Bot-Buddy:
AI Customer Support Platform
Product Manager on a 13-person cross-functional team building an embeddable AI chatbot SaaS from scratch. Real sprints, real Jira, real team dynamics, and a live product shipped through beta.
Context
Bot-Buddy was a real product build, not a case study exercise. I served as one of three Product Managers on a cross-functional team with developers, business analysts, and a UI/UX designer. Real sprint ceremonies, real team conflicts, real deadlines.
Build an AI chatbot platform that revolutionizes customer support for SMBs — 24/7 support through a customizable, embeddable widget powered by document-trained AI agents that any business can deploy in under an hour.
The Full Team (13 People)
| Role | Count | Responsibilities |
|---|---|---|
| Product Lead + PMs (incl. me) | 3 | Roadmap, backlog, sprint planning, stakeholder alignment |
| Business Analysts | 4 | Requirements gathering, user stories, acceptance criteria |
| Developers | 2 | Full-stack MERN implementation |
| UI/UX Designer | 1 | Figma wireframes and design system |
| Product Owners | 3 | Backlog refinement, sprint review, UAT |
The Problem
SMBs have no affordable, customizable, document-trainable AI support widget. Enterprise tools (Drift, Intercom) are overpriced. Simpler tools (Chatfuel) only work on social media. No clean embeddable solution existed for a business to deploy on their own website, trained on their own documents, in under an hour.
| Competitor | Gap |
|---|---|
| Answerly — AI knowledge management | Not truly embeddable; limited SMB customization |
| Drift — Conversational sales | Enterprise pricing; overkill complexity for SMBs |
| Chatfuel — Social media bots | No website widget; locked to Meta platforms |
| Bot-Buddy — Embeddable AI widget | Document-trainable, white-labelable, 1-hour deploy for any website |
My PM Contributions
Working Agreement
I authored and maintained the team Working Agreement. My specific updates (June 7, 2024): Definition of Ready, Definition of Done, communication protocols, and participation standards. All 13 team members signed.
All changes tested across staging · Code peer-reviewed · CI checks passed · Merged to main · Documentation updated in Jira · UAT completed · Demo-ready at sprint review
Sprint Ceremonies I Led
- Sprint Planning: Refined backlog, sized stories, committed to sprint goal
- Daily Stand-ups: Async via Circle — tracked impediments, unblocked team
- Sprint Reviews: Demo to stakeholders, feedback into next sprint
- Retrospectives: Identified process improvements, updated working agreement
- Backlog Grooming: INVEST-compliant user stories with clear acceptance criteria
Key Product Decisions
- MVP Scope: Cut multiple specialized agents, analytics, widget customization from v1 — kept team focused on shipping a working core
- Tech Stack: Recommended MERN over PHP/Laravel or Angular — better real-time performance, team expertise match
- Out of Scope v1: Voice interaction, VR/AR, custom avatars, social monitoring — deferred with documented rationale
Product Architecture
Non-Functional Requirements I Owned
- Performance: Response within 2 seconds under normal load
- Scale: 1,000 concurrent users without degradation
- Reliability: 99.9% uptime with automated failover
- Security: TLS encryption, GDPR + CCPA, JWT authentication
Success Metrics
Timeline & Delivery
| Milestone | Date |
|---|---|
| Requirements + Architecture | May 20 – Jun 2, 2024 |
| Backend + Frontend Development | May 27 – Jun 16, 2024 |
| Internal Testing & QA | Jun 17–23, 2024 |
| Beta + Client Feedback | Jun 24–30, 2024 |
| Official Launch | July 4, 2024 |
Managing a distributed 13-person team taught me that the most valuable PM skill isn't writing great PRDs — it's keeping people unblocked and aligned when everything is moving at once. The Working Agreement and async stand-ups were the infrastructure that made everything else possible.