AI Prompting
Field Managers
A diary study + interview series exploring how AT&T field managers would use AI-powered prompts inside Titan to prioritize visits, coach technicians, and close performance gaps.
01 // Objectives
Research Objectives
Field managers currently lack targeted, actionable AI prompts within the platform to help prioritize technician visits and address performance gaps. Existing tools don't reflect the nuanced criteria managers use, nor offer the transparency needed for effective performance management — resulting in missed interventions, inconsistent coaching, and reduced efficiency.
Identify Preferred AI Prompts & Recommendations
Explore which types of AI-generated prompts, suggestions, and notifications managers find most valuable when making visit and coaching decisions.
Understand Criteria for Recommended Visits
Determine what information and metrics managers currently use to prioritize technician visits, and how these criteria should be reflected or enhanced in AI recommendations.
Evaluate Integration into Existing Workflows
Discover how AI prompting can best support managers' daily routines, decision-making, and interactions with technicians, ensuring relevance and ease of use.
Uncover Additional Data or Features Needed
Identify any other information, metrics, or platform features Staff managers believe should be incorporated into AI recommendations for maximum effectiveness.
Explore Impact on Performance Management
Assess how AI prompting influences managers' ability to resolve performance gaps, conduct meaningful coaching, and drive positive behavioral change.
02 // Method
Participants & Methodology
Participants
9 Field managers (4 Ambassadors)
Methodology
- › Diary Study + 1:1 interview
- › 5 interviews conducted
- › Conducted via Microsoft Teams
- › Recruited from Manager & Ambassador lists
03 // Phase One
Before a Visit
Morning routines consistently center on making sure every tech has work and that the plan is geographically efficient before heading to the field.
Field Manager's First Hour
- ›Techs with recent quality or safety deviations get moved to the front of the line.
- ›First hour is operational triage: email, timesheets, load review, route planning, team huddles.
Tools & Dashboard Checkups
- ›Titan is the backbone for planning and assignment.
- ›INFOR and ORCA for timesheets.
- ›Power BI, Team Habits, SQUID, MSOC/ESM, LSBBT, VOC for trends, repeats, and coaching opportunities.
Factors That Trigger a Visit
- ›Safety day, current job/work type (UG, splicing).
- ›Repeats / PSAs, incomplete installs, efficiency concerns.
Needs Before Visiting a Tech
- ›Prior deviations, training/PLE status.
- ›PPE/tools readiness.
- ›Job type and address (to arrive while work is in progress).
Deviations or Concerns Before Visit
- ›Safety gear and scene setup (cones, vest, wheel chocks).
- ›Vehicle/cab standards, Expert Path markers, drop/quality issues.
Trends to Flag Before a Visit
- ›Repeats (especially 0–3 day and chronics), PSAs.
- ›High geo score anomalies, weather.
Before a Visit
- "Summarize today's load and highlight techs without assignments."
- "Show overdue safety observations and repeats flagged in SQUID."
- "Combine [xxyz] data to suggest the most efficient route for my visits."
- "Generate a priority visit list based on Safety Day + PSAs + incomplete installs."
- "Generate a pre-visit summary for Technician X: job type, address, prior deviations, PPE readiness."
- "Show training/PLE status and any overdue competencies for tech Y."
- "Highlight past safety deviations and confirm if corrective actions were taken."
- "Provide a quick coaching script based on last two visits."
04 // Phase Two
During a Visit
Field visits are safety-driven, documentation-heavy, and app-fragmented. Titan can deliver transformative value by reducing documentation friction, consolidating tools, and enabling data-driven tech selection.
Safety as the First Priority
- ›Arrival actions: cones, parking, PPE readiness, greeting the technician.
- ›On-site focus: PPE compliance, ladder safety, ergonomic practices.
- ›Recommendation: Titan should make safety checklists and quick-reference guides readily available.
Documentation Bottlenecks
- ›Slow photo uploads, crashes, and lack of offline capability hinder timely documentation.
- ›Recommendation: Offline-first design and streamlined media handling are critical.
Multi-App Workflow
- ›Tools like MSOC, LSBBT, and Apple Maps are frequently referenced alongside Titan.
- ›Recommendation: Integrating or linking these tools within Titan reduces context switching.
Desired Titan Enhancements
- ›Quick summaries showing past deviations, safety/quality scores, habit progress.
- ›AI rewrite is valued but needs edit-after-the-fact and offline functionality.
- ›Tech prioritization should consider deviations, efficiency, and proximity.
During a Visit
- "Generate an on-site safety checklist for Technician X."
- "Highlight past PPE or truck cleanliness deviations for tech Y."
