growqr-app/docs
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
main
docs/workflow-primitives-inventory.md
-Puter e6685203fe feat: initial docs repo with project inventory and all documentation 
- Added REPO_INVENTORY.md with all repos, branches, remotes, and staging info
- Added .gitignore
- Synced all existing docs from local workspace
- Centralized documentation hub for GrowQR team
2026-06-22 15:04:27 +05:30

1016 lines
22 KiB
Markdown
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# GrowQR Workflow Primitives Inventory
This document lists the basic building blocks GrowQR has available before defining sellable workflows. The goal is to separate **capabilities/primitives** from **packaged workflows**.
A workflow should be composed from these primitives instead of being treated as a standalone microservice.
---
## 1. Product Framing
GrowQR is moving from a collection of user-facing microservices to an **agentic workflow platform**.
Old framing:
> User buys or uses individual tools: resume builder, interview tool, roleplay tool, matchmaking, pathways, social branding, etc.
New framing:
> User buys an outcome workflow. The Grow Agent coordinates internal capabilities, service agents, human experts, and persistent memory to deliver that outcome.
The microservices still matter, but they become internal capabilities. The product surface becomes workflows such as job search, career discovery, interview readiness, personal branding, promotion readiness, or career switching.
---
## 2. Platform / Architecture Primitives
These are not career products by themselves, but they are the infrastructure required to run workflows reliably.
### 2.1 Main Grow Agent
**Purpose:** User-facing coordinator for all workflows.
**Responsibilities:**
- Understand user intent.
- Select the correct workflow.
- Break workflow into steps.
- Call the correct service agents/tools.
- Ask for user approvals where needed.
- Stream progress back to the UI.
- Persist memory and artifacts.
- Resume interrupted workflows.
**Workflow value:** This is the single conversational/product interface. Users should not need to know which microservice is being used.
---
### 2.2 Durable Control Plane / Actor Layer
**Purpose:** Keep long-running workflow state durable and resumable.
**Responsibilities:**
- Per-user or per-workflow actor identity.
- Active thread/session state.
- Runtime lifecycle decisions.
- Recovery pointers.
- Workflow serialization and resume.
- WebSocket/event streaming to frontend.
**Important distinction:** Actors hold small durable state and coordination metadata. They do not hold the full execution environment.
---
### 2.3 Agent Runtime Sandbox
**Purpose:** Disposable execution environment where the agent can reason, call tools, read/write files, and invoke service capabilities.
**Responsibilities:**
- Execute agent steps.
- Call domain tools and service APIs.
- Maintain isolated workspace/worktree.
- Generate intermediate artifacts.
- Emit progress events.
**Workflow value:** Lets workflows be more than API chains. The runtime can plan, inspect files, revise outputs, and coordinate multi-step work.
---
### 2.4 Capability Manifest / Tool Registry
**Purpose:** Standard way for services to advertise what they can do.
**Current contract:** Services can publish capabilities at:
```text
/.well-known/growqr-capabilities.json
```
**Capability examples:**
- `resume.analyze`
- `resume.optimize`
- `pathways.generate`
- `matchmaking.feed`
- `interview.configure`
- `roleplay.review`
**Workflow value:** Workflows can call capabilities by intent instead of hardcoding service-specific endpoints.
---
### 2.5 Event Stream / Progress Protocol
**Purpose:** Real-time UI feedback during workflow execution.
**Useful event types:**
- Workflow started
- Agent thinking
- Step started
- Step completed
- Approval required
- Artifact generated
- Score updated
- Recommendation generated
- Error/retry
- Workflow completed
**Workflow value:** Makes long-running workflows feel alive and trustworthy.
---
### 2.6 Versioned Memory / Git-Backed User Memory
**Purpose:** Persistent source of truth for user career context and generated artifacts.
**Potential memory contents:**
- Career goals
- Resume versions
- LinkedIn/profile rewrites
- Job preferences
- Application tracker
- Interview feedback history
- Roleplay feedback history
- Q-Score snapshots
- Pathway plans
- Weekly progress logs
- Approved/rejected content
**Workflow value:** Every workflow can build on previous work instead of starting from zero.
---
### 2.7 Operational Database
**Purpose:** Store operational metadata, indexes, status, and recovery pointers.
**Likely contents:**
- User records
- Workflow run records
- Actor/runtime pointers
- Payment/subscription state
- Service job status
- Artifact indexes
- Feed/action history
- Provider bookings
- Score history
**Important distinction:** The database stores operational state; durable user knowledge and versioned artifacts should live in memory/files where possible.
