Welcome to Language Agnostic, the blog of Inaimathi! It's now built on top of the Clojure framework http-kit. I'm reasonably confident it's no longer the least performant blog on the internet.

Enjoy the various programming-themed writings availble on offer. The latest post is available below and the archive link is directly above this text.

Antler - Day 6

Sat Apr 6, 2024

There's a birthday on the team today. (Happy Birthday, Alex Wright! You were literally at work; hopefully it was a fun day.)

Today's schedule

  • Idea generation
  • Idea workshop
  • Group work
  • Huddle Feedback
  • Business Canvas with Alan Smith (optional keynote; I think I'm gonna wanna see this)
  • Family Dinner
  • Throughout the day, we've got office hours.
  • Reminder: fill out Week 1 feedback survey if you haven't already.

Bernie's Intro for Idea Generation

  • Sort of like last time, with the design sprint, but smaller groups
  • Framework to identify important problems (due to the Waterloo Problem Lab)
    1. Start looking for important problems in domains you find personally interesting.
    2. Document as many examples of important ideas as you can find.
    3. Select several problems for further investigation.
    4. Analyze the problem by scale, context, history and failures a. Scale of the problem
      1. Who and how are users affected?
      2. What is the priority level of the problem? b. Context of the problem
      3. What are the root causes of the problem?
      4. What circumstances or conditions affect the problem? c. Research the history of the problem
      5. How long has the problem been recognized?
      6. Does the problem appear to be growing in importance?
      7. Has the scale of the problem changed?
      8. Has there been a change in those affected by the problem?
      9. Have the causes or effects of the problem changed?
      10. Have the circumstances and conditions affecting the problem changed over time?
      11. has the primacy of the problem changed over time?
      12. Have there been previous attempts to solve the problem? d. Analyze Past Failures and Attempts
      13. Why did the attempts fail?
      14. Identify actionable mistakes?
  • Something called the fishbone model helps here.
  • Explore the surface area of the problem. Don't get fixated on the first local maximum you see, chart out enough of the space that you can be fairly sure you're mining something like the current global maximum.

Self Inventory Exercise

  1. Knowledge - what was the focus of your education or career?
  2. Capability - what are you most proficient at?
  3. Connections - who do you know that has expertise in different industries?
  4. Financial Assets - What access to financial capital do you have?
  5. Name recognition - What are you well known for?
  6. Past work experience - Go through every important company you've ever worked at. What have you learned that others don't know?
  7. Passion for a market - Does the idea of improving climate tech or healthcare excite you?
  8. Commitment - Do you have the time and effort to devote to this endeavour?
  1. Distributed systems, type systems (lol), communication protocols
  2. Software development/tool building
  3. Lots of connections in the Toronto dev/AI scene
  5. Coding and blogging about it.
  6. Too much to go over in under like 30 minutes
  7. Toolsmithing; I like building enabler technologies.
  8. Yes. No qualifications.

Different angle

  • Market Edge
    • Where have you worked?
    • What did you learn about an industry that would surprise an outsider?
    • Have you been exposed to the idiosyncrasies and challenges of a market or industry?
  • Tech Edge
    • Pretty self explanatory.
    • What tech are you peak good with? What gives you any kind of development/hardware/etc advantages?
  • Catalyst Edge
    • You as an operator
    • When have you quickly assimilated into a new industry?
    • When have you quickly built and commercialized prototype products?

Share your edges. Use it as a starting point for exploration.

  • Articulate your beliefs
    • Can you arrive at a joint, core belief as a cofounding team?
    • Do you share similar views about what the future might look like?
  • Thought starter Qs
    • Combining your edges, what do you believe could be different about a market, industry or the world in five years' time?
    • How do you believe people or companies will behave differently in the future because of your belief?
    • Will that vision fuel you both for the next 5-10 years?

Study what's happening on the Frontier

  • Where is the edge of technology?
  • Where is the edge of industry roadmaps?
  • What are bleeding edge startups building?
  • What are interesting regulatory changes/obstacles in spaces you find interesting?
  • What are geographic arbitrage opportunities you could take advantage of? (If it's been done in US/Canada/UK, can you do it in the other two)

The idea maze

  • A "good idea" is a detailed path through the maze. Why does your path lead to treasure: competitor oversight, new technology or something else?
  • A way to systematize the space of an idea in a way to figure out where good ideas are. The example flowchart is for "I've got an idea for doing music/movies on the internet".
    • Fork points are "open/closed source", "free/not", "streaming/vod", etc
    • Some endpoints have companies in them; Netflix, Youtube, Kazaa/limewire/napster, iTunes are the ones I remember off the top of my head.

