AI Startupese to plain English BS terms dictionary

AI has been quite the hype for a while already. Among some good companies who really do indeed innovate and introduce useful products, however amongst these you can find, to put it plainly, shit wrapped in a golden foil that tries to ride on the wave. Today we'll go over certain terms and what they mean in reality for most AI companies, if you are a VC or a dev on a lookout for a new job, this might come in handy. Also, long time no see Linkedin, it's been half a year.

AI Startup Dictionary for investors and software engineers

We are planning to integrate an advanced AI solution in our product - we don’t know jack shit about AI but want to ride on the wave and pump the bubble to scam you into a bigger round

Our solution bases on SoTA models with some of our secret sauce on top- we are a thin wrapper on top of openai that is one openai devs day from becoming obsolete or one student with full stack skills and a bit too much time during two weekends.

We are fully AI native company we’ve never did anything remotely related to AI however this one new intern put together a RAG PoC which we used to convince our board of investors that we are riding the AI wave, driving the innovation(TM)

We used broad corpus of data for fine-tuning and can handle almost any data plus we’ve already vectorised most of it - our data was a mess and still is but now at least it’s in another (v)db in another format (quite often managed solution with data who knows here) with no having a clue how to reliably use it, maintain it or go about the whole data lifecycle and management.

We have a strong data pipeline just getting put in place leveraging newest SoTA technologies and connectors to everything - we did a PoC of data scrapping with langchain that is absolutely not scalable, has no unified taxonomy or semi-sense in it’s structure.

We expect to develop and launch our first llm feature in two weeks - we never ran anything related to llms in production and are about to find out Our goal is to achieve full autonomy AI on most operations in the next months - again no idea (except for 3-5 companies who do not claim such stuff as they are in the know) what they are doing worst case scenario it’ll work as offshore student in India acting as the ai under the hood doing AI OCR by hand. (can't make this shit sup) We are aiming to increase productivity and enable people to do more with less using AI- prepare for a bigger round of layoffs and increase in responsibility and duties for all the rest of the teams, which other than normal company workload, will also contain lots of time for maintaining the automatic AI system. All our AI systems take security and privacy seriously - we disabled training data collection in our OpenAI subscription.

Look for companies with unique niches with domain expertise where they apply AI pragmatically as a tool rather than purpose itself. More on that soon.