---
## 3. User / Identity Primitives
### 3.1 Authenticated User
**Purpose:** Known user identity, session, permissions, and tier.
**Sources:**
- Clerk authentication
- GrowQR user-service profile
**Workflow inputs:**
- User ID
- Email/name
- Account tier
- Subscription/purchase status
- Connected accounts
- Region/location
---
### 3.2 User Profile
**Purpose:** Base professional context.
**Possible fields:**
- Current role
- Years of experience
- Industry/domain
- Education
- Skills
- Projects
- Certifications
- Work preferences
- Career goals
- Target role/company/industry
- Location and mobility preferences
**Used by:** Almost every workflow.
---
### 3.3 Career Goals & Preferences
**Purpose:** User-stated intent and constraints.
**Examples:**
- Target roles
- Target industries
- Desired salary
- Remote/hybrid/office preference
- Location constraints
- Timeline urgency
- Fulfillment vs earnings preference
- Work culture preferences
- Growth priorities
**Used by:** Pathways, matchmaking, resume targeting, interview prep, course recommendations.
---
## 4. Core Domain Service Primitives
## 4.1 Pathways Service
**Primitive type:** Career direction and journey orchestration.
**Core capabilities:**
- Ingest profile context from LinkedIn/resume/questionnaire.
- Generate career options.
- Group options as close match, adjacent, and stretch paths.
- Produce career identity statement.
- Generate visual career web data.
- Map skill gaps per path.
- Activate a selected pathway.
- Generate weekly plans and tasks.
- Produce pathway report/export.
- Re-plan as Q-Scores and user activity change.
**Inputs:**
- Resume/LinkedIn/profile data
- Questionnaire
- Q-Score profile
- Career goals
- Region/labor market context
**Outputs/artifacts:**
- Career web
- Career options
- Skill gap map
- Weekly plan
- Pathway report
- Pathway tasks
**Workflow role:** Direction-setting engine. Determines what the user should work toward and what other services should do next.
---
## 4.2 Q-Score Service
**Primitive type:** Measurement and scoring engine.
**Core capabilities:**
- Calculate user score from multiple signals.
- Score profile/resume/education/engagement/goals.
- Produce score breakdowns.
- Track score changes over time.
- Feed recommendations into Pathways and other services.
**Current signal sources:**
- LinkedIn profile
- Resume/CV upload
- Education documents
- Engagement activity
- Goals and self-assessment
- Cover letter / additional career artifacts
- Platform badges and points
**Representative Q dimensions mentioned in docs:**
- Social Quotient
- Execution Quotient
- Communication Quotient
- Vision Quotient
- Drive Quotient
- Growth Quotient
- Reputation Quotient
- Domain Quotient
The broader product vision references 25 Q-Scores.
**Outputs/artifacts:**
- Q-Score
- Pillar breakdown
- Readiness tier
- Percentile/rank where available
- Score trend
- Improvement recommendations
**Workflow role:** Measurement layer. It tells workflows what to prioritize and proves progress to the user.
---
## 4.3 Resume Builder / Resume Intelligence
**Primitive type:** Resume creation, parsing, versioning, optimization, export.
**Core capabilities:**
- Create resumes.
- Store resume versions.
- Use templates.
- Parse uploaded resumes.
- Analyze resume quality.
- Improve ATS compatibility.
- Rewrite bullet points.
- Tailor resume for roles/job descriptions.
- Export to PDF.
**Inputs:**
- Existing resume
- User profile
- Target role
- Job description
- Pathway target
- Q-Score gaps
**Outputs/artifacts:**
- Resume draft
- Resume versions
- ATS analysis
- Keyword gap analysis
- Role-specific resume variant
- PDF export
**Workflow role:** Converts user history into strong application material.
---
## 4.4 Matchmaking Service
**Primitive type:** Opportunity discovery and ranking.
**Core capabilities:**
- Store opportunities.
- Support multiple opportunity types.
- Capture user preferences.
- Rank opportunities against user profile/preferences.
- Explain match scores.
- Maintain feedback loop from user actions.
- Recompute feeds.
- Support employer/organization opportunity ownership in current slice.
