| name | analogous-sauce |
| description | Zoom out from the user's problem and find ideas by analogy — "you're thinking about app distribution, let's go look at drop-shipping / record labels / drug dealers / vending machines." Applies a battery of ideation lenses (different market, different era, different scale, biological, black-market, childhood, adjacent profession, reverse, …) to a user's stuck problem, picks 2–4 of the most generative analogies, then invokes the `get-sauce` skill against each analogous domain (and against the curated `design-thinking` topic for meta-process) to pull *real practitioner language* from the analogous world back into the user's problem. Use when the user is stuck on a problem, says "we need a different angle," "give me wildly different ideas," "what would X look like in Y," "how would [other industry] solve this," or wants to break out of category-bound thinking. NOT for execution-level questions (use get-sauce directly) or for "give me 5 variations" within the same frame (use a normal ideation prompt). |
Analogous Sauce
What this skill is for
The user has a problem and is stuck inside the frame of their own category. Same-industry inspiration produces same-industry ideas. The move that breaks this is analogical reasoning — find a structurally similar problem in a different domain, learn from how that domain solved it, and translate the pattern back.
Concrete examples of the move:
- App distribution → look at drop-shipping (logistics-light retail), record labels (taste-making distribution), vending machines (placement = destiny), drug dealers (zero-marketing word-of-mouth), Hollywood agents (talent representation as distribution).
- Onboarding a new user → look at fraternity rush week, military boot camp, a magician's opening trick, a therapist's first session, how casinos onboard whales.
- Pricing a SaaS → look at how 17th-century opera houses tiered seats, gym membership psychology, airline yield management, car-dealer financing, how the cartels structured drug pricing across distribution layers.
- Building a community → look at religious congregations, AA meetings, knitting circles, MMO guilds, dive bars, CrossFit boxes.
This skill is NOT "give me 5 variations of my idea" (that's stuck in the same frame). It's "give me 5 frames I haven't considered, grounded in what people in those frames actually said."
What this skill does, end-to-end
- Restate the problem so we're hunting for the right thing.
- Extract the underlying job/pattern — strip the user's category off.
- Apply a battery of lenses to that abstracted job to generate analogous domains.
- Pick 2–4 analogous domains with the user's input.
- Pull sauce from the
design-thinking topic to ground the analogy-making process itself.
- For each picked analogy, invoke
get-sauce against that analogous domain.
- Synthesize the cross-pollination — for each analogy, surface 3–6 transferable patterns with verbatim quotes from operators in the analogous domain, and translate each into the user's domain.
Workflow
Step 1 — Restate the problem and extract the pattern
Ask the user to confirm the problem in one sentence, then abstract it one level up — strip the noun-of-art off and re-express it as a generic job.
Examples of the abstraction move:
| User's framing | Abstracted job (the analogy hook) |
|---|
| "How do I distribute my app" | "How do I get my product into the hands of strangers at low marginal cost" |
| "How do I price my SaaS subscription" | "How do I extract differentiated value from a heterogeneous customer base for an intangible good" |
| "How do we keep retention up" | "How do I make people keep showing up to a thing they're not contractually required to attend" |
| "How do I get more App Store reviews" | "How do I get satisfied users to publicly testify, when their default is silence" |
| "How do I find my first 100 customers" | "How do I locate the small population of people who care most, before I've earned the right to broadcast" |
Confirm the abstraction with the user before proceeding. The abstraction is the lens-fitting: if it's wrong, every analogy will be wrong.
Step 2 — Apply the lenses
Run the abstracted job through each of the following lenses. For each lens, generate 2–4 candidate analogous domains. Cast wide on this pass — quantity over judgment. We narrow in Step 3.
The lens battery
-
Different-industry lens — what industry has the same structural job? Apply to every major sector: retail, finance, healthcare, education, media, fashion, food, real estate, energy, transport, hospitality, government, military, religion, entertainment, sports.
-
Different-era lens — how did people solve this job in 1900? 1500? Antiquity? Often the constraints were tighter, so the solutions were leaner. (Roman roads → CDNs. Medieval guilds → SaaS onboarding. Door-to-door encyclopedia salesmen → enterprise SDR motion.)
