Discover more from michael's writing
founder debrief - ari global summer 2023
2 week sprint for college students to explore the next big thing in climate tech
the design of this summer program started with the question:
“what’s the minimum amount of time to see meaningful results?”
minimum is based on the assumption that young people have limited time in the summer from other experiences (work, travel, or family obligations).
meaningful results is our intention of helping generate high quality ideas through an accelerated training experience.
this is what i wish i had while in school: the opportunity to work with other smart people on things that mattered ultimately building my blueprint for what impact i wanted to build. i was never satisfied with the bubble existing on campus and was curious to see what was on the leading edge of emerging tech and science.
for more context before designing this program, i’ve had a great experience as a program director and coach in tks working with 300+ teens to see amazing results start to happen after 3-6 months of this kind of program.
while that program focuses on weekend sessions during the academic year, i’m curious to see what a more intensive program looks like for college aged students.
the sprint was designed to explore the question:
“what is the next big thing in climate tech?”
this is a question that many chat over coffee in the bay area, and perfect for a structurally unstructured sprint.
it’s relevant as we have yet to solve large challenges in decarbonization or climate adaptation. even for current investments, there is an opportunity to scale at a speed similar to the rapid pace that we have not seen since the last industrial revolution.
it turns out two weeks was enough time to get to some interesting ideas paired with important problem spaces.
different teams explored topics like:
how might we capture some of the 1.35 t watt hours lost from solar in 2 months in the united kingdom using graphite for thermal storage?
what are ways to ensure the cost per ton of co2 removed from the atmosphere goes down by making valuable byproducts like carbon black?
is there a hypothesis to shrink the cost of tritium to accelerate nuclear fusion development?
when most people see this, they think these might have been ideas from graduate level students who already had ideas and had been primed with years of experience.
that was not the case. 70% of the students were pre-freshmen, meaning they had little to no formal engineering or science training.
everything was a result of the intentional environment with sessions for new knowledge, 1-1 coaching for upskilling and a clear structure to explore.
for more transparency, check out this process of evolution in thinking. on day one, we asked “what is the next big thing?” this was the same question posed on day five as well as the final day. the first two days were a couple of hours to get ideas out, and the final was a series of workshops to explore what makes a good idea.
day 1 looks like an average college classroom. templates, bullet points, and clip art. ultimately, chat gpt could do better with ideas and design.
day 5 has more clear framing for problems paired with exploring the technical depth in new ideas. students have started to read research papers and implement feedback.
day 10 final projects combined all of the above with problem statements, new idea discovery, understanding economic incentives, first customers, as well as an intentional narrative.
multiple repetitions were important for the rapid development of ideas. high standards is as trainable as bezos mentions.
one metric for success is if the students are proud of their own work. this means they had a level of effort and result that could not be achieved without the program. we pair this with outside perspective judges from founders, investors, and other builders to check for quality as a litmus test.
our judges for this sprint had more positive feedback than critical feedback, which was hard earned from the student side.
if curious, check out the final pitches starting at 7:16:
results alone are not enough, accelerated education needs to be a great experience
here’s what the final data in program surveys revealed:
the program was rated 4.6/5 for overall enjoyment. after running 200+ sessions in the last few years, the average benchmark at scale for virtual sessions is usually around 4.2/5. so this is a strong start.
some hypotheses for why this was so strong:
we used 🧈butter for a better virtual experience. as a platform it’s more fun than zoom and a lot more flexible for collaborative work.
the theme of climate tech was self selecting. those in the program already had one level of shared enthusiasm as well as common ground to connect on.
we’ve done this before and know the importance of 1-1 coaching. what we do outside of the session is just as important as what gets done in the session.
qualitatively, here is what students had to say:
“ari global has been an amazing experience. i had a background in climate tech and entrepreneurship, but as 2 separate things. here I learned about the intersection of both of them, which actually helped me rediscover my passions. before ari global I knew I wanted to found a climate tech startup, but didn't know how to start. now, I have a better understanding on how to do it.” —incoming freshman @ stanford
“it was a sweet balance between intense and enriching. it truly pushed me to my limits and i could make meaningful connections with my cohort to talk about the topics that matter to us.” —incoming freshman engineer from eric schmidt’s global rise scholars
“really great experience. best extracurricular thing i've done this summer. it expanded my mind a ton with real pitching experiences, strategies, feedback. extended my horizon line as well for what I can get done and what's possible.” —cs student @ waterloo
imagine if this was the baseline experience for young people in school.
they found individual meaning through their personal experience
had the opportunity to train new skillsets like project management, writing engaging blog posts, or doing research in unfamilar topics
explored projects with a metalearning structure they can always fall back on for future builds
overall, the quality of results and experience are two things we care a lot about. this should be more obvious in education, but it’s not always incentivized.
great education is coaching + cohort based training
the core ingredients to make those two happen are:
intensive cohort based sessions that train mindsets, skillsets, and knowledge. learning is better together as a multiplayer game.
