### vision and T4
T4 Topic1, 2 (learning and improving dynamic human decisions) aligns with my interest in developing AI mentor (nss_navigator) that helps entrepreneurial operations. nss_navigator discusses with entrepreneurs on how and when to segment, automate, processify, automate, acculturate, professionalize, collaborate, platformize, capitalize, evaluate. Furthermore, it requires entrepreneurs to continuously test the co-designed model by providing smart and scalable hypothesis testing and experiment design framework. In other words, it emulate mentor's insights and democratize consulting. Let me elaborate how this vision and choice (entrepreneurs and AI mentor) can fit into and be supported by T4.
### customer: entrepreneurs
Entrepreneurs have a critical role to play in economic development. One need look no farther than the entrepreneurial founders of the large tech companies of the past few decades to see their impact. Generative AI has catalyzed the blossoming of many AI startups, and many more will likely be created in the coming years. Additionally, many more entrepreneurs will be affected by AI, as a tool to help them, or sometimes as a tool that puts them out of business, if they are not sufficiently nimble and clever. A large fraction of all startup companies fails, yet there is also evidence that more experienced and mature entrepreneurs probabilistically have greater success. Work at the MIT Sloan School by several entrepreneurship scholars/educators, including Charlie Fine (my advisor) on [Nail, Scale, Sail framework on operations for entrepreneurs](https://operations4entrepreneurs.com/nss-framework)), Scott Stern on [entrepreneurial startegy](https://www.entrepreneurial-strategy.net/), Bill Autlet on [disciplined entrepreneurship](https://www.d-eship.com/) and tactics, seeks to codify some of the knowledge that may help entrepreneurs improve their likelihood of success. Admittedly, the decision space for entrepreneurs is large and complex. However, we believe that there are many case examples of successful and unsuccessful entrepreneurial ventures that can support a database to help us codify the knowledge we seek.
### technology: AI mentor
Based on my practical experience of founding startup that supplies forecast demand and optimize inventory for small business people, my theoretical preference of pre-asymptotic to asymptotic, instability/disequilibrium to stability/equilibrium, tools to reach normative to analysis from normative, growth diagnostics to growth theory; my interactions this summer with practitioner and scholars in entrepreneurship, I identified three areas to technologies to develop. `pitfall modes` described existing problems (I felt or observed), `as-is` is how practice and academia approach this problem (which has problem on its own way). `to-be` column explains the direction of improvement. It is our hope that generative AI may open up new avenues of research to speed the development of our codification of the knowledge most needed by entrepreneurs.Â
| | | | |
| -------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| pitfall modes | as-is (practice) | as-is (theory in academia) | to-be (theory, tool in nss_navigator) |
| unprincipled and inefficient hypothesis and experiment testing | startups grapple with validating their business hypotheses | traditional hypothesis testing like NHST aren't tailored for startup environments, leading to inefficiencies and missed opportunities | Bayesian model testing, action and decision based testing + diagnostics monitoring tools |
| underestimated and not modeled uncertainty within oneself | many producers and consumers do not know what they want and can supply and demand. Uncertainty within oneself is underestimated so learning oneself, from its past decisions ex-post where conditioning on a less uncertain environment is attemptable, is one way to preparing for its decisions ex-ante. | rooms for improvement in modeling belief state for sequential decision making/RL literature, unclear distinction of what should be learned (e.g. risk preference), could be learned (e.g. belief on certain idea, technology, market which can sometime be more and sometimes less precise), could be improved (e.g. startup level decision process, use of decision support tool), should be improved (e.g. community level communication standardization to facilitate learning pooling) | prior elicitation, gut feeling analyzer |
| unstructured experience pooling deterring knowledge diffusion | startups find conflicting mentors' advice troubling | incommensurable, replication crisis, locally tested behaviors are (sometimes intentionally) reported as if it applies to general setting, to gain the status of the "theory". This leads to conflicting advices, which are troubling to startups. | system level typification and mapping, mechanical tool to adopt is prescribed base on object's character (size and speed) in pysics. Likewise, operation and strategy tool to adopt can be prescribed base on startup's character (business model feature: customer, technology, organization, product, competition + choice: orientation (execute vs control) and investment (compete vs collaborate)) |
| less actionable prescription to startup | language is less standardized, less actionable | unrealistic assumption (e.g. decision from an agent with unrealistically high level and low uncertainty of knowledge, confusing global and local view by assuming fully mixed model (e.g. differential equation based vs agent-based, one global s-curve vs global s-curve as an envelope of several s-curves | nss_navigator's prescription is tool-based. Tools implement and guide the strategy and dynamic theory exist (by Charlie Fine) on how `segment, automate, processify, automate, acculturate, professionalize, collaborate, platformize, capitalize, evaluate` tools can be used in nail, scale, sail stage across startup lifetime. Tool (:=abstracted action bundle) is handy interface that lowers user's (e.g. startup) and process designer's (e.g. mentor) cognitive load. Through discretization, user can not decide or evaluate instead of optimize. Through modularized states, the entrepreneurial journey. Also, milestone allows reverse-engineering and planning. |
### organization
Other than constant communication on the above idea with Charlie, Scott, Jinhua, Sandy, I submitted nss_navigator to [MIT ignite](https://entrepreneurship.mit.edu/mit-ignite/) (generative AI entrepreneurship competition) to get feedback. Also, I'm proactively reaching out to Singapore networks to ask for support.`Yanling Liu` from Changi Airport Group (Sloan fellow). She promised support in applying the theory we learn in the class we're taking together (15.911 entrepreneurial strategy) using her experience and ongoing projects in Changi airport operations e.g. self-service check-in, biometric measures, using AV to transport baggage). `Pengfei Chen` from the Singapore Economic Development Board (Sloan fellow) can help analyze ventures in growth industries targeted by EDB and to distribute nss_navigator.`YoungGun Song`(Hyundai Motor Group Innovation Center in Singapore) can help Charlie and Angie visit local factory and we can discuss innovation strategies for established companies. Also we can learn auto-tech entrepreneur ecosystem in Singapore.`Howard Califano`(Singapore Innovation Center) can provide data and insights on local startups. `Prof. Hahn Jungpil`, Vice Dean at NUS. He could advise on AI governance perspectives.`Prof. Moon Seung Ki` from NUS Engineering. He could provide a technical perspective. `Robert Fu`, CEO of CellWave Technologies (SMART CatalystÂÂ). We can interview him to learn his entrepreneurial decisions and ask for some connection to other members of SMART ecosystem.
### journey log
in [NSS_GPS](https://github.com/Data4DM/BayesSD/discussions/159) (nss_navigator's mothership) and [Applying NSS_Navigator for EV architecture decisions](https://github.com/Data4DM/BayesSD/discussions/167)