Show HN: Some Pay Millions for McKinsey, Others Use BrainTerms

Show HN: Some Pay Millions for McKinsey, Others Use BrainTerms(brainterms.ai)

4 points by Jderenne 1 year ago | 11 comments

Hi all !

We built a fantastic full agentic ai system for funders and product managers explore and pilot their product strategies.

With a team of orchestrated, objective drive and self enhanced AI Agents doing strategic expertise you benchmark your pricing, define your GTM strategy, roadmap and more.

We simulate your users and for each segment the system give you deep insights (churn, adoption, which features to develop etc).

Objective is to have a digital twin of your business and take decision based on simulated outcome

Need your help :

We have lots of visitors but few (free) sign up (ok we launched some weeks ago, but we expected more converstion).

We need to understand why. :-)

Tks ! Jerome.

Jderenne 1 year ago |

We added a strategic agents for GTM strategies and use it for ourself as well. The results from it and from the community feedback is great with more clarity on our message and an improvement of sign-ups.

By the usage of multiple AI models and a fine tuned architecture (https://www.linkedin.com/feed/update/urn:li:activity:7342834...) , some strategic agents leverage the hallucination as a great features (ideation for exemple) while hallucinations are very limited and controlled for other agents (references provided, specific setup, cross checking, usage of other technologies than LLM etc..)

codingdave 1 year ago |

There are a few problems I see here...

- As others have pointed out, if you need to ask us how to GTM, then your product isn't effective, as your entire claim is that it will tell you.

- AIs hallucinate badly when it comes to numbers. The techniques for preventing hallucinations need an valid data set to contrast responses against, which either won't exist or exists because you have hard-coded valid GTM strategies into a data set. So the end result is either prone to hallucinations, or is just regurgitating a canned GTM playbook.

In short, while the idea is interesting, I can't imagine it working well in reality under the current LLM stacks.

Jderenne 1 year ago | |

Tks,

We didn’t hard-code GTM strategies, the agent uses strong prompting + internal logic to get high-quality output from the LLM. It’s not just regurgitating a playbook, but it’s also not freestyling.

Actually our GTM strategy does not cover the design of a high converting homepage. (it give the strategic alignement a.o) We could add that later and see the impact on our homepage.

Financial benchmarking use their specific models and fine tuning. We constrain them hard, structured prompts, known-good data, and refined knwoledge from the agentic system. No room for AI "vibes." The results and projection (financial strategy is not accounting) is actually quite impressive tbh.

For others agents we see hallucination has a strong feature, while obviously it's not the case for financial benchmarks.

A_D_E_P_T 1 year ago |

> We need to understand why. :-)

Two thoughts:

- Most people will want to try something before handing over their personal details. Many startups, once they have your email address, will start spamming with emails, which gets annoying very fast.

- It's not clear what your product does that o3-pro/Claude4/Gemini2.5/DeepSeekR1 can't do. (Or do well.) Your video is mostly just a marketing spiel. If you can't offer an unrestricted, no-questions-asked free trial, what you want to do is show people a sort of A/B test, with your product vs. a well-prompted leading competitor like o3-pro.

Also just imagine hiring McKinsey in 2025. lmao.

Jderenne 1 year ago | |

thanks! appreciated feedback !

-There are two video, one positioning the product the other (at the end) walking through it. This one gives more insights ? not the right format ?

-I can let all people in (with some limitation) without any sign up (to remove all friction). I was afraid that It will only increase our AI cost usage without attracting serious "clients" ... wdyt ? any experience with that ?

A_D_E_P_T 1 year ago | | |

Video probably isn't the right format. A lot of people (myself included!) won't want to view them. Tables, charts, images, etc. are generally better, easier for any short-attention-span person who just happened across your site and wants to skim through what you have to offer.

I think what you really need to answer is: "What differentiates my product from a good prompt on o3-pro?" Make a comparative table or a whitepaper, and use that for marketing and outreach.

If, and only if, that doesn't work, I'd suggest you offer a zero-friction free trial. Maybe one or two free attempts?

Lionga 1 year ago |

> We have lots of visitors but few (free) sign up. We need to understand why. :-)

Why not ask you genius AI? It surely knows, it is made for that.

Well I can consult you: This thing is shit and will spew out brain dead bullshit. Much like McKinsey, so that part is on point.

How would you like to pay my 250K consulting fee?

Jderenne 1 year ago | |

Fair point we actually did ask our AI. It gave us 6 hypotheses, a funnel analysis, and a competitor teardown. Now we’re asking humans too they tend to be better at sarcasm :-)

We’ve been impressed by the depth so far (we uses domain-specific agents), often comparable to a solid strategy consultant. Still early, though, and always learning from edge cases and critiques.

2muchcoffeeman 1 year ago | | |

What good is your product if you yourself are asking a human?