MVP Documentation
Research
6 Weeks
Defining the MVP for a Conversational Sales Agent
Reducing query response time from 2 days to immediate using an AI agent.
Collaborator: Shivani Kolala Jayaprakash
My contributions include MVP research, documentation, initial prototypes and testing. The final designs are further developed by other designers and engineers.
Check out the agent in action
This is a fully developed demo, which included other engineers and designers contributions as well.
Problem
Static "Book a demo" buttons create delay and cause friction.
Sales teams lack context on incoming inquiries.
No way to understand if the user is a right fit.
Solution
Conversational AI agent that understands context and answers questions 24/7.
Closed knowledge loop to prevent hallucinations.
Smart qualification and routing to sales team.
Outcome
MVP requirements doc delivered to engineering.
Onboarded 5 design partners for MVP validation.
Prototype development initiated successfully.
Context
Every B2B SaaS website has a "Book a demo" problem, users visit a website, like what they see and try to understand what the product does, but cannot because it's behind a paywall and demo button.
Many companies customize their software to ensure the user's problems are addressed, but in order to understand if the company can solve user's problems, it takes 3 calls on average and 80% of leads get disqualified after 2 calls. They wanted to replace this first call with a context aware agent that can answer questions and qualify the right fit.
My Role
Led product discovery sessions with stakeholders
Conducted competitive analysis of AI chatbot solutions
Defined user flows and conversation design patterns
Created MVP requirements document and feature prioritization
Designed initial wireframes and interaction prototypes
Figma
Meet
Docs
How might we...
Create an AI agent that feels helpful and human, not robotic or intrusive?
Low Lead Quality
Form submissions often lack context, forcing sales teams to spend time qualifying unqualified leads.
AI Trust Issues
Users worry about AI hallucinations and getting incorrect information about products or services.
Data Concerns
Businesses hesitate to share proprietary information with AI systems they don't control.
Research & Insights
Context is everything
Sales teams spend 40% of their time gathering basic context that could be captured upfront.
Users do not want to provide unnecessary data
73% of users reported they do not want to provide data to a company if they do not know if the company will solve their problem.
Trust requires transparency
Users need to know when they're talking to AI and have an easy path to human support.
73%
Prefer to know if company is a right fit
40%
Time saved on context
Design Strategy
Three core pillars guided our MVP definition, ensuring we built something technically feasible and genuinely useful.
Context Understanding
The AI must understand the user's needs and business context to provide relevant, accurate responses without hallucinating.
Closed Knowledge Loop
All responses come from verified company data sources. The agent only answers what it knows, and escalates when unsure.
Sales Insights
Capture actionable data during conversations to give sales teams the context they need to close deals faster.
Solution Highlights
Natural Conversation Flow
Instead of rigid form fields, Path AI guides users through a natural conversation that adapts based on their responses. The agent asks clarifying questions when needed and keeps the tone friendly and professional.
"Feels conversational, not scripted."
The agent automatically qualifies leads based on predefined criteria, routing high-value prospects directly to sales while capturing detailed information about their needs, timeline, and budget.
"Only the best leads reach your team."
Actionable Sales Insights
Every conversation generates structured data that sales teams can use. Pain points, use cases, competitive mentions, and buying signals are automatically extracted and surfaced Slack/CRM tools.
"Turn conversations into insights."
Results & Impact
MVP Document Delivered
Comprehensive requirements doc with user flows, feature specs, and technical considerations delivered to engineering team.
Cross-Functional Alignment
Achieved buy-in from engineering, leadership, marketing on the focused MVP scope and 3-month timeline.
Prototype Development Initiated
Engineering team began building the first working prototype based on the documented specifications.
First Customer Pilots Scheduled
Secured 5 early-stage customers willing to test the MVP in their production environments through the design partner program.
1 month
From concept to MVP spec
3 teams
Aligned on product vision
3 months
Timeline to first prototype
Reflection
"Clarity beats complexity. The best MVPs aren't about cramming in every feature, they're about identifying the one thing that matters most and doing it exceptionally well."
Start with the problem, not the solution. We spent 3 weeks just understanding pain points.
AI isn't magic—it's a tool. Focus on the user experience, not the technology.
Document everything. Future you (and your team) will thank you.
Mockups
Fin