**Opportunity types:**
1. Full-time jobs
2. Contracts/gigs
3. Events/meetups
4. Board advisory roles
5. Surveys/research
6. Temporary earning opportunities
7. Future pathway opportunities
**Scoring components from docs:**
- Skills match: 40%
- Personality fit: 30%
- Company culture: 15%
- Career growth: 10%
- Compensation: 5%
**Feedback actions:**
- View
- Save
- Dismiss
- Apply
- RSVP
**Outputs/artifacts:**
- Opportunity feed
- Match percentage
- Explanation scores
- Saved/applied/dismissed history
- Feed history
**Workflow role:** Finds external opportunities that match the user and learns from their behavior.
---
## 4.5 Interview Service
**Primitive type:** Live mock interview simulation and evaluation.
**Core capabilities:**
- Configure interview session.
- Select interviewer persona and interview type.
- Generate questions based on role/context.
- Conduct live voice interview.
- Transcribe candidate/interviewer turns.
- Archive audio artifacts.
- Generate post-interview review.
- Score performance and recommend improvements.
**Inputs:**
- Target role
- Job description
- Resume
- Interview type
- Duration
- User history/Q-Score
**Outputs/artifacts:**
- Interview session
- Transcript
- Audio artifact
- Interview review
- Scores
- Recommendations
**Workflow role:** Converts preparation into simulated practice and measurable readiness.
---
## 4.6 Roleplay Service
**Primitive type:** Workplace communication simulation and coaching.
**Core capabilities:**
- Generate roleplay preview/plan.
- Allow user approval before start.
- Run live audio-first roleplay with avatar persona.
- Support workplace scenarios.
- Enforce timing/moderation.
- Capture transcript/audio.
- Generate rubric-based review.
- Provide improvement roadmap and historical comparison.
**Example scenarios:**
- Sales objections
- Customer success conversations
- Support diagnosis
- Stakeholder conflict
- Salary negotiation
- Difficult manager conversation
- Founder/investor pitch practice
- Custom professional scenario
**Evaluation metrics:**
- Voice and tone
- Content quality
- Scenario adherence
- Adaptability
- Emotional intelligence
- Technical/session quality
**Outputs/artifacts:**
- Roleplay plan
- Transcript/audio
- Rubric scores
- Review
- Improvement roadmap
- Trend data
**Workflow role:** Practice layer for soft skills and workplace performance.
---
## 4.7 Assessment Service
**Primitive type:** AI-generated assessments and grading.
**Core capabilities:**
- Generate assessments from text, topic, YouTube URL, or course JSON.
- Create STAR-framework scenario MCQs.
- Support manual assessments.
- Grade submissions.
- Support multiple question types including multiple choice, coding, and scenario.
- Process jobs asynchronously.
**Inputs:**
- Raw text
- Topic
- YouTube URL
- Course module JSON
- Manually authored questions
- Difficulty
**Outputs/artifacts:**
- Assessment
- Questions
- User submission
- Score/result
- Skill evidence
- Completion status
**Workflow role:** Diagnosis and checkpoint layer. Verifies whether user has learned or is ready.
---
## 4.8 Courses Service
**Primitive type:** Learning recommendation and courseware layer.
**Core capabilities:**
- Curate/index external courses.
- Map courses to skill gaps and user personas.
- Track enrollment status.
- Track progress/completion.
- Generate certificates where eligible.
- Support future third-party creator content.
- Support future GenAI-generated course IP.
**Three-phase courseware strategy:**
1. External course mapping and indexing
2. Third-party creator platform
3. GenAI-generated personalized course IP
**Inputs:**
- Skill gaps
- Pathway target
- Q-Score breakdown
- Assessment/interview/roleplay feedback
- User persona/life stage/industry
**Outputs/artifacts:**
- Recommended courses
- Learning path
- Enrollment record
- Completion status
- Certificate
**Workflow role:** Turns identified gaps into structured learning actions.
---
## 4.9 Social Branding Service
**Primitive type:** Professional presence and content engine.
**Core capabilities:**
- Connect social/professional accounts.
- Scrape/collect profile and engagement data.
- Analyze profile completeness and audience alignment.
- Calculate Brand Score.
- Rewrite profiles.
- Generate content calendars.
- Generate post drafts.
- Queue items for approval.
- Schedule/post approved content.
- Track performance and feed data back into RQx/Q-Score.
**Target platforms:**
- LinkedIn
- Instagram
- X/Twitter
- YouTube
- Git/profile surfaces where relevant
**Personas from docs:**
- Job seeker
- Career transitioner
- Thought leader
- Founder/entrepreneur
- Creator/consultant-style personas
- Enterprise/company brand contexts
**Brand Score components:**
- Profile completeness
- Content consistency
- Audience growth
- Engagement rate
- Recruiter visibility
- Content quality
- Cross-platform consistency
**Outputs/artifacts:**
- Brand Score
- LinkedIn/profile rewrite
- Content strategy
- Content calendar
- Post drafts
- Approval queue
- Analytics report
**Workflow role:** Makes the user externally visible and credible.
---
## 4.10 Marketplace Service
**Primitive type:** Human expert supply layer.
**Core capabilities:**
- Provider onboarding and validation.