Force Constraints

What is a company that can be worth $1B, that can be built in seven years using less than $50M of total capital.

  • What does $1B of equity look like? Revenue/profitability targets? # users/customers?
    • How far into the future does mainstream market adoption happen?
      • For example, culturing meat/fat in the lab. One question they'll have to answer from investors is: how far out in the future is customer acquisition/moneymaking?
    • How capital intensive is the production/solution and customer acquisition?
      • One of the reasons B2C is tough today; high cost of customer acquisition (because competition is huge unless you have a potential network built out already. This is why marketing/sales is really intense here)

Bernie's Q&A

  • If you're starting with a horizontal tech, and you're looking for a vertical, what do you think of a "spray-n-pray" approach?
    • You have to have a starting point. So like, just pick one you're passionate about. You need a vertical to focus on and crack it, if you "spray-n-pray", things get diluted very quickly. Especially if you're starting with like two people. Thiel asks: What's a space where you can niche down repeatedly to a small enough sliver of the market that you can monopolize it. And once you've done that, you can grow your monopoly out.
  • Given that we're looking for $1b in 7 years, do we want to pick a niche where $1b is directly possible?
    • No, your first market doesn't need to take you all the way to $1b. The main reason you're doing this is to paint a picture for the investor. Because if you're not looking for VC, you don't really need to do this.
  • We have to aim on a problem solution and dominating a market. How do we think about roadblocks for competitors?
    • Yeah, you're talking about the classic moat question. One of the questions you're going to get about your company is "Ok, you're solving this problem, so why can't other people just do the exact same thing?" And that's something you should be credibly able to answer. Either with IP answers, technical edge answers, or something. But you do need an answer to the question.

Break and Ideation Sprints

  • No specific notes here. Really, cool people and great energy, really really interesting ideas. Learned a lot, mostly about my cohort peers, rather than any businesses. Also, a surprising number like Pickleball? I've only vaguely heard about it; in my brain it maps to "squash, but for old people".

Alan Smith on Business Canvas

  • Audience demographics
    • about 80/20 B2B/B2C split
    • about 20% first-time entrepreneurs (including me :D)
  • Book: Value Proposition Design
  • Book: Testing Business Ideas
  • This is a talk about business model intricacies. There really isn't such a thing as a simple blueprint for a company, so you can't write "The 100 Best Business Models", much to our publishers chagrin, it'd be kinda like "The 100 Best Foods". You can boil out like 10 or 20 principles and processes and then explain them, but now we're in "How to Cook n Food" territory rather than "Top m Food Ideas".
  • Early stage startup is Searching for a business model, later stage startups are Executing a business model
  • In the early stage, you're looking for (audience answers)
    • What you're selling
    • Who your customers are
    • How much are they willing to pay
    • What is the problem that they're willing to pay you to solve
    • Is this a real prolem?
  • The search phase is extremely chaotic (he draws a spaghetti-looking jumble on the whiteboard to illustrate this)
  • Once it's stabilized, ideally, you're going to try to grow whatever ended up stabilizing, but there's a space of almost pure chaos beforehand that you need a model to work through properly.

In Chaos

You should

  1. Ideation (generate ideas)
  2. Problem validation (is this a real problem?)
  3. Solution validation (can you solve this problem, and what is the MVP?)
  4. Formation (growing the company, getting a team together do the thing)

At that point, you've stabilized and you can move on.

The Business Model Map

| |_| |_| |

  1. Customer Segments (who are you selling to?)
  2. Value Proposition (what are you selling them? What do they get out of buying you?)
  3. Channel (how are you selling to them? internet/malls/delivery system)
  4. Relationship (what's your relationship to the client? what keeps them coming back to you?)
  5. Key activities (what are the things you need to do in service of our goal/customers?)
  6. Key resources (what are the things you need to get in order to enable your key activities?)
  7. Key partners (outsiders to the company who perform key activities?)
  8. Cost Stream (what do you need to pay in service to your goals?)
  9. Revenue Stream (how do you make money as a result of your goals?)

