Settings - INtelliBOX

⚙️ Settings

Program / Project Name

Shown in the navigation banner alongside INtelliBOX. Leave blank to hide.

Default Priority

Fallback for actions that don't match any rule below.

Due Date Threshold

Escalate to HIGH when due date is within this many days.

days or fewer

High Priority Senders

Emails from these addresses always escalate to HIGH.
One per line — full address or @domain.com.

High Priority Keywords

Subject or body containing these words escalates to HIGH.
One per line. Case-insensitive.

Confidence Threshold

Actions with AI confidence below this score are discarded.
0.0 = keep everything · 1.0 = only certain actions.

0.50

Timezone

Used for timestamps in email reports.

Changes apply to the next email processed.

▸ How priority rules are evaluated

AI analyzes the email and suggests an initial priority
Rules are checked in order: sender match → keyword match → due date threshold → default
The first matching rule wins and sets the final priority

3 members

Excel format: columns named first_name, last_name, email

Last Name	First Name	Email
Perkins	Blake	b.p@example.com
Russo	Nick	n.r@example.com
Vargas	Annika	a.v@example.com

6 categories

Name	Description
RFI	Request for Information requiring a data or document response
data_call	Request to provide specific data, metrics, or a report
stakeholder_request	Request from an executive, client, or external stakeholder
meeting_action	Action item assigned during or following a meeting
deliverable	A work product or artifact that must be produced and submitted
other	Any action item that does not fit the above categories

Insights System Prompt

This prompt is sent to GPT-4 when generating the Insights page analysis. Use {placeholder} syntax for dynamic data.

You are generating a program intelligence briefing for a director. Cross-reference the action items below against the Program Knowledge Base documents provided in your system instructions.

## Current Action Pipeline

{action_details}

## Pipeline Metrics

- Total actions: {total_actions}
- Awaiting assignment: {unassigned_count} (High: {high_priority_count}, Medium: {medium_priority_count}, Low: {low_priority_count})
- In progress: {assigned_in_progress}
- Completed: {completed_count}
- Overdue: {overdue_count}
- Due this week: {due_this_week_count}

## Recent Email Traffic ({days} days)

{email_summaries}

## Analysis Instructions

Cross-reference the actions and emails against the Program Knowledge Base documents. Generate analysis in these four areas:

1. **SOW Alignment** — Map actions to deliverables, CLINs, or milestones from program documents. Flag actions that don't match any documented scope (potential scope creep) and documented deliverables with no matching actions (coverage gaps).

2. **Risk Radar** — Identify risks by cross-referencing actions against known issues, deadlines, or concerns from program documents. Include severity and suggested mitigation.

3. **Process Compliance** — Compare workflow patterns (triage speed, assignment rates, priority distribution) against documented team processes and procedures.

4. **Trend Intelligence** — Connect action patterns to program context. What do the current actions tell us about program trajectory?

Be specific and factual. Use exact numbers from the metrics above. Name the specific program document that informed each finding.

Respond with this JSON:

{{
  "executive_summary": {{
    "headline": "Short, punchy status line — max 12 words. Lead with the most critical number or finding. Example: '4 of 5 SOW deliverables at risk — 27 actions unassigned'",
    "key_finding": "One sentence: the most important insight from cross-referencing actions against program documents. Name the specific document.",
    "top_risk": "One sentence: the single biggest risk right now, with evidence.",
    "recommended_action": "One sentence: the single most important thing the director should do today."
  }},
  "sow_alignment": {{
    "mapped_deliverables": [
      {{
        "deliverable": "SOW deliverable or CLIN name",
        "related_actions": ["action title 1", "action title 2"],
        "status": "on_track|at_risk|behind",
        "note": "Brief observation about this deliverable's status"
      }}
    ],
    "unmapped_actions": [
      {{
        "action_title": "Action that doesn't map to any SOW item",
        "concern": "Why this might be scope creep or a missing SOW line item"
      }}
    ],
    "coverage_gaps": ["SOW deliverable with no matching actions in the pipeline"]
  }},
  "risk_radar": [
    {{
      "risk": "Specific risk description",
      "severity": "high|medium|low",
      "evidence": "What actions or patterns indicate this risk",
      "mitigation": "Suggested next step"
    }}
  ],
  "process_compliance": {{
    "observations": [
      {{
        "area": "Process area (e.g., Triage, Escalation, Assignment, Response Time)",
        "finding": "What the data shows vs what the process docs prescribe",
        "recommendation": "What to adjust"
      }}
    ],
    "overall_health": "green|yellow|red"
  }},
  "trend_intelligence": {{
    "patterns": [
      {{
        "pattern": "Description of the trend",
        "context": "How it relates to program documents",
        "impact": "What this means for the team"
      }}
    ],
    "workload_forecast": "Forward-looking statement about expected workload based on program documents and current action patterns"
  }},
  "recommendations": [
    "Specific, actionable recommendation grounded in program documents",
    "Another recommendation"
  ]
}}

If no Program Knowledge Base is loaded, provide general analysis based on action content and email patterns.

JSON response:

▸ Available Placeholders

{action_details} — Numbered list of all actions with title, priority, category, due date, and assignment status
{total_actions} — Total number of actions in the pipeline
{unassigned_count} — Number of unassigned actions
{high_priority_count}, {medium_priority_count}, {low_priority_count} — Priority breakdown of unassigned actions
{assigned_in_progress} — Number of actions currently assigned
{completed_count} — Number of completed actions
{overdue_count} — Number of overdue actions
{due_this_week_count} — Number of actions due within 7 days
{days} — The lookback period selected (e.g. 14)
{email_summaries} — List of recent email subjects and senders

Note: Use {{ and }} for literal braces in the prompt (e.g. JSON examples). The Program Knowledge Base is injected into the system message automatically — you don't need a placeholder for it.

Changes apply on next Insights refresh.

Documents

6.9 KB

Total Size

7,081

Characters Extracted

Upload Document PDF, DOCX, or TXT · max 10 MB

Filename	Type	Size	Extracted	Description	Uploaded
pilot-kb-team-roster.txt	TXT	1.9 KB	1,918 chars	—	Feb 18, 2026
pilot-kb-project-beta-brief.txt	TXT	1.6 KB	1,663 chars	—	Feb 18, 2026
pilot-kb-program-charter.txt	TXT	1.8 KB	1,806 chars	—	Feb 18, 2026
pilot-kb-coding-standards.txt	TXT	1.7 KB	1,694 chars	—	Feb 18, 2026

▸ How the Knowledge Base works

Uploaded documents are extracted to text and injected into the AI system prompt.
This gives GPT-4 program context (SOW terms, deliverables, processes) to produce more relevant action extraction and insights.
Supported formats: PDF, DOCX, TXT. Only successfully extracted documents are used for AI context.