March 5, 2024

How BetterFutureLabs Leverages Conversational AI Agents

Dive into how we leverage Conversational AI Agents at BetterFutureLabs through our idea analysis tool MagicJ.


BetterFutureLabs stands at the forefront of transforming innovative ideas into groundbreaking realities. Our mission—to rapidly identify and nurture the seeds of tomorrow's technology—presents a unique set of challenges. The sheer volume and rapid pace at which new ideas are generated demand an urgent need for a method to evaluate their potential that is both swift and cost-effective.

The Validation Bottleneck

Our venture studio model thrives on discerning the potential in a sea of ideas, aiming invest in only a select few that show real promise of becoming successful companies. However, the challenge lies in the sheer volume and variety of these ideas. Sourced from universities, corporate partners, entrepreneurial co-founders, and our own brainstorming. These ideas span from nascent thoughts to early-stage MVPs with user traction.

Sifting through these ideas requires a rigorous, detailed research process focused on identifying the highest potential for success. This process is manpower-intensive and time-consuming, involving deep dives into market landscapes and investment climates. Given our position as an early-stage company, the necessity for an innovative approach to bypass validation bottlenecks without sacrificing quality is clear.

Introducing MagicJ

Our solution, MagicJ, draws inspiration from the research departments of leading consulting firms. Envisioned as a specialized department within BetterFutureLabs, MagicJ comprises teams of Conversational AI Agents, each focused on specific criteria crucial to our idea screening process. This structure enables collaboration and streamlined idea validation, overcoming the limitations of traditional, labor-intensive processes.

How MagicJ Works

At the core of MagicJ's innovative approach is a structured hierarchy of Conversational AI Agent Teams. This hierarchy is designed to delineate responsibilities clearly and foster effective collaboration within a team.

Here's a closer look at the hierarchy and the specific responsibilities of each agent:

  • Director: Receives and analyzes new ideas, setting specific research objectives.
  • Research Manager: Translates these objectives into actionable tasks for the Researcher. Reviews the Researcher’s work for accuracy, proper citations, and completeness.
  • Researcher: Executes tasks using tools like Google searches, webpage scraping, and accessing private data sources via APIs, sending completed tasks to the Manager for review.

At the helm of the team is the Director, who receives new ideas logged into an Airtable. From there the Director analyzes each idea and formulates specific objectives for the team.

Here is an example of the objectives the Research Director sets for the team:

The Research Manager formulates these objectives into actionable tasks for the Researcher to complete. This breakdown ensures scoped tasks with concrete deliverables.

The Researcher executes the assigned tasks using the tools available (like the web searching, scraping, and APIs mentioned above) at its disposal and sends the Manager the results for review.

Feedback Loop

The Researcher sends the completed tasks to the Manager for review. The Research Manager looks over the work and provides feedback. When the Research Manager spots something it thinks needs improvement, it critiques the work (sometimes a bit too harshly) and then provides a few recommendations for the researcher to improve the quality of work.

The Researcher then begins incorporating the Manager’s feedback into it’s analysis.

This Feedback loop is what drastically improves the quality of our research over other methods and just asking questions to ChatGPT. The conversations between the team members mimic conversations that would be had between different stages of the review process at a market research department.

Outcomes and Impact

MagicJ represents a revolutionary shift in idea validation, offering a faster and more cost-effective approach. On average, a MagicJ report is generated in about 7 minutes, achieving approximately 80% of the depth of analysis that we would expect a venture analyst would provide us. This accelerates our decision-making process and allows us to assess a wider array of ideas with the same level of scrutiny.

How to Help

As we evolve MagicJ, we invite feedback from investors, corporate partners, and the TechUnited community to further improve MagicJ. If you see potential for MagicJ to benefit your team, we're eager to explore collaboration.