Building a project-management platform where AI can actually do the work.
Morphlix designed and built Seshira as a full project-management platform for modern teams — combining Kanban boards, agile sprints, timelines, documentation, reporting, collaboration, automation, and an AI assistant in one workspace. Rather than treating AI as a separate chatbot, Seshira connects its agent to the same workspace capabilities used by people. The result is a system where teams can plan work, execute it, understand what is happening, and ask AI to act on the work itself.
- SaaS
- AI
- AWS
- Product Engineering
- Timeline
- 3 months
- Team
- 4 engineers
- Industry
- AI · Productivity
- Engagement
- Studio build
Results
330+
API endpoints powering the workspace
The platform's Go backend exposes more than 330 endpoints across project management, collaboration, authentication, billing, reporting, integrations, and AI capabilities.
24
AI tools in the shared capability registry
Seshira's AI agent works through a shared set of workspace tools, allowing the agent and the application interface to operate on the same underlying capabilities.
5
Work views for different ways of managing work
Teams can move between Kanban, table, swimlane, calendar, and timeline views without maintaining separate project data models.
The challenge
Teams were managing projects across too many disconnected systems.
Project management has evolved beyond a simple task board. Teams need backlogs, sprints, documentation, dashboards, reports, communication, files, dependencies, and increasingly AI assistance. But these capabilities are often spread across different products or layered together without a shared understanding of the work.
Work loses its context
Tasks, documents, conversations, reports, and project information can live in different systems, making it difficult to understand the complete state of a project.
Different teams work differently
A development team may think in sprints and story points. A product team may need timelines and milestones. Leadership may want dashboards and reports. A useful platform needs to support these workflows without forcing every team into the same view.
Project data is difficult to act on
Traditional project-management software stores enormous amounts of information but leaves the team responsible for interpreting it. Blockers, scope changes, workload imbalance, and project health often require someone to manually inspect the data.
AI is usually disconnected from the work
A chatbot can generate text, but generating a useful project plan requires access to the actual workspace. The valuable step is not simply asking AI a question. It is allowing AI to understand the project and safely act on it.
Real-time collaboration increases infrastructure complexity
When multiple people are working on the same board, changes need to appear quickly and consistently without forcing every client to repeatedly poll the backend.
Morphlix saw an opportunity to build a project-management platform where the workspace, collaboration layer, automation system, and AI agent share the same underlying model of work.
The solution
One workspace for planning, execution, collaboration, and intelligence.
Morphlix built Seshira as a modular project-management platform designed to support both everyday task management and structured agile workflows. Teams can create workspaces, organise projects, manage boards, run sprints, collaborate in real time, document decisions, automate repetitive work, and use AI to accelerate the work itself.
Workspaces
Organisations create dedicated workspaces with members, teams, roles, permissions, and project structures.
Boards & cards
Teams manage work through cards containing descriptions, assignees, labels, priorities, due dates, story points, checklists, comments, attachments, dependencies, subtasks, custom fields, and time tracking.
Multiple project views
The same work can be viewed through Kanban, table, swimlanes, calendar, and timeline / Gantt.
Agile project management
Backlogs, sprints, epics, story points, velocity, retrospectives, and guided retrospective ceremonies.
Projects & milestones
Multiple boards can be grouped into projects with milestones and Gantt-based planning.
Documentation
Teams can create documents, goals, dashboards, reports, and public forms that turn incoming requests into work.
Automation
Rules can trigger actions when cards change state, while recurring cards automate repeated work.
Collaboration
Real-time board updates, notifications, mentions, comments, guest links, and shared workspace activity keep teams working from the same project state.
Integrations
Seshira connects project work with external systems including Slack, GitHub, and Figma, while supporting imports from existing project-management workflows.
Enterprise capabilities
The platform includes capabilities for authentication, role management, auditability, SSO, two-factor authentication, custom roles, and custom domains.
Seshira is not one project-management workflow. It is a workspace that adapts to how different teams plan and execute work.
Architecture
A cloud-native workspace with AI connected to the same system of record.
Seshira is split across dedicated application, AI, infrastructure, and frontend components.
The core backend is implemented in Go with PostgreSQL, pgvector, and Redis. The web application runs on Next.js and React, while the AI agent is implemented in Python using Strands and Amazon Bedrock AgentCore.
AWS infrastructure is managed through Terraform, with separate development and production environments.
The architecture separates the concerns of the product without separating the underlying model of work.
Application layer
The Go backend provides the core project-management API, authentication flows, workspace operations, real-time communication, scheduling, and integrations.