Example - AI Headshot service

  1. CS
    • Dating sites
    • Grads
    • Real estate sales
    • Founders
    • Models
    • Pets
    • Plastic surgery
    • etc
  2. VP
    • Cheap
    • Convenient
    • High variety
    • Quality
    • Fast
    • Ease of use (pets are hard to get in front)
  3. Channel
    • Direct marketing to campuses
    • Tinder/Bumble/Whatev (channel partners)
    • Word of mouth
    • Modeling agencies
    • Influencers
  4. Relationship
    • Referral program
  5. KA
    • Develop software
    • Marketing
    • Customer service
    • Legal
  6. KR
    • Cloud resources
    • Models (AI models)
    • Servers
  7. KP
    • Realestate boards?
  8. R$
    • Pay per use?
    • Credits
    • Possibly subscriptions?
  9. C$
    • Salaries
    • Servers
    • Cloud services
    • Legal fees

Developing the above

JTBD Funnel

 ________ ________
| \      |      / |
|  \_____|_____/  |
|  /     |     \  |

(This was a square on the left and a circle on the right in the talk, I'm not going through that much trouble with ascii art)

  • JTBDs(Jobs To Be Done) are an interesting concept here
  • Lets run through that example:
    • We're picking real estate services as our CS
    • What are their JTBDs?
      • Sell houses
      • Advertise
      • Reputation management
    • What is the pain when it comes to the photos part? (Pains)
      • Photo cost
      • time
      • printing
      • being satisfied with the outcome
      • static (if they've changed their haircut since, they might need to re-do them)
    • What happens if everything goes amazing? (Gains)
      • Looks hotter (generates leads)
      • Instant turnaround time
      • 1:1000 cost
      • Always accurate
    • On the other, end you want to put together a feature list that connects either to addressing a Pain or enabling a Gain.
      • Open API a-la gravatar so people can link their portrait in anywhere addresses instant turnaround and always accurate
      • Being really cheap addresses the cost
      • Review step addresses being satisfied
      • etc.

What assumptions have we made?

evidence <-----> unknown
  • That our customer segment trusts AI
  • That it's legal to do this
  • That it's technically feasible
  • That there's a real pain
  • That we're cheaper than the competition
  • That our customers actually want better solutions to this pain
  • That we can raise funding

The point of this exercise is to point out three different spaces in the assumption problem:

  • Can we? ("Feasability" can we build it?)
  • Do they? ("Desirability" do they want it?)
  • Should we? ("Viability" does it make sense from an economic perspective? Can we make money doing this?)

Once we have these assumptions (business risks), we need to figure out what to do about them.

  • "Test them all" doesn't really work, because you won't have the time to test everything.
  • The question is about prioritization. A procedure (which he walks us through deriving) is:
    1. Get your assumptions and stack rank them in order of important/not important
      • ex: If it's literally illegal to do what you're doing you're done no further questions. On the other hand, if you're charging more than your competition, you might still win out if you have better UI, better throughput or better quality output. So, you want the top to be "problems that kill the business" and the bottom to be "problems that take it from $1b to $0.999999b". Non-issues don't rank here; don't bother thinking about them except to the extent that you might want to verify that they're non-issues)
    2. Assay amount of evidence you have for these, and see if you can become confident that any individual threat won't bite you

Experiment Funnel

  • B2B: Cold email -> Interview -> LOI
  • B2C: Ads -> landing page -> signup -> survey -> interview

Alan's Q&A

  • If you have a B2B2C, do you follow the B2B experiment funnel or the B2C?
    • It's basically wherever your biggest risks are. Point yourself to the big unknowns on big risks and hit those first.
  • Of the startups that use this process, how many of them succeed?
    • This isn't a guarantee for success; it helps you notice that you're failing.
  • What's an appropriate number of use cases per sample to help you solidify assumptions?
    • Depends on the strength of the assumption. If you've got a big bet on it, you need more testing.
    • Keep in mind that B2B vs B2C is different here; getting an LOI from someone saying "I'll buy that for $10k/year" is much different from getting an interview response of "Sure, I'd pay $4.99 for that app"
  • Between customer interview 1 and 10, is there any point where leading questions might start happening?
    • Read the book "The Mom Test". Basically, there's a way of talking to customers that give you false positives.

A lot of this rhymes with rationality training. But like, at a Company level rather than an individual level. It has to do with how much evidence you have for particular claims, and how likely mistakes are to be fatal. Given how formalized and templatized this process is, my intuition is that it might be automateable in the next year or two.

Creative Commons License

all articles at langnostic are licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License

Reprint, rehost and distribute freely (even for profit), but attribute the work and allow your readers the same freedoms. Here's a license widget you can use.

The menu background image is Jewel Wash, taken from Dan Zen's flickr stream and released under a CC-BY license