Idea Validation: ProofMate

Hi everyone! I’ve been sitting on an idea and I’d love to gather early feedback from you. The Problem As a software engineer, the annual performance review is one of those tasks that always ends up draining more time and energy than it should. Personal experience: Year 1 - I dug through Teams messages, Pull Requests, Jira tickets, and more, trying to piece together what I’d done. Year 2 - I started recording my "achievements" (with links) in Confluence throughout the year, but still had to look for extra details when writing my summary. Year 3 - Same as Year 2, plus ChatGPT polishing before pasting it into Workday. The Idea I have this idea. Let's call it "ProofMate". It will aggregate data from various platform for us, and generate a summary using AI. PlantUML Text @startuml actor User as Engineer package "ProofMate System" { rectangle "Data Collector" as Collector rectangle "AI Assistant" as AI rectangle "Self-Review Draft" as Draft } cloud "GitHub" as GitHub cloud "Jira" as Jira cloud "Confluence" as Confluence cloud "Microsoft Teams" as Teams cloud "Slack" as Slack Engineer --> Collector : Authenticate & Connect Accounts Collector --> GitHub : Fetch commits & PRs Collector --> Jira : Fetch tickets & issues Collector --> Confluence : Fetch documents Collector --> Teams : Fetch chat summaries Collector --> Slack : Fetch conversations Collector --> AI : Send structured activity data AI --> Draft : Generate draft review Engineer --> Draft : Review & Edit @enduml Considerations 1. Privacy Work data is often sensitive. Feeding raw material to AI models risks exposure. Should this be an offline-first app like jwt.io? More research is needed. 2. Data Retention The app could be designed to store only the final summary (perhaps even save directly to your Google Drive). No long-term logging of your raw work artifacts. 3. Security vs. User Friendliness Should API tokens be destroyed after each session? How can re-authentication be made as smooth as possible? Your thoughts I’d love to know: Does your company ask for self-reviews? How do you usually prep for them? Would you trust an AI to generate a draft based on your real work data? What integrations would be a must-have for you? I’d really appreciate your feedback!

Apr 13, 2025 - 09:19

Hi everyone!

I’ve been sitting on an idea and I’d love to gather early feedback from you.

The Problem

As a software engineer, the annual performance review is one of those tasks that always ends up draining more time and energy than it should.

Personal experience:

Year 1 - I dug through Teams messages, Pull Requests, Jira tickets, and more, trying to piece together what I’d done.

Year 2 - I started recording my "achievements" (with links) in Confluence throughout the year, but still had to look for extra details when writing my summary.

Year 3 - Same as Year 2, plus ChatGPT polishing before pasting it into Workday.

The Idea

I have this idea. Let's call it "ProofMate".

It will aggregate data from various platform for us, and generate a summary using AI.

PlantUML Text

@startuml
actor User as Engineer

package "ProofMate System" {
    rectangle "Data Collector" as Collector
    rectangle "AI Assistant" as AI
    rectangle "Self-Review Draft" as Draft
}

cloud "GitHub" as GitHub
cloud "Jira" as Jira
cloud "Confluence" as Confluence
cloud "Microsoft Teams" as Teams
cloud "Slack" as Slack

Engineer --> Collector : Authenticate & Connect Accounts
Collector --> GitHub : Fetch commits & PRs
Collector --> Jira : Fetch tickets & issues
Collector --> Confluence : Fetch documents
Collector --> Teams : Fetch chat summaries
Collector --> Slack : Fetch conversations

Collector --> AI : Send structured activity data
AI --> Draft : Generate draft review
Engineer --> Draft : Review & Edit

@enduml

Considerations

1. Privacy

Work data is often sensitive. Feeding raw material to AI models risks exposure. Should this be an offline-first app like jwt.io? More research is needed.

2. Data Retention

The app could be designed to store only the final summary (perhaps even save directly to your Google Drive). No long-term logging of your raw work artifacts.

3. Security vs. User Friendliness

Should API tokens be destroyed after each session?
How can re-authentication be made as smooth as possible?

Your thoughts

I’d love to know:

Does your company ask for self-reviews?
How do you usually prep for them?
Would you trust an AI to generate a draft based on your real work data?
What integrations would be a must-have for you?

I’d really appreciate your feedback!