Real-time collaboration
A WebSocket hub enables changes made by one user to propagate to other connected clients without relying on continuous polling.
AI agent layer
The Seshira agent runs separately from the primary API and connects to backend agent tools, so the AI system can evolve independently while operating against the same workspace capabilities.
Data layer
PostgreSQL provides the primary relational data store, with pgvector supporting vector-based capabilities and Redis providing fast-access application state and caching.
Infrastructure
Managed through Terraform: Amazon ECS Fargate, RDS PostgreSQL, ElastiCache, Application Load Balancer, AWS WAF, CloudFront, S3, Cognito, Route 53, Amplify, Bedrock / Bedrock AgentCore and SES.
Environments
Development deployments are triggered automatically from the development branch, while production deployments are promoted through a controlled pipeline from the main branch.
The AI layer
An AI assistant that understands the workspace — and can act inside it.
The defining difference in Seshira is that AI is connected to the project-management system rather than operating as an isolated conversational feature.
The Ask Seshira experience can work with the actual state of a workspace and use tools to perform actions.
Turn ideas into structured work
Seshira transforms natural-language requirements into task descriptions, checklists, acceptance criteria, task breakdowns, story-point estimates and clarity assessments, so teams start with an intention rather than an empty card.
Turn existing information into tasks
Seshira can extract actionable work from product specifications and meeting notes, reducing the gap between discussion and execution.
Generate project intelligence
The system produces standups, end-of-day summaries, weekly reports, changelogs and digests, turning project activity into information teams can consume without reconstructing what happened.
Analyse the state of the project
Seshira inspects workspace activity for blockers, scope creep, workload balance and comment sentiment, surfacing what would otherwise need manual inspection.
Take action
The agent uses workspace tools to create cards and move work: the human asks, the AI understands the workspace, uses tools, and the workspace changes, instead of copying a chatbot's answer by hand.
Keep actions reversible
Agent actions include an undo capability, giving users a way to recover from an automated change rather than treating AI actions as irreversible.
From human actions to agent actions
One capability layer for people and AI.
A key architectural decision in Seshira is the shared capability model: the application interface and the AI agent invoke the same workspace capabilities.
A capability such as creating a card exists once and is exposed to the web application, the AI agent, automation workflows and future integrations, instead of being built separately for each.
The AI doesn't maintain a separate version of the project. It works with the project itself.
Shared tools
Workspace capabilities are registered as reusable tools that can be invoked by the agent.
Agent execution
The AI agent selects and invokes tools based on the user's request and the current workspace context.
Workspace state
Actions ultimately operate against the same project data used by the rest of the application.
Reusable foundation
The capability layer creates a foundation for future AI workflows, automations, and integrations without rebuilding the underlying business logic.
Technology stack
A multi-service AWS platform built around a Go application and Bedrock-powered AI.
Application
- Go
- Chi
- PostgreSQL
- pgvector
- Redis
- WebSockets
Frontend
- Next.js
- React
- TanStack Query
- Tailwind CSS
- shadcn/ui
- AWS Amplify
AI
- Amazon Bedrock
- Bedrock AgentCore
- Strands
- Claude models
Compute & infrastructure
- Amazon ECS Fargate
- Amazon ECR
- Amazon RDS
- Amazon ElastiCache
- Application Load Balancer
- AWS WAF
- Amazon CloudFront
- Amazon S3
Identity & security
- Amazon Cognito
- AWS Secrets Manager
- AWS WAF
- Amazon Route 53
Events & communication
- Amazon EventBridge
- Amazon SQS
- Amazon SES
Infrastructure as code
- Terraform
Outcome
From project-management foundation to an AI-native workspace.
Morphlix took Seshira from an empty repository to a live product in approximately three months with a four-person engineering team.
The resulting platform spans project planning, agile execution, collaboration, documentation, automation, reporting, integrations, billing, and AI-assisted workflows.
The backend now exposes more than 330 API endpoints backed by 58 database migrations, while the AI system operates through a shared capability layer rather than a disconnected chatbot interface.
The architecture also provides a foundation for continued expansion: new AI capabilities, workflow automation, integrations, and enterprise features can build on the same workspace model rather than creating parallel systems.
Seshira's central idea is simple: the project-management system and the AI assistant should understand the same work.
Seshira was designed on a modular architecture, so workflow automation, AI copilots, document intelligence and enterprise integrations extend the platform rather than fork it.
More from the library.
All case studiesWork with the studio
Ready to future-proof
your infrastructure?
Tell us what you are building and where the current stack is holding you back. We will come back with an architecture and a plan.