Workflow
Improve List Creation
TEam
1 Product Manager, 3 Engineers, and a Customer Success Team
Overview
Bankers use the Cyndx platform to search for companies and secure their potential list of targets to a list that will be used for deals that involve raising capital, investments, or acquisitions.
Problem
When bankers search for targets on a search platform like Cyndx, securing them in a list is the first course of action. The current user experience limits how users create a list and manage it within the platform. They prefer exporting the list to Excel for editing and post-processing company data.
Goal
Help bankers procure a list of potential targets in collaboration with their team and manage it easily
Current Workflow
After assessing
Understanding the concept of "concept"
Users find it challenging to understand the concept of "concepts" why they are not referred to as industries or sectors and how they operate.
Complex set of filters
The search features numerous filters that can be overwhelming for newcomers, and even experienced users may need help finding the right combinations for desired results.
Long load times and multiple trials
Search results took over 8 secs to load and which was the biggest frustration especially users who had to do multiple tries to get the right result.
Market Research
Some of Cyndx's direct competitors are evolving quickly with AI integration in the search & discovery of targets. This is one of the core reasons for product reassessment and to realign product strategies. We identified the following key elements that align well with Cyndx's product strategy.
Deal-sourcing platforms with NLP search
Leading M&A platforms are incorporating NLP to enhance free text search capabilities. This allows users to search using natural language queries, making it easier to find relevant information quickly.
AI-based insights
Artificial intelligence (AI) is being used to not only retrieve data but also to surface insights, predict trends, and suggest potential deals based on historical data and user behavior
Personalized experience
Customizable search filters and personalized search experiences are becoming standard. Users can set preferences for the types of deals, industries, or geographies they are interested in, and the platform will prioritize these in search results.
Roadmap
Design Exploration
"Free text search" or NLP is fairly new among the deal-sourcing search & discovery tools. This means that there is no standard design convention to follow and opens the room for experimentation to learn user behavior. However, "free text search" is not completely new as there are a few tools in the market that users are exposed to and that have established a design pattern to learn from. ChatGPT, Google Gemini, and perplexity.ai are a few of the most popular ones.
Through deep research, I have uncovered design principles and best practices that are very specific to generative AI product design.
Principle #1
Designing for Mental Models
Orient the user to generative variability. Help the user understand the AI system’s behavior and that it may produce multiple, varied outputs for the same input.
Teach effective use. Help the user learn how to effectively use the AI system by providing explanations of features and examples through in-context mechanisms and documentation.
Understand the user’s mental model. Build upon the user’s existing mental models and evaluate how they think about your application: its capabilities, limitations, and how to work with it effectively.
Train AI Model to work effectively with the user. Capture the user’s expectations, behaviors, and preferences to improve the AI system’s interactions with them.
Principle #2
Designing Responsibly
Use a human-centered approach. Design for the user by understanding their needs and pain points, and not for the technology or its capabilities.
Identify & resolve value tensions. Consider and balance different values across people involved in the creation, adoption, and usage of the AI system.
Expose or limit emergent behaviors. Determine whether generative capabilities beyond the intended use case should be surfaced to the user or restricted.
Test & monitor for user harms. Identify relevant user harms (e.g. bias, toxic content, misinformation) and include mechanisms that test and monitor for them.
Principle #3
Designing for Trust & Reliance
Calibrate trust using explanations. Be clear and upfront about how well the AI system performs different tasks by explaining its capabilities and limitations.
Provide rationales for outputs. Show the user why a particular output was generated by identifying the source materials used to generate it.
Use friction to avoid overreliance. Encourage the user to review and think critically about outputs by designing mechanisms that slow them down at key decision-making points.
Signify the role of the AI. Determine the role the AI system will take within the user’s workflow
Principle #2
Designing for imperfection
Make uncertainty visible. Caution the user that outputs may not align with their expectations and identify detectable uncertainties or flaws.
Evaluate outputs using domain-specific metrics. Help the user identify outputs that satisfy measurable quality criteria.
Offer ways to improve outputs. Provide ways for the user to fix flaws and improve output quality, such as editing, regenerating, or providing alternatives.
Provide feedback mechanisms. Collect user feedback to improve the training of the AI system.
Best Practices
Check the validity of prompts - asking users to verify if the generated content is accurate.
Uncertainty estimates (indicating the model’s confidence level in a given output) bake human oversight into the generation process.
Explainability features can further increase visibility into the AI’s “black box”: for example, using heat maps to visualize which inputs most influenced the model’s output, including citations to external sources, and showing alternative responses the model considered.
Interfaces that let users tweak prompts and parameters and observe the effects in real-time could further boost transparency into a model’s inner workings.
Design elements that can help optimize “time to trust”: that pivotal moment when a user gains enough confidence to rely on an AI system for complex, mission-critical tasks.
Wireframes & Iterations
Based on customer feedback from the support team and observations made during the sales demos with the current platform, we discovered three recurring pain points.
Iteration #1
Iteration #2
Iteration #3
The MVP
Based on customer feedback from the support team and observations made during the sales demos with the current platform, we discovered three recurring pain points.
Retrospect
This project was developed in such a short timeline, with a lot of iterations, exploration, and discussion with the stakeholders. It was fast-paced and here are some of the important lessons that I learned along the way
Being intentional about the terminolgy
This new product enables users to search for companies using Natural Language Processing (NLP), essentially functioning as a "free text search." However, it took some time for our internal teams to adapt their language to this new concept. I had to be deliberate in explaining what "free text search" entails and how it operates in the context of company searches, as it isn't as open-ended as one might expect. Users are still encouraged to apply filters to refine their search results.
Learning AI-Model & data
Creating an AI-powered search tool provided me with valuable insights into the capabilities of AI algorithms. It allowed me to anticipate user behaviors and expectations in advance. Collaborating with the data team was also beneficial, as it helped me understand the complete data landscape and discover ways to present data within the context of searches.