on scaling
my position is that of absolute pragmatism, a sense of optimization for the market alone. in an ideal scenario where the company has made its initial, key deals, from that point on the hierarchy of capital sources are as follows:
- revenue
- dilutive or non-dilutive raising
regardless of stages, non-dilutive funding should never be the engine because it always provides at a speed slower than the market. non-dilutive funding should always be the supplementary fuel, never the engine.
initially, however, dilutive funding is absolutely necessary, but the startup needs to make its transition into having revenue be the primary source after 1 or 2 rounds (preseed and/or seed) by introducing real customers. later dilutive rounds should only be justified to massively scale things such as full automation in manufacturing, which is a one-off thing. any potential r&d effort down the line should be approached with capital alone. look at deepmind being backed by google. deepmind is the r&d sector of google which provides capital without external assistance.
i can again argue that non-technical sf vcs that often moves too fast (critical trying to give me 200k without me providing any results or technical showcase), and government resources often move too slow for the sake of absolute certainty to consider ethical safety (likely causing loss of first-mover advantage to establish monopoly when competitors like Engram, which i should mention, exists). the middle ground, for first time founders, is a funnel that can lead to convos with technical investor networks, with reasonable tradeoffs and also fast enough. and my verdict is ultimately going for that sosv sf/ny route. it is being despised by both ends because i believe it sits in that exact pragmatic center.
SAFE is a good way to approach things, standard for the accelerator at sosv and also the broader sf network. i should explain that the valuation cap is the definite value of how we approach things. this is exactly like the overfitting vs. underfitting situation! sosv sf/ny and most inception stage investors use SAFE before a priced round. and they set their equity to align for the valuation of the startup during the priced round. it is worst to give them more control over a company because the valuation is lower than than their valuation cap and thus their initial SAFE check divided by that actual value lower of their cap would give them more control at that priced/seed round. it is slightly better, though not ideal, to have a round that is exceedingly higher than the valuation cap since that would overestimate the company's performance until the next round. for a CRO, it is never a short-term moonshot sprint but a steadily scaled expedition. so the sweet spot is somewhere marginally higher than the valuation cap. after the priced round, their ratio is no longer fixed but the actual amount of shares, so any future rounds would dilute both the founders and investors who hold a stake in prev. rounds. and to circle back to the dependence argument, there shouldn't be more than 3 rounds (preseed, seed and series A) to scale everything to full automation. so there won't be much dilution. by the third round the business should already be established with multiple ongoing deals.
going back to the short term... if the experiment result is at least moderately successful, the startup narrative should already be clearly established for applications for sosv sf/ny, or direct convos with VCs, a corporate and/or IP attorney already helping with everything IP wise, team structure and equity split (4 year vesting + 1 year cliff blah blah blah) established, and scale everything in summer. if we are doing sosv sf/ny specifically, we need to aim for the application being submitted by early july (july 1-7).
ofc if beatriz wants to do a paper, we do that after submitting the application. the "rolling basis" is probably season locked. publication is just a lower priority in terms of importance. a good addition for credibility! but additional. speed is crucial since we are not playing games. we are in a war w/ a very young competitor that already raised a series A.
IP ownership (in order of relevance to ME)
part 1: the QC software/pipeline/thesis IP
the main concern about this is that the data is generated within umass using their microscopes, their plate readers, their organoids, their lab capital, and their nutrients/reagents. i contributed on the software, ran it on my own computer and came up with the entire pipeline structure. so this is in a gray zone that I have to talk to letao about before walking into that meeting with sun and give him a straight answer on how this entire predictive QC thesis is definitively mine, not UMass's.
the question to ask letao is: given my current situation, would umass have any viable IP claim over software i independently developed, and if there's any risk (gray area?), what do i do about it before things get formalized?
facts that she needs to know:
my status: amherst college freshman. not umass student, not umass employee, not paid by umass, not funded by any umass grant. i have nens account (non-employee non-student designature) with research assistant title on my ucard, which actually means visitor credential for IT and building access purposes. eh that research assistant title (an administrative default on the UCard but not a reflection of an employment or funding relationship) could introduce more ambiguity and give the umass TTO (tech transfer office) more argumentative strength but i will have to see... w/ letao.
i completed scishielf safety trainings, registered for CORUM acct for facility booking. i don't think i signed a TTO participation agreement through CORUM or the safety trainings but im not 100% certain with potential IP language buried in onboarding clickthroughs. so i should also ask letao if i should verify this definitively, and if so, how? email TTO directly? check w/ EHS (Environmental Health & Safety department in umass) or pull records from scishield? etc.
what i used from umass: i used cytation 5 plate reader for automated imaging in 1st epoch. a leica microscope for manual imaging in 2nd epoch. the incubator, plates, reagents, the windows computers to retrieve data as an intermediate to eventually transfer into my machine, the organoids derived from iPSCs cultured in sun's lab using whatever funding they received. i didn't pay for any of this. the imaging were part of beatriz's routine experimental workflow that i participated in.
what i did independently: i wrote the entire QC pipeline on my personal Mac machine. I used m yown claude account, my own coding skills. nobody from umass helped me write the code, design the architecture, select features, or build the model. the software stakc (SAM3 MLX segmentation, feature extraction, ENR) is entirely my intellectual contribution. no umass compute resources were used.
the dependency: my software is trained on the imaging data that was generated using umass equipments, from organoids grown with umass-funded materials, in a umass lab with their student. without that data, the model doens't exist in its current form. but the model methodology and architecture could be applied to imaging data from any source. the data is the input, not the invention.