-
Different-scale lens — how does this job play out at three radically different scales? A single person solving it for themselves; a small team / village / shop; a planet-scale system (Google, the US gov, evolution). Each scale reveals different constraints.
-
Biological / natural lens — how does nature solve this job? Evolution, ecosystems, swarm behavior, immune systems, mycelium networks, predator-prey dynamics. "How does an ant colony do load balancing." Distant analogies from biology are unreasonably effective.
-
Adjacent-profession lens — how would a [profession outside the user's world] approach this? A magician, a journalist, a stand-up comedian, a chef, a sniper, an emergency-room nurse, a hostage negotiator, a librarian, a midwife, a tour guide, a Pixar story editor, a museum curator, a wedding planner.
-
Black-market / illegal-industry lens — illegal industries can't advertise, can't enforce contracts, can't use mainstream rails — so they evolve brutally efficient solutions to distribution, retention, trust, and pricing. Drug dealers, smugglers, counterfeiters, pirate streaming sites, illegal gambling. Treat as inspiration, not endorsement; the constraints are the lesson.
-
Childhood / amateur lens — how would a 7-year-old solve this? How would a hobbyist with no resources do it? Strips professional cruft. Often surfaces what's actually load-bearing.
-
Reverse / inversion lens (de Bono) — what's the opposite problem, and how is it solved? If the job is "retain users," the opposite is "make a user leave" — study churn from the attacker side. If the job is "drive growth," the opposite is "stay tiny on purpose" (cf. 37signals / Pieter Levels).
-
Game-design lens — what if this job were a video-game mechanic? What's the player's loop? Where's the reward schedule? What's the boss fight? Game designers have ruthless intuition for the engagement-and-difficulty curves underneath every behavioral problem.
-
Constraint-flip lens — what would solve this if [the resource you have most of] were suddenly zero, or [the resource you have least of] were infinite? "What if you had no engineers but infinite money." "What if you had no money but a 6-figure email list." Forces you to find a different solution path.
-
Genre-shift lens — re-imagine the problem as a different narrative genre. What's the horror-movie version of this onboarding? The romantic-comedy version? The Western? The murder-mystery? Surprisingly generative; uncovers tonal possibilities most product thinking misses.
-
Anti-domain lens — pick a domain that has the most opposite values to the user's, and look at how they handle the analogous job. (Luxury fashion ↔ Wal-Mart. Monastery ↔ Las Vegas. Hospital ICU ↔ stand-up comedy club. Often the contrast reveals what's truly load-bearing vs. just convention.)
Output the lens-pass as a structured grid:
ABSTRACTED JOB: <one-sentence job from Step 1>
LENS | CANDIDATE ANALOGIES
---------------------------+------------------------------------------------
Different industry | drop-shipping, record labels, vending machines, ...
Different era | door-to-door sales, medieval guild, ...
Different scale | personal homepage, neighborhood bulletin, Google index
Biological | seed dispersal, mycorrhizal networks, ...
Adjacent profession | librarian, hostage negotiator, magician
Black-market | drug dealers, smugglers, ...
Childhood / amateur | lemonade stand, school book fair
Reverse / inversion | how do I get UNINSTALLED → study churn attackers
Game-design | quest-giver mechanic, daily-login reward, ...
Constraint-flip | what if I had no engineers?
Genre-shift | onboarding-as-horror-movie, ...
Anti-domain | if the user is luxury → study Wal-Mart, etc.
You will not mine all of these — most are throwaway prompts. But generating the full grid first prevents premature commitment to the obvious analogies.
Step 3 — Pick 2–4 with user input
Surface the 5–8 most generative candidate analogies (your judgment; do not exhaust the user with the full grid). Use AskUserQuestion to let them choose 2–4 to actually mine.
Phrasing example:
Your stuck problem abstracts to: "how do I get my product into strangers' hands at low marginal cost." Here are 6 analogies that the lens-pass surfaced as most generative. I'd default to drop-shipping (industry) and drug dealers (black-market) — both are unreasonably good at "distribution-with-zero-marketing." Pick 2–4 to mine, or say "all" if you want a wide net.
