Agentic AI · Enterprise · production
MyPop
AI agent and agile coach for product teams.
Production AI agent that joins the tools teams already use—meetings, Slack, Jira, GitHub—to produce summaries, action items, risk alerts, and continuous-improvement coaching from real work signals.
Problem
Hybrid product teams lose signal across meetings and boards. Delays, rework, and blocke…
System
An agent pipeline that ingests meetings and project activity, retrieves prior patterns,…
Role
AI Architecture & Agent Orchestration
Status
production
Outcome
Multi-tool Jira · Slack · Zoom · GitHub
My role
AI Architecture & Agent Orchestration
- AI Architecture
- Agent Orchestration
- Retrieval
- Integration
- Productionization
Period / 2024–Present
AI agent for product delivery teams
Plays once · tap to replay
Problem
Hybrid product teams lose signal across meetings and boards. Delays, rework, and blockers surface late because no system learns from past delivery mistakes in context.
System
An agent pipeline that ingests meetings and project activity, retrieves prior patterns, orchestrates summaries and ticket actions, and surfaces coaching recommendations with early risk detection.
Architecture
Meetings and collaboration tools feed an ingestion layer; the agent retrieves team history, produces structured outputs, writes tickets/messages via tools, and surfaces coaching and risk alerts on a dashboard.
- Meetings → Ingestion
- Collab Tools → Ingestion
- Ingestion → Team Agent
- Team Agent → Team Memory RAG
- Team Agent → Tool Actions
- Team Memory RAG → Agile Coach
- Team Memory RAG → Risk Alerts
- Tool Actions → Team Dashboard
Execution path plays once on view · hover a node or tap the canvas to replay
System anatomy
Architecture
Ingest → agent orchestration → tools and coaching surfaces.
- Multi-source ingest
- Agent runtime
- Tool adapters
- Dashboard
Engineering decisions
Decision
Agent over the existing stack
Constraint
Teams will not abandon Jira, Slack, and Zoom for a new system of record.
Approach
Integrations pull work signals into the agent so coaching and automation land where teams already operate.
Result
Unified insights without forcing a workflow migration.
Decision
Meeting → structured work objects
Constraint
Raw transcripts are not actionable for delivery teams.
Approach
Agent extraction turns discussion into summaries, decisions, blockers, and ticket candidates with tool writes back to Jira/Slack.
Result
Less manual note-taking and fewer dropped action items.
Decision
Retrieval of past delivery patterns
Constraint
Coaching without history repeats the same mistakes.
Approach
RAG over prior meetings and project context grounds risk alerts and improvement tips in this team's reality.
Result
Recommendations tied to observed patterns, not generic agile slogans.
Decision
Security-first enterprise packaging
Constraint
Meeting and code-adjacent data is sensitive.
Approach
Encryption, GDPR/ICO compliance, and SSO on higher tiers make agent access acceptable for product orgs.
Result
Deployable coaching agent with compliance controls.
Reliability
- Encrypted data at rest and in transitACTIVE
- GDPR / ICO complianceACTIVE
- SSO (Teams+)ACTIVE
- Scoped tool integrationsACTIVE
Evaluation
Outcome / Results
- Automated meeting transcription, summaries, and action capture into team channels.
- Voice/context-driven Jira ticket creation with intelligent task suggestions.
- AI agile coaching recommendations grounded in team activity and delivery history.
Stack
Product visuals
