Building a ‘Second Brain’ AI Platform for Efficient Development Teams

Industry:	IT, AI
Location	Belgium
Client since:	2025

Successfully tested three major releases in 2024

52% reduction in meeting time

99.5% accuracy in automated ticket generation

Services used

AI Services Application Development Cloud & DevOps Cloud Native Development Data & Analytics

Background

Modern software development teams are buried in context switching. Project details live in Jira, Confluence, GitHub, chats, and email, while AI coding assistants usually see only a small slice of that picture. Developers lose hours jumping between tools, sitting in status meetings, and rewriting the same documentation, and even then AI still lacks the organizational context and project history it needs to be genuinely useful.

Client

GUTT is a Belgian startup building an an AI-driven development platform. The company aims to transform how software development teams work by providing AI assistants with organizational memory and context. The result is AI assistance that’s grounded in the organization’s real history and standards, reducing correction loops and helping teams move faster with traceable, source-backed answers.

Challenges

The client set out to build an AI-powered “second brain” for development teams. The GUTT platform would capture meetings, index codebases, plug into existing tools, and give both developers and autonomous agents the full project context they need to work effectively.

The real challenge was turning scattered, fast-moving organizational knowledge into something AI could use in real time: capturing it automatically, feeding it into AI coding assistants, and enabling autonomous ticket implementation without losing quality or human oversight.

To make this possible, the team needed to:

Build a graph-based memory system to store and retrieve organizational knowledge
Implement real-time meeting transcription and context extraction
Create GitHub webhooks for automatic codebase indexing
Integrate with Jira, Confluence, Gmail, and key development tools
Develop MCP (Model Context Protocol) servers for AI tool integration
Build autonomous agents that can implement tickets with human validation
Create mobile apps (iOS/Android) for on-the-go access
Ensure enterprise-grade security and multi-tenant architecture

Solution

Symphony Solutions assembled a cross-functional team of backend and mobile developers, QA, DevOps, and business analysts to build the GUTT platform from the ground up. The team adopted an AI-driven development approach, using the very tools they were creating to speed up their own delivery and validate real-world workflows in practice.

The solution architecture consists of three main components:

1. Organizational Memory System

Built a Neo4j-based graph database that stores entities (people, meetings, decisions, lessons learned) and relationships between them. The system automatically extracts knowledge from meeting transcripts, code commits, and tool integrations, creating a searchable organizational memory that captures “who said what, when, and why.”

2. Real-Time Context Integration

Implemented GitHub webhooks that automatically index pull requests and commits, Jira API integration for ticket context, and meeting transcription services that capture conversations in real-time. This ensures AI assistants always have up-to-date project context without manual updates.

3. AI Agent Framework with Human-in-the-Loop

Developed autonomous agents that can research, plan, and implement features using the organizational memory as context. The system includes human checkpoints at critical stages: trigger approval, plan review, code review, and lesson validation, ensuring quality while enabling automation.

Key technical implementations included:

Build a graph-based memory system to store and retrieve organizational knowledge
Implement real-time meeting transcription and context extraction
Create GitHub webhooks for automatic codebase indexing
Integrate with Jira, Confluence, Gmail, and key development tools
Develop MCP (Model Context Protocol) servers for AI tool integration
Build autonomous agents that can implement tickets with human validation
Create mobile apps (iOS/Android) for on-the-go access
Ensure enterprise-grade security and multi-tenant architecture

The team also applied a “shift-left” approach, letting agents handle deep research and planning before implementation. This reduced rework, improved architectural alignment, and ensured that both humans and agents were working from the same, well-structured context from the start.

Results

The GUTT platform has been successfully deployed in pilot with the first client, a mid-size IT company, and is ready for cater more B2B customers. Below are the measurable business outcomes it’s already delivering:

Development Velocity:

Sprint velocity increased by 50% (from 24–25 to 35–37.5 story points).
Cycle time dropped to under 3 days from ticket start to deployment.
20–30% of work is now completed autonomously by AI agents.

Meeting Efficiency:

Meeting time was reduced by 52% (from 8.5 hours to 4 hours bi-weekly).
Async-first standups cut live attendance by 50% (people join only when they have blockers).
AI-generated pre/post meeting summaries save around 45 minutes per week.

Quality and Accuracy:

99.5% accuracy achieved in AI-generated Jira tickets and Confluence documentation.
Defect rate remains below 5% despite higher delivery speed.
Human-in-the-loop checkpoints safeguard quality on all autonomous work.

Knowledge Capture:

5–10 “lessons learned” are automatically captured each week.
100+ meetings have been transcribed, indexed, and made searchable.
Real-time GitHub integration indexes code changes within minutes.

Team Satisfaction:

Developers rated GUTT 8–10/10 in sprint retrospectives.
Reported context switching has decreased, addressing a key pain point.
Teams feel more productive and less frustrated with day-to-day workflows.

The platform is now entering B2B pilot phase with external companies, including logistics organizations and other software development teams. The solution provides:

A scalable, cloud-native architecture ready for enterprise deployment
Multi-tenant security with audit logging and RBAC/ABAC
Integration adapters for Microsoft Teams, Slack, and Google Workspace
Mobile apps for iOS and Android providing on-the-go access
Continuous learning system that improves agent performance over time

Overall, the project demonstrates how AI-driven development can not only build products faster but also fundamentally transform how software teams work, moving from reactive documentation to proactive knowledge capture and from manual task execution to supervised autonomous development.