Event intelligence / product design

Signal clarity before the daily deadline.

Because this project is under NDA, this portfolio version shows the problem, workflow, design approach, high-level architecture and measurable impact using sanitized language, abstracted diagrams and generalized examples.

Sanitized interface modelPrivate work

Daily intelligence brief

Evidence reviewed. Claims checked. Ready for editorial approval.

10week product strategy and design sprint
9average hours saved each week
3 PMdaily publishing target
Privategeneralized flows without production artifacts
01High-value signal cluster identified from multiple public and internal inputs.High
02Evidence check flags weak support before the brief reaches stakeholders.Review
03Final brief captures what happened, why it matters and what to watch next.Ready

Role and context

Designed at JOJA as confidential product work.

I worked as the product strategist and designer, turning a manual intelligence workflow into a structured product concept for collecting event signals, evaluating evidence and producing a credible daily brief.

Disclosure note

I do not show production screenshots, proprietary source lists, prompts, real outputs, internal data, client-specific details or implementation internals. The case study focuses on my role, process, workflow decisions and approved impact.

Company
JOJA

Company name is safe to show as workplace context.

My role
Product strategist and designer

Defined the use case, workflow, information architecture and product narrative.

Timeline
10 weeks

From problem framing through product experience and case-study structure.

Outcome
9 hours saved weekly

Reduced the average time needed to review signals, connect evidence and draw insights.

Concept flow

From scattered signals to a reviewed daily brief.

The case centers on one repeatable job: help a strategy team understand what matters during a live program, explain why it matters and publish a source-backed brief before the daily deadline.

Session transcript answer flow

The public version of the transcript Q&A logic shown in Carbon system styling.

Transcript retrieval
Input
User question
Scope
Current session or full archive?
Choose the retrieval boundary before searching.
Full archive
All session transcripts
Current session
Current transcript
Retrieval
Retrieve relevant passages
Return the evidence excerpts that answer the prompt.
AI synthesis
Synthesize answer
Cited output
Answer with source links
Follow-up
Suggested follow-up questions
The diagram recreates the transcript flow without exposing production prompts, source indexes or client data.

End-to-end product flow

The safe public version of the product logic.

Primary flow
01 / Setup
Configure workspace
Dates, audience, topics and output format.
02 / Context
Map the program
Turn the event structure into reviewed operating context.
03 / Plan
Create watch plan
Translate priorities into sessions, sources and tasks.
04 / Synthesis
Synthesize signals
Deduplicate, classify and cluster evidence.
05 / Approval
Review and brief
Check claims, edit output and archive the record.
The public diagram shows the workflow shape without exposing proprietary mechanics.

Daily operating loop

The repeated rhythm behind the time savings.

Live mode
Scan
Collect signals
Capture
Add context
Clean
Normalize inputs
Draft
Generate takeaways
Review
Check claims
Send
Publish brief
Carry unresolved signals into the next daily scan.
The loop saves time by replacing open-ended research with a structured review path.

Problem

The work was not finding information. It was making sense of it fast.

During a live program, meaningful signals appear across schedules, sessions, social commentary, announcements, media coverage and internal observations. The team needed a way to move from fragmented inputs to a trusted daily point of view without spending hours manually reading, comparing and synthesizing.

Before

Manual scanning across many sources, followed by ad hoc synthesis.

After

A structured flow that captures signals, evaluates evidence and produces a reviewed daily brief.

Impact: average 9 hours saved each week

What I designed

A safer summary of the product work.

The public case study focuses on design decisions and workflow logic rather than production details. These are the parts that can be shown without exposing confidential project material.

01

Product framing

Defined the core job to be done: produce timely, evidence-backed event intelligence for a team working against a daily deadline.

02

Workflow design

Mapped the process from setup and watch planning through capture, synthesis, review and publication.

03

Information architecture

Organized the experience around workspace setup, signal capture, daily output, archive and review states.

04

Evidence model

Designed guardrails for confidence, attribution, weak claims and human approval before a brief is shared.

High-level architecture

The architecture can be shown as capability blocks, not screens.

To keep the project private, I represent the system at the level of inputs, processing stages, review gates and outputs. This shows the product thinking without exposing production UI, source strategy, data, prompts or implementation details.

01 / Setup layer

Workspace and priorities

Defines the event context, audience, deadline, watch priorities and output expectations.

02 / Signal layer

Capture and normalization

Turns public inputs and team observations into structured evidence that can be reviewed.

03 / Intelligence layer

Synthesis and ranking

Clusters related signals, identifies themes and ranks takeaways by relevance and evidence strength.

04 / Review layer

Human approval

Flags weak claims, missing support and unclear attribution before the brief is shared.

05 / Output layer

Daily brief and archive

Publishes the daily view and preserves the record for later synthesis.

Session video intelligence

Long-form sessions become cited evidence.

The product also included a video intelligence capability. In the public case study, I show it as an architecture flow rather than a screen.

Session assetRecorded or captured long-form session material enters the evidence pipeline.
TranscriptSpeech becomes searchable text with time-based references.
SegmentingLong content is broken into coherent idea-level chunks.
RetrievalQuestions can pull back relevant cited transcript moments.
Brief evidenceUseful session signals can support daily takeaways with attribution.

Impact and reflection

Less manual synthesis, more decision-ready context.

The clearest outcome was time saved: the workflow reduced the average effort required to go through signals, connect evidence and draw useful insights by 9 hours each week.

9 hrs/week

Average weekly time saved by replacing scattered research and manual synthesis with a structured capture, review and brief-generation workflow.

What changed

The team could move from open-ended signal review to a repeatable daily operating rhythm.

What I learned

For AI-assisted strategy tools, the design challenge is not only output quality. It is trust, reviewability and clear human control.

What I can show

My role, process, workflow diagrams, generalized product structure and approved impact metrics.

What stays private

Production screens, real data, proprietary workflows, source strategy, prompts, internal architecture and client-specific outputs.

Previous projectGarden Strategy Agent <- HomeBack to Home -> Next projectPolis Design System ->