- Provider profiles.
- Service listings.
- Booking flows.
- Availability/calendar.
- Reviews and ratings.
- Provider categories mapped to career needs.
- Pathway-triggered recommendations.
**Provider types:**
- GrowQR-certified external coaches
- Verified senior users
- Corporate/institutional providers
- GrowQR in-house experts
**Delivery formats:**
1. Live 1:1 video session
2. Async review
3. Group workshop
4. Done-for-you service
5. Retainer package
**Service categories:**
- Interview preparation
- Resume and cover letter
- Social and personal branding
- Career coaching
- Job search strategy
- Domain and industry guidance
- Skills development coaching
- Entrepreneurship and business
- Academic and graduate guidance
**Outputs/artifacts:**
- Provider recommendation
- Booking
- Session notes
- Async review result
- Follow-up action plan
**Workflow role:** Brings in human wisdom when AI is not enough or trust is required.
---
## 4.11 Dashboard Service
**Primitive type:** User progress and home surface.
**Core capabilities:**
- Load user dashboard state.
- Show recent progress.
- Display Q-Score and updates.
- Surface recommendations.
- Present active workflows/pathways.
- Show tasks, suggestions, and activity.
**Outputs/artifacts:**
- Dashboard summary
- Progress cards
- Recommendations
- Current task list
- Score/activity widgets
**Workflow role:** The user’s command center for active workflows and progress.
---
## 4.12 User Service
**Primitive type:** Identity, profile, and user source of truth.
**Core capabilities:**
- Ensure/create user record.
- Store user profile.
- Store account-level metadata.
- Manage profile photo/QR-related identity surfaces.
- Expose user state to frontend and orchestrator.
**Outputs/artifacts:**
- User profile
- Account tier/pro status
- Basic identity and preferences
**Workflow role:** Baseline identity and profile context for all workflows.
---
## 4.13 Frontend / Client Experience
**Primitive type:** User interface and approval surface.
**Core capabilities:**
- Chat-first interaction.
- Generated UI cards.
- Workflow progress display.
- Approval/rejection/edit flows.
- Direct page experiences for some services.
- Realtime WebSocket session with orchestrator/actor.
**Workflow role:** Where users see, approve, and trust the workflow execution.
---
## 5. Cross-Cutting Data / Artifact Primitives
These are the reusable objects workflows can create, consume, update, and version.
### 5.1 Profile Artifacts
- User profile snapshot
- LinkedIn profile snapshot
- Social profile snapshot
- Education documents
- Skills inventory
- Career goals/preferences
### 5.2 Resume Artifacts
- Uploaded resume
- Parsed resume JSON
- Resume draft
- Resume versions
- Job-specific resume variants
- ATS analysis
- PDF export
### 5.3 Career Direction Artifacts
- Questionnaire response
- Career identity statement
- Career web
- Pathway options
- Skill gap map
- Activated pathway
- Weekly plan
- Pathway report
### 5.4 Opportunity Artifacts
- Opportunity record
- Opportunity feed
- Match score
- Fit explanation
- Saved/applied/dismissed actions
- Application tracker
- Employer/company enrichment
### 5.5 Practice Artifacts
- Interview config
- Interview transcript
- Interview review
- Interview audio artifact
- Roleplay plan
- Roleplay transcript
- Roleplay review
- Practice trend history
### 5.6 Learning / Assessment Artifacts
- Course recommendation
- Learning path
- Enrollment record
- Course completion
- Certificate
- Assessment questions
- Assessment submission
- Assessment result
### 5.7 Branding Artifacts
- Brand Score
- Profile audit
- LinkedIn/profile rewrite
- Content pillars
- Content calendar
- Post drafts
- Approval queue
- Published content history
- Engagement analytics
### 5.8 Human Support Artifacts
- Provider profile
- Provider recommendation
- Booking
- Session notes
- Async review
- Follow-up plan
### 5.9 Score / Progress Artifacts
- Q-Score snapshot
- Q-Score breakdown
- Brand Score history
- Readiness score
- Weekly progress log
- Workflow completion report
---
## 6. Approval Primitives
Many workflows should not execute irreversible actions without approval.