- Drop-shipping
- Record labels (taste-maker distribution)
- Hollywood agents (talent representation)
- Drug dealers (word-of-mouth at scale)
- Vending machines (placement = destiny)
- Mycorrhizal networks (biological)
Recommend defaults explicitly. "All of them" is allowed but expensive — flag the cost.
Step 4 — Pull the meta-sauce on analogical thinking (one-time)
Before mining the user's chosen analogies, invoke the get-sauce skill against the design-thinking topic to load up on how to actually do this move well. This is the meta-pass — it grounds the rest of the workflow in real practitioner language about analogical reasoning, not your priors.
Concretely:
-
Read ~/moonlight/sauce/skills/get-sauce/references/topics/design-thinking.md directly. (It's curated for exactly this purpose.)
-
Spawn the get-sauce workflow with this question:
"How do experienced founders, designers, and thinkers actually use analogies from distant domains to solve their problems? Specifically: how do they pick which analogy to pursue, when does the analogy break down, and what's the move that turns 'cute observation' into 'real product decision'?"
-
Use Founders Podcast, Acquired, IDEO/d.school, and Brian Chesky talks as the default corpus (these are flagged in the topic file).
-
Run get-sauce's full loop (sauce-hunter → critic → optional augment).
The output of this pass is process sauce — quotes and tactics about how to make analogies work. Keep it in your context for the rest of the workflow; cite it in Step 7 when translating analogies into recommendations.
Don't skip this step even if you "know how analogies work." The user gets way more value when the analogy-translation move itself is grounded in Chesky-said-X / Munger-said-Y / IDEO-said-Z than in your priors. This is the same anti-pattern that motivates get-sauce in the first place — apply it to ourselves.
Step 5 — For each picked analogy, invoke get-sauce
Loop over the 2–4 analogies the user picked. For each one:
-
Treat the analogous domain as the actual sauce target. "Drop-shipping" is now the topic of the sauce hunt, not "app distribution."
-
Check references/topics/ first — there may already be a curated topic file for that domain. If not, fall back to grepping references/sources.md.
-
Compose a sauce-hunt question that's expressed in the analogous domain's vocabulary — not the user's. Examples:
| User's actual problem | Analogy | Sauce-hunt question (in analogy's vocabulary) |
|---|
| App distribution | Drop-shipping | "How do successful drop-shippers acquire customers at near-zero CAC?" |
| Onboarding | Magician's opening trick | "What do master magicians say about the structure of a great opener?" |
| Pricing SaaS | Opera-house seating | "How did opera houses historically segment audiences via seat tiering?" |
| Retention | AA meetings | "What do AA organizers say about why people keep showing up?" |
-
Invoke get-sauce end-to-end against that question. If a relevant topic file exists, use it; otherwise grep the master registry; otherwise the user may need to nominate a source.
-
If no sauce corpus exists for the analogous domain in the registry, do one of:
WebSearch + WebFetch for primary-source interviews/essays in the analogous domain
- Ask the user if they know a podcast/book/operator in that domain
- Skip and note as a limitation. Don't fall back on LLM priors about that domain — that defeats the whole skill.
-
Note any new sources discovered and consider contributing them back to the registry per get-sauce's "How to contribute" section.
Step 6 — For each analogy, build the translation table
For each mined analogy, build a translation table mapping patterns from the analogous domain back to the user's domain:
ANALOGY: Drop-shipping
ABSTRACTED JOB: low-CAC distribution of products you don't make yourself
PATTERN IN ANALOG | QUOTE | TRANSLATES TO IN USER'S DOMAIN
--------------------------------+------------------------------------------+----------------------------------
"Winning product > winning ad" | <Operator quote, file cite> | "Your paywall conversion ceiling
| | is set by the product, not
| | the funnel"
"Test 50 SKUs in week 1" | <Operator quote, file cite> | "Ship 50 onboarding variants
| | in week 1, not 5"
"The customer is the algorithm" | <Operator quote, file cite> | "TikTok organic is the new
| | AliExpress storefront"
... | ... | ...
Three rules:
- The pattern must be a verbatim quote, not your summary of the operator. Cite the file (per
get-sauce discipline).
