What Building Talisman Has Taught Me About AI

Building Talisman has given me a way to explore what artificial intelligence can do inside a business. AI becomes interesting to me when it helps someone accomplish something specific. With Talisman, that purpose has been helping B2B sales teams manage LinkedIn outreach, c

onversations, and meeting booking. A problem I quickly identified in the world of B2B is that meetings with potential clients is what drives revenue. Working on that idea connects my interests in technology, entrepreneurship, and how businesses communicate with customers. When a founder is responsible for sales, outreach competes with building the product and running the business. There are prospects to contact, messages to write, replies to review, and conversations to continue. Talisman grew around the idea of making that process easier to manage.

What interests me about AI in this setting is its ability to work on behalf of a user. An outreach message needs to reflect what a business offers and why that might matter to the recipient. Generating a sentence is only one part of the task. The more difficult question is whether that sentence makes sense for the person receiving it. This relates more than I initially realized to my love of music and DJing. Both the premise of Talisman and DJing as a whole is about sending a message to a recipient with the goal of them understanding in its simplest form. That leads to a broader conversation and reaction.

That distinction has shaped how I think about the product. A message can sound polished while still being vague or irrelevant. Adding more words does not necessarily make it better. Sometimes a short question creates a clearer starting point than a paragraph explaining every feature. Building Talisman has made communication feel like a central decision in everything I do.

One area I have worked on is making campaign creation more conversational. Instead of asking someone to complete a form, the idea is to let them explain their business through a chat. Many products are realization that the GPT style interface is truly an optimal way to allow a user to essentially "word vomit" and come away with new knowledge and context.

A website or document can provide information about the product, audience, tone, and desired next step. That information can then become a starting point for the campaign.

The challenge is making that process easy without losing important details. A business might have specific language it wants to avoid or examples that show how its team speaks. It might already have campaign settings that should stay in place. I want the setup experience to help organize those choices and give the user an opportunity to review them before moving forward.

Working on Talisman has also made me think carefully about the limits of automation. AI can produce something that sounds confident without fully understanding the situation. In sales communication sometimes an incorrect detail or an awkward reply can affect how someone views the business.

There is also a difference between demonstrating a feature and building something people can use repeatedly. A demonstration can focus on one successful interaction. One important lesson I learned is that most users do not care at all about a new feature. All they see is how can this make my life easier, which I understand.

Overall, building Talisman has required decisions about problems, audiences, interface, and the communication happening through it. Each piece influences the others. A complicated setup can make a useful feature difficult to access, while a simple interface still needs enough context to produce a relevant result.

My experience building Talisman has strengthened my interest in AI as a tool for business problems. I want to keep exploring how it can reduce repetitive work while giving people meaningful control. The most exciting part is taking a technical possibility and developing it into something useful for another person.