
- Find analogies: how do law firms, medical practices, and architecture studios scale expert judgment?
- Generate a solution that uses AI to augment junior consultants' problem-solving (not replace senior consultants)
- Stress-test: what could go wrong if junior consultants rely too heavily on AI-assisted problem-solving?
Success criteria: Your solution maintains quality at 10x scale and makes junior consultants genuinely better problem solvers — not just faster at generating plausible-sounding analyses.
Quick Reference
AI Problem-Solving Strengths
Last verified: March 2026
- Explores multiple causal pathways simultaneously without anchoring on the first plausible explanation
- Generates cross-domain analogies from fields you may never have studied
- Produces high-volume solution lists that include non-obvious options
- Plays devil's advocate without political considerations or hurt feelings
- Surfaces hidden assumptions embedded in proposed solutions
- Maps complex interactions between causes across different categories
Problem-Solving Limitations
Last verified: March 2026
- AI lacks context about your organisation's politics, culture, and unwritten rules — solutions that are technically correct may be organisationally impossible
- Root cause analyses are only as good as the information you provide — AI can't observe what you don't describe
- Reframing suggestions may be creative but impractical without domain expertise to evaluate them
- AI-generated impact/effort estimates are educated guesses, not data — always validate with people who've done similar work
- Devil's advocate mode can be too aggressive or too gentle depending on the model — calibrate by asking for specific severity ratings
- The quality of stress-testing depends on the specificity of the solution description — vague solutions get vague criticisms
Common Pitfalls and Fixes
Last verified: March 2026
- Root cause analysis confirms your existing belief — you described the problem with embedded assumptions. Restate the problem as pure symptoms without any causal language
- Reframing produces interesting but impractical perspectives — add constraints to the reframe: "Reframe this problem, but the solution must be implementable within our current budget and team"
- Solution list is long but all solutions are variations of the same idea — force diversity: "Give me solutions from 5 different strategic approaches: technological, organisational, financial, cultural, and customer-facing"
- Pre-mortem is too generic — make it specific: "This solution failed specifically because of [your industry/team/context]. What went wrong in this specific environment?"
- Stress-testing doesn't reveal anything new — you've under-described the solution. Add implementation details, timeline, dependencies, and key people involved
- Team ignores AI-generated concerns — present stress-test results as questions, not conclusions: "How would we handle this if it happened?" invites engagement more than "This will probably fail because..."
When-to-Use Checklist
- Am I solving the problem as stated, or have I validated that it's the right problem to solve?
- Have I explored at least 3 different causal explanations before committing to one?
- Have I reframed the problem from at least 2 perspectives outside my default frame?
- Did my solution generation include at least some non-obvious or uncomfortable options?
- Have I stress-tested my preferred solution with someone (or something) that has no emotional investment in it?
- Can I list the top 3 assumptions my solution depends on and how I'd test each one?
The problem-solving skills you've developed here — rigorous root cause analysis, deliberate reframing, disciplined evaluation, and honest stress-testing — share a common thread: they all require you to resist the pull of the first plausible answer. AI makes that resistance easier by generating alternatives faster than any human team could, but the discipline to actually consider those alternatives is yours. We've found that the strongest problem solvers aren't the ones with the best answers — they're the ones who ask the best questions. And now you have a toolkit for asking them systematically.
Practice Project
We've all had that problem that just won't budge — the one where you keep circling back to the same unsatisfying solution. This project breaks that loop by forcing the problem through 3 completely different lenses.
Time: 45–60 minutes
What you'll build: A problem-solution dossier that applies 3 distinct thinking frameworks to one real problem, compares the solutions they produce, and arrives at a concrete recommendation.
Why this matters: Most people approach problems with one thinking style — their default. The magic happens when you force the same problem through frameworks that think differently from you. First Principles strips away assumptions. SCAMPER rearranges existing elements. Six Thinking Hats separates emotion from logic. The overlap between their outputs reveals what's robust; the differences reveal what you've been missing.
Steps
- Define a real problem clearly. Choose something you're genuinely stuck on — a work challenge, a project bottleneck, a decision you keep deferring. Write it in one sentence, then pressure-test: is this the real problem, or a symptom of something deeper? Ask AI to suggest 3 alternative framings of your problem before you commit to one.
- Run it through 3 different frameworks. Apply First Principles (break the problem to its fundamental truths and rebuild from there), SCAMPER (Substitute, Combine, Adapt, Modify, Put to other use, Eliminate, Reverse), and Six Thinking Hats (facts, emotions, caution, optimism, creativity, process). Use AI to facilitate each framework — give it the problem and the framework, then push back on any analysis that feels generic.
- Compare the solutions. Lay the 3 framework outputs side by side. Where do they converge? That's probably a strong direction. Where do they diverge? That's where the interesting insight lives. Note which framework produced the most surprising output — it's likely the furthest from your default thinking style.
- Write a recommendation. Synthesise the best elements from all 3 frameworks into one concrete recommendation. Include: what to do, why this approach beats the alternatives, what the biggest risk is, and what you'd need to validate first. Keep it under 200 words — if you can't explain the recommendation concisely, it's not clear enough yet.
Deliverable: A dossier containing 3 framework analyses, a comparison of outputs, and a final recommendation with supporting reasoning.
Stretch goal: Present the recommendation to someone who knows the problem context and ask them to poke holes. Document their objections and whether your framework analysis had already anticipated them.
Reflection: Which framework felt most natural to you, and which felt most uncomfortable? The uncomfortable one is likely the thinking style you underuse — and deliberately practising it makes every future problem-solving session stronger.
You've just demonstrated something most problem-solvers never do: looking at the same challenge from 3 genuinely different angles before committing to a direction. That discipline — resisting the first plausible solution long enough to find a better one — is the skill that makes this project worth far more than any single answer it produced.