PulseCheck

PulseCheck

Less time reporting. More time moving work forward.

An AI-powered internal tool that turns weekly check-ins into a shared view of progress, blockers, and priorities.

COMPANY | Spectro Cloud

ROLE | UX Design Intern (second summer internship)

TEAM | Two software engineers, a product manager, and the CTO

DURATION | 10 weeks, Summer 2026

TOOLS | Figma, Figma Make, Claude, Cursor, Lovable, ChatGPT

OUTCOME | Launched company-wide on June 12, 2026. Reached 80% weekly participation.

Turning scattered updates into shared context

When I returned to Spectro Cloud for a second summer, the CTO brought me onto PulseCheck to address an internal problem: progress updates were spread across standups, Slack, and project tools. Getting a clear picture of what was moving forward or blocked meant piecing information together from different places.

PulseCheck is built around one weekly workflow. Contributors submit a check-in, an Analysis Agent synthesizes the updates, and a dashboard brings progress and priorities together across the organization.

I designed seven product surfaces: Check In, Manager Dashboard, Projects, Performance, Org Chart, Ask AI, and Feedback. I worked closely with the product manager, the CTO, and two engineers.

AFTER

BEFORE

Updates scattered across Slack, Jira, monday.com, and GitHub come together in one weekly view.

One check-in, structured for useful summaries

The check-in asks four questions, each capturing context the Analysis Agent needs to produce a useful summary:

Project status: Current state and risks
Progress this week: Work completed
Risks and mitigations: Blockers and responses
What’s up next: Upcoming priorities

The same form gives contributors a lightweight way to report their work and gives managers a consistent view across the team.

The check-in asks four questions, each capturing context the Analysis Agent needs to produce a useful summary:

Project status: Current state and risks
Progress this week: Work completed
Risks and mitigations: Blockers and responses
What’s up next: Upcoming priorities

The same form gives contributors a lightweight way to report their work and gives managers a consistent view across the team.

Every contributor answers the same four prompts, so updates can be compared and summarized across teams.

PERSONAL CONTEXT

TEAM OVERVIEW

A weekly rhythm instead of a constant feed

Check-ins close on Sunday, and the Analysis Agent runs on Monday morning. Batching updates creates one predictable moment to review progress and blockers, rather than a feed that never stops changing.

Check-ins close Sunday, analysis runs Monday morning, and teams start the week from the same snapshot.

Redesigning the dashboard for teams of every size

The first dashboard placed everything in one long view. Smaller teams were left with too much empty space, and larger teams were hard to scan. I reorganized it into focused tabs for Blockers, Upcoming Work, Key Initiatives, Team Check In, and Performance, with a more flexible team breakdown.

VERSION 1 DASHBOARD

VERSION 2 DASHBOARD

Before and after: one long view becomes focused tabs with a flexible team breakdown.

Following work across team boundaries

Individual and team-level updates couldn’t show how cross-team initiatives were progressing. I designed Projects to close that gap, bringing an initiative’s status, blockers, recent activity, and relevant check-in mentions into one place.

Individual and team-level updates couldn’t show how cross-team initiatives were progressing. I designed Project View to close that gap, bringing an initiative’s status, blockers, recent activity, and relevant check-in mentions into one place.

Mentions from different teams’ check-ins roll up under the initiative they relate to.

Putting engineering signals in context

I extended the dashboard with selected DORA-style engineering indicators (common measures of software delivery performance), alerts, AI-flagged patterns, and a custom chart builder. The goal was to place engineering signals alongside weekly reporting, not to replace existing analytics tools.

The chart builder, with an AI-suggested filter based on team conventions.

Bringing AI into the existing workflow

Ask AI adds a conversational layer to Dashboard, Org Chart, and Projects, so users can explore information inside the view they’re already using.

Ask AI adds a conversational layer to Dashboard, Org Chart, and Project View, so users can explore information inside the view they’re already using.

Questions are answered inside the current view, so users don’t need to leave what they’re looking at.

I also collaborated with the two software engineers on an MCP integration that extends check-in preparation into Claude. The engineers led the implementation; working alongside them, I learned how the integration worked and explored how it fit into the check-in experience. Contributors can draft updates from material already in Google Docs, Slack, and GitHub pull requests, while PulseCheck remains where reports are submitted.

A check-in drafted in Claude from existing work, then prepared for submission in PulseCheck.

When visibility crossed a boundary

On launch day, feedback revealed that Ask AI had auto-filled a check-in with content from private direct messages. The information was retrievable, but it ended up somewhere it didn’t belong.

Ask AI drafts a check-in using content from a private direct message.

The incident showed that information is only useful when it reaches the right person in the right context. I refined the visibility model so Ask AI and Org Chart follow the same hierarchy as the rest of the product:

Individual contributors see their own work and their team’s work.
Managers and leadership can drill down through team summaries to find specific answers about work across their organization.
Confidential teams, such as HR and Finance, work differently: an individual contributor’s work is visible only to their manager, not to teammates. Managers keep visibility into their team’s work.

The experience reinforced that privacy and access boundaries are part of the product, especially when AI brings together information from different sources.

Launch and results

PulseCheck launched company-wide on June 12, 2026, and reached 80% weekly participation. Feedback from the product, Slack, and team conversations continued to shape refinements after launch.

Participation shows engagement with the reporting workflow. It doesn’t establish time saved or better decision-making, which would need to be measured separately.

PulseCheck launched company-wide on June 12, 2026, and reached 80% weekly participation. Feedback from the product, Slack, and team conversations continued to shape refinements after launch.

Participation shows engagement with the reporting workflow. It doesn’t establish time saved or better decision-making, which would need to be measured separately.

What I learned

Design for the range, not just the average. Different team sizes and reporting needs should shape the information architecture from the start, not after the first version.

Treat privacy as a product constraint. AI access should be built around roles, permissions, and context from day one.

Measure outcomes alongside adoption. Next, I’d evaluate reporting effort, how useful updates are, and how early teams identify blockers.

PulseCheck brought together interaction design, information architecture, AI-assisted workflows, and cross-functional product development. It reinforced that making information easier to retrieve isn’t enough. It also needs to be useful, understandable, and visible only to the right people.

Design for the range, not just the average. Different team sizes and reporting needs should shape the information architecture from the start, not after the first version.

Treat privacy as a product constraint. AI access should be built around roles, permissions, and context from day one.

Measure outcomes alongside adoption. Next, I’d evaluate reporting effort, how useful updates are, and how early teams identify blockers.