- The translation must be defensible — explicitly check: what's the same about the situation? What's different? Does the difference invalidate the analogy?
- Flag where the analogy breaks down. Every analogy fails somewhere. Saying "this maps cleanly except for X, where the user's domain has constraint Y that drop-shippers don't" is more useful than a perfect-sounding translation.
Step 7 — Synthesize and relay
Compose the final answer:
- One-paragraph framing: what the abstracted job was, which analogies you mined, why those.
- Per-analogy section: the translation table (Step 6), plus a "what specifically I'd try first" 2–3 bullet recommendation drawn from the operator quotes.
- Meta-synthesis: across all analogies, what converging patterns showed up? (Convergence across distant domains is the strongest signal that you've hit a real underlying structure.) What contradicting patterns? (Contradictions are forks the user has to choose between.)
- Process meta-note (from Step 4): one or two quotes from Chesky / Munger / IDEO / Senra about how to evaluate whether the analogy actually applies — give the user the tool to second-guess your translation.
- Where the analogies break: which translations the user should not trust, and why.
Don't re-summarize operators' words — quote them. The framing is a wrapper around their actual language.
When to use this skill vs. get-sauce directly
| Situation | Use |
|---|
| "How are other founders in my exact space doing X?" | get-sauce directly |
| "What's the playbook on paywall design for apps?" | get-sauce directly |
| "I'm stuck — give me a totally different angle on this problem" | this skill |
| "What's [my problem] look like through the lens of [some other industry]?" | this skill |
| "Help me ideate without being trapped in my own category" | this skill |
| "Give me 5 variations of my idea" | Neither — that's a regular ideation prompt |
The diagnostic: if you can name a podcast/founder who's literally solved this exact problem, use get-sauce. If the user needs re-framing before they even know what to look for, use this.
Anti-patterns
- ❌ Skipping Step 1's abstraction. If you don't strip the user's category off, every analogy will be a near-neighbor and you'll generate same-industry ideas in costume. The abstraction is the move.
- ❌ Picking near analogies. "App distribution is like website distribution" — too close. The Innocentive research is unambiguous: distant domains outperform near ones for novel ideation. Push further.
- ❌ Generating analogies but not mining them. A list of "we should look at how X does it" is not the deliverable. The deliverable is operator quotes from domain X translated into the user's domain. That requires
get-sauce-grade source mining for each analogy.
- ❌ Falling back on LLM priors about the analogous domain. If you don't have real source material on drop-shipping, you'll generate generic "drop-shipping intuitions" — which is exactly what
get-sauce exists to prevent. Either find primary sources or flag the gap.
- ❌ Pretending the analogy is perfect. Every analogy fails. Naming the failure mode is part of the value.
- ❌ Stopping at one analogy. Convergence across 2–4 distant analogies is the strongest signal that the underlying structure is real. One analogy is just a clever observation.
- ❌ Letting the user pick "all 12 lenses". This is a bait-and-switch; you'll generate shallow translations for all of them. Push back: 2–4 deep beats 12 shallow.
Repo hygiene
This skill is dependent on the get-sauce skill in the same repo (skills/get-sauce/). It assumes:
skills/get-sauce/SKILL.md is the canonical sauce-hunting workflow.
skills/get-sauce/references/topics/design-thinking.md is the meta-process source (Step 4).
skills/get-sauce/references/topics/*.md and skills/get-sauce/references/sources.md are the discovery layers for each analogy in Step 5.
If get-sauce is missing, this skill cannot run — say so, link the user to install it, and stop.
Further reading
skills/get-sauce/SKILL.md — the underlying sauce-hunt workflow this skill calls.
skills/get-sauce/references/topics/design-thinking.md — the curated meta-sauce topic on analogical reasoning, lateral thinking, and design-thinking method.
- Edward de Bono, Lateral Thinking — the canonical primer on the kind of move this skill operationalizes.
- David Epstein, Range, ch. 4–5 — the empirical case for distant-analogy ideation.
- Steven Johnson, Where Good Ideas Come From — "the adjacent possible" and "exaptation."
- Brian Chesky's talks on Airbnb's cross-industry inspiration ("design Airbnb like a Disney Cruise") — operationalized analogical thinking by a working founder.