H2 2023 Update
This is a longer, semi-annual update originally sent to investors in January, 2024. We send shorter updates every 4-8 weeks.
A related Twitter thread recapping the full 2023 story can be found here.
Overview
Our vision of accessible, life-changing meditation involves two things: groundbreaking tech and world-class instruction. We made great progress on both in 2023. In H1 we showed the technical problem was tractable. In H2, we showed we could teach jhanas without tech with 70% success on weeklong retreats. This led to dramatic testimonials (watch these!), 80%+ NPS, meaningful revenue, starting and appearing on podcasts, and preliminary word-of-mouth growth. We’ll begin to combine these in 2024.
In the short term, we’ll prioritize more retreats, which offer a road to default alive, attract talent and investors, and accelerate R&D all at the same time.
See here for upcoming public retreats.
Unexpected success in H2
To recap, in H1 we:
Built relationships with dozens of jhana experts
Collected what we think is the largest EEG jhana dataset in the world
Built an offline model that could detect jhanas on never-before-seen subjects with above chance accuracy
From this starting point, we set out to adapt our model to work for novices. This required many novices spending many hours under our slow, difficult-to-use equipment.
To entice novices to spend time with us, we experimented with multi-day jhana retreats. We figured if we could get 20-30% of people in jhana, we’d be able to sustainably continue R&D.
This was very ambitious. Jhana teachers spent years in apprenticeship models before setting out on their own.
Since our equipment would only complicate things and hurt our odds of success, we decided to run our first retreats without tech. We also recruited well-known teachers to observe and give us feedback.
To our delight, we succeeded beyond expectations!
65-90%+ of participants, totaling over 40 people, reached jhana in <40 hours on all 4 retreats, matching rates of top teachers, and defying the conventional belief that jhanas require hundreds or thousands of hours of practice.
Testimonials (same as above) were hyperbolic and NPS scores ranged from 70-100%
Unexpected revenue pushed back the need to fundraise by months
New talent and investors asked to get involved after attending retreats, suggesting retreats double as recruiting and fundraising tools
We completed two retreats online, at lower cost, with equivalent success
Experiments with evening workshops, a weekend retreat, and a cohort model also had surprising success rates, ranging from 10-30%
Challenges in H2
H2 also saw failed experiments. We’re just as proud to report these lessons as we are our unexpected success – in a world of power laws, maximizing ROI is in part a function of maximizing the number of low cost experiments.
Our biggest failure was investing 3 months of R&D into a detector for "jhana milestones” using consumer-grade hardware, rather than continuing work on the existing jhana model using research-grade hardware. For technical reasons, we thought this could offer low-cost, short term wins.
These did not work and we stopped investment in December. We took lessons on the difficulty of modeling jhana milestones and the usability of certain consumer-grade hardware.
H1 2024 strategy and priorities
We believe to scale jhanas to millions, we’ll need an intervention that teaches jhanas reliably in under 10 hours, or ideally under 2 hours. Such such speed and reliability will require a tech product.
But for now, we have a special opportunity in starting to scale low-tech retreats:
A “default alive” business means we’ll have more time to iterate
Recruiting and fundraising will be much easier
Teaching hundreds of people to enter jhana will accelerate and inform our tech roadmap
Gradually layering on tech will create a virtuous cycle with R&D
In H1 of 2024, our top priorities will be scaling our budding retreat business to reach cash flow positive and 50 new people in jhanas per month by September.
We’ll defer our raise until we’re too good to ignore: a positive, compelling business with demonstrated scientific breakthroughs, and a cornered resource on jhana data for R&D.
This breaks out into four initiatives:
Go-to-market – So far all customers have been organic. We’ll set up paid customer acquisition channels and iterate on our CAC
Build a content universe – Since our product requires a week off, we expect customers to require multiple touch points before a purchase. Fortunately, our stories are novel and dramatic; we think our podcasts will do well
Improve jhana speed, reliability, and cost – we can increase our retreat TAM by shortening retreats, or building a cohort model that doesn’t require time off work
Technical progress – In parallel, we plan to opensource our data, publish a paper(s) on our classification benchmarks, and do “spikes” on ways to make this data useful on retreats. We’ll build relationships with strong CTO and possibly Head of Research candidates, bring them on a retreat, and make hires as soon as we find the right people.
