# traction is limited as the idea exists for only 6 weeks but there is some during these 6 weeks i had 2 workshops, where participants used my 2 metaprompters ![[Pasted image 20250510235332.png]] [botmaker](https://poe.com/botmaker-0000) ![[Pasted image 20250510235344.png]] [prompter-3000](https://poe.com/prompter-3000x335) ** more importantly, the idea itself was born in **hundreds of conversations** with end users during my workshops i collected over **70 use cases** from them, learned about the challenges of using ai tools as my first business, i run an ai consultancy, which builds ai tools for companies. building an agentic app costs between 30 and 100k, although all 70 use cases are excruciatingly similar - while still having a ton of customization need in my work, i built several repositories, eg, rag pipeline, chatbot, sequencer for prompts, integrations, etc - which i can adapt and repurpose to accelerate cogit perhaps most tellingly, in these 6 weeks, i have sold **2 mini-projects for about 4k EUR** each, dropping the price significantly. for these customers, i make a prompt with @botmaker, a simple metaprompter, generate an ai-readable dataset from source material (with other 2 prompts), and load as a custom gpt / poe bot for the customer to use all that could be done in 60 seconds with **cogit** it would create an agent that writes prompts, munches data, and uses a browser to setup a custom bot. simple enough, but there is no tool on the market that can do that try in chatgpt, manus, cursor - they won't load data from outlook, they don't know the format we use, they can't build a custom llm as a judge evaluator, and will bang their head against openai's login form the worst part is that they **can't learn**. pretraining cycles are expensive and slow, while in-context learning capabilities are underestimated + no one notices that reasoners can be created for any task. see my post [[expect ai magic to continue]] for more detail so lastly, a few words about --> [[team]] ---