### Approval points to support
- Approve resume rewrite/version.
- Approve job applications before submission.
- Approve recruiter outreach messages.
- Approve social profile changes.
- Approve social posts before publishing.
- Approve roleplay/interview plan before session starts.
- Approve paid human expert booking.
- Approve pathway activation.
- Approve final report/export.
### Approval actions
- Approve
- Reject
- Request revision
- Edit directly
- Save for later
- Auto-approve rule for low-risk repeated actions
---
## 7. External Integration Primitives
### 7.1 Professional / Social Platforms
- LinkedIn
- Instagram
- X/Twitter
- YouTube
- Git/profile surfaces where relevant
### 7.2 Job / Opportunity Sources
- LinkedIn Jobs
- Naukri
- Indeed
- Glassdoor
- Monster
- Company career pages
- Employer direct posts
- Event/meetup sources
- Advisory/survey/gig sources
### 7.3 Learning Platforms
- Coursera
- LinkedIn Learning
- Udemy
- Moodle
- Thinkific
- Other indexed course providers
### 7.4 Communication / Live Session Surfaces
- Live voice/video interview sessions
- Live roleplay sessions
- Marketplace expert sessions
- Calendar/reminder integrations later
---
## 8. Commercial / Packaging Primitives
These primitives help determine how workflows can be sold.
### 8.1 User Tiers
From Pathways docs:
- Free/basic
- One-time report purchase
- Premium subscription
### 8.2 Purchase Models
Potential models:
- Free diagnostic
- One-time workflow purchase
- Subscription workflow access
- Premium bundle
- Human expert add-on
- Credit-based usage
- Outcome-guarantee package where appropriate
### 8.3 Gating Concepts
- Limited free nodes/recommendations
- Full report unlock
- Workflow activation unlock
- Unlimited opportunity feed for premium
- Human marketplace add-on
- Advanced Q-Score insights for premium
- Longer pathway durations for subscription users
---
## 9. Current Service Inventory Summary
| Primitive / Service | Main Role | Key Outputs |
|---|---|---|
| Main Grow Agent | User-facing workflow operator | Plans, tool calls, progress, artifacts |
| Durable Actor Layer | Resumable workflow control | State, recovery pointers, event stream |
| Runtime Sandbox | Agent execution environment | Tool execution, generated files, service calls |
| User Service | Identity/profile source | User record, profile, tier |
| Dashboard Service | Progress command center | Dashboard cards, recommendations, active tasks |
| Pathways | Career direction/orchestration | Career web, pathway, weekly plan, report |
| Q-Score | Measurement engine | Scores, breakdowns, trends, recommendations |
| Resume Builder | Resume intelligence | Resume versions, analysis, exports |
| Matchmaking | Opportunity discovery | Feed, match scores, feedback history |
| Interview | Mock interview practice | Live session, transcript, review, score |
| Roleplay | Workplace simulation | Scenario practice, transcript, rubric, roadmap |
| Assessment | Skill/readiness testing | Assessment, graded result, evidence |
| Courses | Learning recommendations | Course plan, enrollment, certificates |
| Social Branding | Professional visibility | Brand Score, profile rewrite, content calendar |
| Marketplace | Human expert supply | Provider booking, session notes, expert reviews |
| Frontend | User interaction/approval | Chat, cards, approvals, progress UI |
| Versioned Memory | Durable user knowledge | Files, diffs, artifacts, history |
| Operational DB | Workflow/service metadata | Status, indexes, pointers, transactions |
---
## 10. Notes for Next Step: Workflow Design
When defining actual workflows, each workflow should specify:
- User promise
- Target user segment
- Required inputs
- Services/capabilities used
- Step sequence
- Approval points
- Generated artifacts
- Success metrics
- Pricing model
- Memory written
- Retry/resume behavior
A workflow should be packaged around a user outcome, not around a service name.
Example:
```yaml
workflow_id: job_search_apply
promise: Find relevant jobs, tailor application material, and prepare the user to apply.
uses:
- user_profile
- qscore
- resume_builder
- matchmaking
- social_branding
- interview
- marketplace_optional
approval_points:
- approve_resume_variant
- approve_application_targets
- approve_outreach_messages
outputs:
- curated_job_list
- tailored_resume
- cover_letter_or_message
- application_tracker
- interview_prep_plan
```
This inventory should be treated as the base map for building the actual GrowQR workflow catalog.
Reference in New Issue View Git Blame Copy Permalink