KnowledgeWorking knowledge
The context that builds up as you work: notes, readings of requirements, policies, retros. It keeps changing and feeds the AI context.
- Used as AI context
- Linked by relationship
- Continuously updated
So you can focus only on the work,
without breaking your flow.
With a process that stays clear.
People and AI share the same context, split the roles,
and build the product as one team.
ApexStack/Product/Project Alpha
Search⌘K
JHSYMK
Project Alpha
In progressSprint 12OverviewTasksTimelineDocumentsMembers
Progress68%
Open tasks24
Due this week6
Documents38
TaskStatusAssigneeDue
ALP-142Redesign the checkout flowIn reviewJHSep 24
ALP-139Clean up the onboarding email sequenceIn progressSYSep 25
ALP-131Finalize the Search API response specBlockedMKSep 26
ALP-128Sprint 12 retrospective docTodoJHSep 29
ALP-124Review workspace permission policyTodoSYOct 2
Problem
Are you explaining it all over again, every time?
With a general-purpose chat AI, you often have to restate the project's background and structure, its requirements and earlier decisions every time you start something new.
Hand over context→Ask→Check the result→Ask for changes→Hand over context againREPEAT
The conversation is fast, but the longer a project runs, the more work it takes to hand over context and fold the results back into the actual work.
Human carries the context
Project BackgroundRequirementsTech StackPrevious DecisionsCurrent TaskRelated DocumentsConstraints
AI AssistantGeneric LLM
"This project runs on Next.js 16, and…"
"The existing auth structure is…"
"What we decided last time was…"
"For this task we need…"
Human carries the context
Project BackgroundRequirementsTech StackPrevious DecisionsCurrent TaskRelated DocumentsConstraints
AI AssistantGeneric LLM
"This project runs on Next.js 16, and…"
"The existing auth structure is…"
"What we decided last time was…"
"For this task we need…"
It isn't that AI is slow —
it's that explaining the work to AI keeps repeating.The bottleneck is not always the model. Sometimes, it's the workflow around it.
PJT
In PJT, the project is the context.
Plans, tech specs, tasks, documents, decisions and work results are all connected inside one project.
AI agents work from the project context they need, and the results flow back into tasks and knowledge.
BEFORE · carried by a person every time
Project BackgroundRequirementsTech StackPrevious DecisionsCurrent TaskRelated DocumentsConstraints
PlanPlan · 3 phasesPhaseMS 2 in progressEpic6Task24 · 9 openTech Specv3 finalizedDocs38Knowledge112 sourcesDecisions14HistoryFull change historyAgent ResultRun logs · artifacts
Cycle
PROJECT CONTEXTContext connected to the project
AI AGENTRuns on that context
WORKWork at the task level
RESULTArtifacts · verification
KNOWLEDGESources and decisions on record
↺ Results become context again
Spend less time explaining,
and more time on the actual work.
You don't just use AI —you work with it.
People and AI understand the same project and each play their role inside the same workflow.
People set the direction, AI agents carry out the work, and the results become the project's context again.
Agent
HumanYRYoungrok
Approved the direction for passkey authenticationDecision · Sep 18
Requested a review of Tech Spec v2Review · Sep 19
Finalized the scope of 3DS authenticationDirection · Sep 20
DirectionDecisionReviewApproval
Shared Project ContextProject Alpha
PlanTech SpecTasksDocsKnowledgeDecisionsActivityHistory
People and agents look at the same information.
AI AgentIdle
Understanding context
Planning
Working
Verifying
Completed
ALP-128Implement Passkey Authentication
AnalyzePlanExecuteVerify
Work · ResultWaiting
Implement Passkey Authentication
Files changed7
TestsPassed
ReviewRequested
Knowledge
Implementation note · passkey authentication flow
It lands in Project Alpha's context, so people and agents both reference it on the next task.
Back into contextContext accumulates as the project moves forward.
A real collaboration moment · after a tech spec change
- HumanTech spec change approved
- PJT ContextTech Spec Updated
- AgentAnalyzing affected tasks
- AgentTASK-128 implementation startedWorking
- ResultImplementation CompletedTests 42/42 passed · 3 files changed
- HumanReview Requested
- KnowledgeDecision / Result Stored
Human decides.Agent executes.PJT keeps the context connected.
HumanContextAgentWorkKnowledgeContext
Quiet Workspace
Focus on your
workspace.
PJT cuts back on unnecessary visual noise so your projects and your work are what you see first.
Step away from a sea of browser tabs and use PJT in a workspace of its own.
Project Alpha/ALP-142
In progress
Redesign the checkout flow
List the payment methods
Define the error states
3DS authentication step screens
Write the QA scenarios
Personalization
Quiet by default.
Rich when you need it.
A workspace theme is not decoration that floods the screen — it expresses selection and state. Backgrounds and surfaces stay neutral; color goes only to the selected item and the primary action.
ApexStack/Project Alpha
Search⌘K
Project Alpha
In progressNew taskOverviewTasksTimelineDocuments
Progress68%
Open tasks24
Due this week6
Documents38
ALP-142Redesign the checkout flowIn reviewJHSep 24
ALP-139Clean up the onboarding email sequenceTodoSYSep 25
Space
The unit of work is not a project —
it's a Space.
