Current Scenario
12%
Project Failure rate due to unclear goals.
25–40%
Stale Data within hours of manual entry
15–20 mins
Time Spent on board setup
Primary Users of Plaky
Founders & Small Team Owners
Startup founders, agency owners, and early-stage leaders managing multiple projects alongside broader business responsibilities.
Functional Team Leads
Business, marketing, and engineering leads coordinating tasks, timelines, and team deliverables.
Operations & IT Managers
Professionals overseeing workflows, internal systems, and tool adoption across teams.
As the research scope for this project was intentionally lightweight, we focused on desk research. Insights were gathered from publicly available online user reviews of Plaky. The data was then organised, analysed, and interpreted to uncover recurring themes, friction points, and opportunity areas, which directly shaped the problem framing and solution direction.
What Users Needed?
As a manager, I want automated status updates and reminders so I don’t have to manually maintain boards.
As a user, I want built-in chat linked to tasks so communication stays contextual.
As a team lead, I want board maintenance to be automated so I can focus on decision-making rather than administration.
Generative AI Integration
Current User flow
Plaky currently functions as a "Static Container." It excels at holding data but requires significant manual effort to input, structure, and interpret that data. Our focus is on the daily maintenance phases, where the lack of proactive intelligence and high manual maintenance creates a barrier to entry for non-professional project managers (e.g., CEOs of small agencies).
So we Ideated…
What if…
Generative AI worked like a Proactive Project Coordinator?
A new AI-powered chatroom transforms everyday conversations into structured, actionable work.
AI Detects a Potential Task
Key metadata like task names, assignees, and deadlines are extracted automatically from the natural conversation.
AI Suggests. User Confirms.
A "suggestion card" appears in the chat, allowing the user to review, edit, or confirm the task details before they hit the board.
Confirmed & Synced
Upon confirmation, a visual task card is shared in the chat, and the item is automatically added to the corresponding project board.
View Breakdown
Generative AI Information Architecture
Agentic AI Integration (In Progress)
Current Scenario
Project timelines and status updates are tracked and analysed manually when delays occur in the board.
Users must identify how a delayed task impacts dependent tasks.
Available scheduling windows must be manually reviewed to find a new feasible date.
Timeline updates must still remain within the project’s defined timeline limits.
The Problem
Delays require users to manually analyse the timeline and adjust schedules, which is time-consuming and takes focus away from actual work.
The Objectives are simple…
Analyse user activity on the board and within the systems.
Analyse task dependencies and timeline impact.
Provide rescheduling options for user selection.
What if
An Auto-Rescheduler could reorganize timelines when delays happen?
The Auto-Rescheduler monitors user activity on board and system. When a delay occurs, it analyses dependencies and generates rescheduling options for the user to choose from.
How it Works?
Auto-Rescheduler Workflow
Agents Used
The Auto-Rescheduler operates through multiple agents that analyse the project timeline, plan scheduling adjustments, and improve recommendations over time.
1.
Model Based Reflex Agent
Maintains a dynamic model of the project timeline beyond individual task cards.
Tracks team availability through integrated calendars
Maps hard and soft task dependencies
Evaluates how delays create ripple effects across the board
2.
Goal Based Agent
Plans rescheduling strategies once a disruption is detected.
Generates three possible scheduling outcomes
Evaluates each option against project timeline limits
Ensures all proposed dates remain feasible
3.
Evolution Based Agent
Improves scheduling suggestions by observing user decisions over time.
Tracks which rescheduling options users select most often
Identifies team preferences for buffer vs urgency
Adjusts future recommendations based on past behaviour
Agentic AI Information Architecture
Peas Model
The PEAS model describes how an intelligent agent interacts with its environment by defining its goals, surroundings, actions, and the information it observes.
