AI meets Negotiation Expertise
The art and science of negotiation is undergoing its biggest transformation in decades.
Our research, involving 120 experienced negotiators in complex business deal simulations, demonstrates that LLMs fundamentally change negotiation dynamics.
When only one party has access to LLM support, they achieve notably better outcomes: buyers gained 48.2% and sellers 40.6% more value compared to their counterparts.
Even more compelling, when both parties use LLM support effectively, joint gains increase by 84.4% compared to traditional negotiations.
However, achieving these results requires mastering both negotiation fundamentals and LLM capabilities. Neither alone is sufficient.
AI meets Negotiation Expertise
Beyond the Headline: Enhancing Behavioral Insights with LinkedIn Data
One problem we are facing in Cassidy is that whenever we scrape a LinkedIn profile with the built-in scraper, the information is limited to the headings.
If you enjoyed this episode, please leave a review and check out our website: www.negoai.ai
I welcome any suggestions, questions, or comments at yrana@negoai.ai
--------------------------------------------------------------------------------------------------
Copyright Disclaimer Under Section 107 of the Copyright Act 1976, allowance is made for "fair use" for purposes such as criticism, comment, news reporting, teaching, scholarship, and research. Fair use is a use permitted by copyright statute that might otherwise be infringing.
Hello. Let’s return to the behavioral assistant we recently upgraded. We’ve made solid progress, but two practical issues still hold us back.
Problem 1 – Shallow LinkedIn Scraping
One problem we are facing in Cassidy is that whenever I scrape a LinkedIn profile with the built-in scraper, the information is limited to the headings. It doesn’t capture the About section, and it has no details beyond the experiences.
As a result, the information I can give the behavioral assistant is quite limited—even though the assistant itself is very comprehensive. This is the first problem we face, and here is our usual report. It has been upgraded a little bit; you can see it’s more comprehensive and more verbose than the previous one.
Problem 2 – Character-Limit Constraints
Most of our clients aren’t running workflows inside Cassidy. Instead, they use Microsoft Copilot or ChatGPT.
While our Daniel Behavioral Assistant is very comprehensive— counting at roughly 85 thousand characters — so we must slim it down, because ChatGPT allows only 8 000 characters and Copilot 10 000.
Today we’ll tackle the first problem—how to provide more comprehensive LinkedIn information to our assistant so it can build a better profile.
The tool we’re going to use is Website to Markdown. It converts the main content of a web page and lets you download it as a Markdown file.
After downloading, we attach that Markdown file to the behavioral assistant inside Typing Mind and run the analysis.
The assistant converts the Markdown into JSON, now containing skills, recommendations, courses, groups, newsletters, and publications—far richer than Cassidy’s default output.
With the expanded profile data, the assistant’s confidence score rises to 95 %. We also receive a fuller behavioral analysis, including:
Thomas-Kilmann conflict-mode profile, Tips for interacting with a collaborative style, Transparency notes and inference basics, Key caveats for interpretation
Our assistant converts Markdown into JSON, giving us much richer data than Cassidy’s built-in scraper. The remaining challenge is compressing our 85 000-character system instructions to fit within Copilot and ChatGPT limits—something we’ll cover in next week’s video.
Thanks for listening and if you enjoyed the episode subscribe to our newsletter at negoai.ai