Product, operations and marketing teams work to different rhythms. A Space accepts that difference as structure, so each team moves only inside its own context.
Product
Product Development · 8 projects · 12 members
Booking web revamp72%Active
Mobile app v2.138%Active
Payment module swap12%Planned
Requirement extraction14 items · done
IA draft5 pages · done
Design system32 tokens · running
Booking flow requirements v312 min ago
Design system tokens3 hours ago
June release notesYesterday
Reading of the cancellation policy8 AI sources
Competitor analysis12 sources
Accessibility checklist5 sources
Mon
Tue
WedMS 2
Thu
FriRelease
DDohyun Kim3 tasks completed
SSeoyeon Park2 documents written
JJunho Lee4 PRs merged
DHSYJHMSMJ+7
9 widgetsActive Projects · Recent Documents · Knowledge · Assets · AI Activity · Calendar · Team Activity · Members · Timeline
Space scopeEverything shown is limited to the current Space.
AI activity recordSee who ran which AI action, right on the dashboard.
Search Spaces
ProductProduct Development
Active Projects
Recent Documents
Knowledge
Assets
AI Activity
Calendar
Icon · color · descriptionYou tell Spaces apart by sight, not by reading names.
Template · members · project countKnow the size and character before you go in.
Recent activityPreview what happened last, right from the switcher.
SYSTEM TEMPLATES
Modules included · 6
OverviewProjectsKnowledgeDocumentsAssetsCalendarMembersOperations
Product Development · Product — From requirements to sprints to release
Preview which modules are included before you create it.
Knowledge · Documents · Assets
Knowledge, deliverables and resources
each have a clear role.
Lump everything together as 'files' and nobody knows where anything belongs. We separated the three roles and only let them link to each other.
DocumentsFinal deliverables
What you have agreed on: proposals, specs, reports. Versions are pinned and they are what you share and approve.
- Version · approval history
- Meant for sharing outside
- A settled state
AssetsWorkspace resources
Images, logos, icons, video, files, templates. Shared resources referenced from any project or document.
- Linked to projects · documents
- Preview by type
- Reusable templates
Enterprise Security
Security isn't a settings page —
it's a dashboard.
We turned the items a procurement review asks for into screens. Check the state, keep secrets, grant permission by layer, and leave a record of every change.
86
Security Score3 recommended actions left
MFA enforcedSSO readyAudit log 90 days
EncryptionAES-256 · TLS 1.3 in transitHealthy
MFA21 of 24 members activeAction needed
Active Sessions9 sessions · 3 devicesHealthy
Connected AppsSlack · GitHub · FigmaHealthy
Login History0 anomalies in the last 30 daysHealthy
API Keys6 keys · 1 expiring soonAction needed
Production DBPasswordInfra2 hours ago
Stripe Live KeyAPI KeyBillingYesterday
GitHub Actions TokenTokenPlatform3 days ago
deploy@apexstackSSH KeyInfra1 week ago
*.pjt.aiCertificateInfraExpires in 42 days
prod.envEnv VariablesPlatform5 hours ago
A 1Password-style table. Type, owner and last use on one line.
WorkspaceAdminSet directly
Space · ProductEditorSet directly
ProjectsEditorInherited
KnowledgeEditorInherited
DocumentsViewerSet directly
AssetsEditorInherited
MembersViewerSet directly
Permissions are inherited from above and narrowed only where you need it. Exceptions stay visible.
10:24Dohyun KimSpace permission changedProduct · Documents → Viewer
09:51Junho LeeAPI key issuedpjt-mcp-prod
09:12Sarah ChenVault item viewedStripe Live Key
Yest. 09:00SystemMFA not set alert3 members
AES-256 at restTLS 1.3 in transitSSO · SCIM (Business)Audit log kept 90 daysNever used for training
PJT does not stop at managing projects and tasks.
It connects plans, tech specs, tasks, documents, knowledge and work results into one project context, so people and AI agents can collaborate inside the same workflow.
With a general-purpose chat AI you often have to restate the project background, the requirements and earlier decisions every time you ask for something.
PJT is built so AI agents can work from the context already connected inside the project, which cuts down repeated explanation and context switching.
An AI agent can analyze, plan, execute and verify work based on the project context and the task assigned to it.
People set the direction and review the outcome, and the agent's results flow back into the project's work and knowledge.
No.
PJT is not about replacing people with AI — it is an environment where people and AI each hold a role and work together.
People own direction, judgement, priority and approval; AI agents support the analysis and execution of the work.
Yes. Existing projects, not just new ones, can be brought into PJT's work structure and carried on there.
Once a project's existing plan and work information is connected to PJT, it extends into tasks, documents, knowledge and the AI agent workflow.
Yes.
On a personal project you can manage projects, tasks, documents and knowledge, and work alongside an AI agent.
As the team grows you can share the same project and work context with everyone and collaborate.
Customer data is never used for AI training. It is encrypted with AES-256 at rest and TLS 1.3 in transit, and access is recorded in the audit log.
Yes. PJT runs on the web, and the desktop app gives you a workspace of its own.
Pick whichever fits the way you work.
Now go focus on the work.
Start projects, documents, tasks and AI
in a single workspace.