1-1 coaching across skillsets, knowledge, and different areas of life, like time management, how to be a project manager, as well as human potential coaching to help increase personal velocity for growth.
collaborative learning community that shares the same values. we don’t use grades or prior experience for admission. we filter for curiosity and wanting to have an impact. we care more about what students will do v. what they have done.
more specifically, here's a look at how the program sessions were structured:
intro to the next big thing
mindset: high standards ∙ skillset: slide design
an overview understanding of the current climate tech landscape to identify key trends in the space.
ai in climate
mindset: curiosity ∙ skillset: 1 pager
explore how to think on how the latest developments in machine learning can change different sectors.
net zero materials
mindset: figure it out skillset: problem statements
discover how industrial materials will evolve to be net zero for carbon emissions. learn how to write clear problem statements.
mindset: bias towards action ∙ skillset: competitive landscape
research the potential of different approaches to fusion energy and understand what's unique about different approaches.
syn bio sprint
mindset: think 10x ∙ skillset: pitch
work in a team to identify a clear problem and a 10x solution using different tools in synthetic biology.
more info on the program website for schedule and outcomes.
note, these are topics that are not taught in school because they are either too new or they don’t fit neatly in a prescribed department’s box.
all skill sets are based on frameworks from leading founders, researchers, and companies. if we find a better way to do something, we implement it. this is similar to how matt mochary built his methods for his ceo coaching practice.
there are also constant feedback loops with session based coaching, debriefs, and 1-1 chats. this is all paired with async resources like community slack and a program platform, like 2-3 hour rabbit holes to go 0 to 1 in a topic.
bigger vision is the idea olympics
it’s normal for student athletes to spend 10-15hrs a week training in their off season and 25+hrs weekly on their sport of choice during their season.
we’re curious to see what that looks like for curious students who want to have an impact. what would the structure look like for an idea olympics for the next big thing?
train skillsets and build projects 10-15 hrs a week in the off season (fall + spring)
compete seasonally for resources and support for next steps (winter + summer)
integrate llm coaching for increased velocity with freedom to explore seriously on things that matter to each student individually as well as to the world
we’re always learning to get better, a few things we noticed
with a feedback culture, we like saying ‘it’s the best it’s ever been and the worst it will ever be.’
as we experiment with different hypotheses as well as explore what’s possible, we deeply care about student feedback to make this even better.
some things we want to improve for what we’re building next:
more students. this program had 10 students, 4 countries. we designed and launched ~1 month before running everything, so it was very much a pilot. now that we know it works, we’ll have more lead time to get more interest. ideal for next iteration would be 50 students per sprint to start.
more community experiences. sprinting for 2 weeks is an alone together kind of journey. next program we’ll intentionally start with more community sessions to build relationships as we get started while hitting the ground running. we’ll also find ways to continue to connect students and alum to the larger ecosystem.
explore different scalable models. right now it is a lot of manual time coaching in and out of session. next up, we’ll be experimenting with llm models for ai assistant skillset coaches to enable high quality coaching at scale.
what’s next for us
we’re doing 100+ customer interviews to see what an ideal fall program could look like. current ideas include a gamified discord, or strategically lite weekly program.
winter take 2 for next big thing. this program worked well enough, we’re ready for the next one. alongside a new and improved climate program, we even might do the next big thing in ai… sign up for updates below.
explore different ways to build out the core product. we’re exploring llm assistant coach for students. to build the mvp, we’re starting to look for funding options through south park commons and other programs to get the first product shipped by the winter.
final notes on future of education + ai
while the present opportunity to update areas that are missing in higher ed may start with these kind of sessions and community, ultimately for an accelerated program we’ll have to find the right way to integrate different machine learning models.
we’ve already experimented with students hiring their personal tutors with a chat training session… but that was v0.1.
up next we’re exploring what personalized coaching looks like for specific skillset training.
stay tuned for more updates. if you’re a college aged student and you’re curious to be invited for a future opp, sign up below.
if you’re a full stack engineer and want to explore building with us or if you’re a potential sponsor for future programs, email me email@example.com
our goal at ari global is to help increase the ambition of the most ambition to ensure global progress.
having run accelerators the last 10+ years for thousands of early stage founders from harvard, cornell, and tks, this is my next accelerator build. more on my background here.