TTO's role:
this is the entity that would asset a claim on behalf of umass if they believed my software fell under their IP policy. a few possible outcomes analyzed by claude:
- letao thinks tto clearly has no claim because i'm not a covered individual and didn't sign a participation agreement, then i'm done. i document that conclusion and move on to the meeting with sun with no concern about IP ownership conflicts.
- she thinks it's ambiguous and recommends i proactively reach out to the TTO to get a written acknowledgment that i am not subject to their IP policy. the umass policy actually has a mechanism for this - cover individuals can submit a disclosure with evidence that their work didn't use significant university resources and the TTO issued a written confirmation for individual ownership. whether i qualify for that process as an NENS visitor is itself a question letao can advise on.
- she thinks umass has plausible claim, the strategy ight shift to bundling the software IP conversdation with the device licensing conversation. meaning that when the company eventually negotiates with the TTO for an exclusive license on the device patent, i could simultaneously resolve the software question asp art of that same deal - getting the TTO to release any potential claim on the QC pipeline in exchange for the broader licensing relationship. this would be a negotiated outcome and not necessarily any hostile legal fight.
the argument i need to make: I'm not a UMass student or employee. I was never funded by UMass (as in salary/stipent in my bank account; they clearly funded the experiment itself). I did not sign the participation agreement. The software was written entirely on my own hardware using my own skills. The data I used as training input was observational imaging data from a collaborative experiment, not prepared to a UMass dataset created specifically for my use. The university's significant use of resources standard was designed for people working inside a university system, not external collaborators who happened to be in the building. The policy specifically says that the use of library facilities available to the general public and occasional use of office equipment does not constitute significant use. I could argue that imaging sessions supervised by Beatriz fall closer to this category than to deep embedded use of university research infrastructure.
This is probably what I need to say to Letao and she can evaluate how strong this argument is.
additional note: UMass has three specific Participation Agreement forms — faculty/postdocs, graduate students, and visiting scientists. Ask Letao whether the visiting scientist version could have been triggered by my NENS status even if you don't remember signing it, and whether there's a way to confirm with the TTO whether one is on file for you.
And the conclusion that comes out of this call/meeting with Letao, plus whatever steps are needed to clarify who owns what with EHS, TTO, etc is what i hollistically need before walking to the meeting w/ sun.
part 2: the hardware IP
I don't need you to solve this problem. I don't need Letao's advice on this. This is Sun and Beatriz's co-invention filed through UMass as a provisional patent and the licensing path is between them and the TTO. My role in this meeting is strictly asking where things stand and making a point that exclusivity is absolutely critical and necessary because competitors exist. That's it. I'm not negotiating, I'm not advising anything. I'm just probing to understand the current state of flagged and flagging urgency. If Sun says we haven't talked to the TTO yet, that tells me there's work to do and I can offer to help drive the process. If he says we are already in discussions, great, I asked about exclusivity and move on. either way this isn't my problem and to prepare for it is a question to ask.
team structure and roles
sun's wants
i listen. he could just want an advisory role of the startup while staying on the sidelines in academia. he could want unlimited access to the software in his lab, so sure as long as it's running with their hardware? but if he is mainly working on hardware engineering but not organoids derived from the hardware, perhaps what i just imagined is false. he could talk about equity if he wants to be more involved with everything operational and commercial. i just need to understand his expectations so i can structure something forward that works for everyone in the picture.
regardless, i just listen, process his wants, and structure things accordingly afterward.
beatriz's involvement
ask sun about her timeline given her PhD. she would most likely be most beneficial to this venture by leading commercialization and continuing enginieering efforts to scale the custom hardware size. should she be a cofounder and/or science lead? first employee?
if she comes out of academia, her future inventions and IP will be much easier to manage than if she has ties with a university. therefore, asking about her timeline is especially important when it comes to future IP creations, not just availability and commitments.
my role
i should know this already. i want to operate the business. incorporate and scale the startup. negotiate with the TTO (if necessary), pitches to investors and expand my network with investors. drive this entire thing forward.
on the hardware itself (mental note)
per claude:
the custom device is general-purposed for patterned (meaning guided instead of unguided like cerebral) brain organoids. this is the best case scenario. no need to ask about if the device has any near term limitations besides geometrical scale. save that for the beatriz convo in april.
the device also enables better morphogenetic culturing protocol. versus what we have right now is giving every cell the same signal. the morphogenetic-enabling nature of the device creates more relevant-to-human-brain-biology organoids by making each organoid a compartmentalized differentiation path. parts of a organoid gets higher concentration of a signal than the other, etc. this is just for basic understanding of whats going on. the current PoC involves no morphogens. there you go. undersand this internally and talk more about this with beatriz if necessary.
time to invite letao to a call.