- "Confirm corrective actions on past safety deviations during this visit."
- "Provide quick coaching tips for reinforcing Expert Path compliance."
- "Show common errors for fiber installs and what to look for today."
- "Provide quick coaching tips for reinforcing ladder safety and ergonomics."
05 // Phase Three
After a Visit
Post-visit work is dominated by driving and administrative wrap-up. Titan can materially reduce friction and increase field coverage through routing, multi-app consolidation, and offline-capable documentation paired with timely reminders.
End-of-Day Workflow
- ›Driving back to the work center, handling calls/emails/overrides, finishing documentation.
- ›Many defer full documentation until end-of-day to maximize field time.
Documentation Friction
- ›Titan is central but photo uploads are slow, pages reload, mobile performance lags.
- ›Participants ask for offline mode, faster saves, and compact forms.
Risk Reduction Needs
- ›Accurate, timely info and reminders for monthly tasks (competencies, random driving observations).
- ›Customizable dashboards and pop-ups help avoid misses.
Trust & Sharing Boundaries
- ›Trust grows with correct data and helpful suggestions — users 'trust until given reason not to'.
- ›Sensitive personal information stays confidential; only pertinent operational info is shared with techs.
After a Visit
- "Summarize all pending documentation tasks for today and prioritize by urgency."
- "Generate a quick end-of-day checklist: overrides, calls, photo uploads."
- "Show which visits still need submissions and attach direct links."
- "Optimize my route back to the work center while flagging any missed observations."
- "Show jobs with incomplete documentation and provide one-click completion links."
- "Provide shortcuts to frequently used tools (Titan, Power BI, LSBBT) in one click."
06 // Designs
Performance: AI Prompts & Recommendations
Managers prioritize behavioral and quality-driven metrics (HPC, scoping, true testing) over raw efficiency, and want integrated dashboards that combine these with historical trends for real-time coaching.
AIQ — Quality
- ›Repeat analysis broken down by cause (cut drops, bad ends, RG issues) with category percentages.
- ›Technician drill-down: who is struggling over 1–3 months and why.
- ›Root cause categorization: customer vs tech vs drop-related.
- ›Quick metrics: 0–3 day repeats, missing huddle time, missed OTO checks.
- ›Identify techs responsible for bringing down team numbers.
AIQ — Quality
- "Summarize repeat jobs by cause category for the last 30 days."
- "Show top 5 technicians with highest repeat count and reasons."
- "Break down AIQ attainment by technician and flag those below threshold."
- "Identify technicians with most ATND issues and missing huddle time."
- "Generate a trend report for tech performance over the last 3 months."
Efficiency
- ›Identify the bottom 10–20%: how long they stay on jobs, how often they overrun.
- ›Job-type breakdown — which types cause inefficiency (copper, fiber, APA).
- ›Root cause analysis tied to specific job types.
- ›Behavioral insights and links to PLE / Expert Path resources.
Efficiency
- "Show bottom 10% of technicians by efficiency and their most frequent job types."
- "List technicians with highest job overruns and link to job categories."
- "Summarize efficiency trends for fiber vs copper vs APA jobs."
- "Provide quick coaching tips for improving efficiency on fiber installs."
- "Generate a link to relevant PLE course for low-performing techs."
ATND
- ›Top and bottom performers identified at a glance.
- ›Timing analysis — start-of-day vs end-of-day patterns.
- ›Daily breakdown with anomaly highlights (e.g., days > 2hrs ATND).
- ›Visual representation: graphs or heatmaps of spikes by day.
- ›Root-cause drill-down for high-ATND days.
ATND
- "Show top 5 technicians with highest ATND this month and their trend."
- "Break down ATND by time of day (morning vs evening) for each tech."
- "Highlight days with ATND > 2 hours for any technician."
- "Generate a graph of ATND spikes by day for the bottom 10% performers."
- "Summarize ATND anomalies for quick coaching conversations."
07 // Field Review
Screen-by-Screen Findings
Baseline
Managers like the screen concept but want it streamlined and context-aware. Show only competencies due soon, surface coaching opportunities as clickable drill-downs, and keep safety trends front-and-center.
Deviations
Surface both current and past deviations with safety emphasis. Include quick blurbs from previous visits so coaching can reinforce positive change ('Great job correcting this since last time').
Strengths
Strengths shift tone from corrective to constructive. Surface them in Titan for quick reference — safety = no deviations, quality = no repeats, team habits = consistent pre-calls.
Badges
Show only when earned, paired with top 3 strengths/opportunities. Swipeable cards for multiple badges, with links to send praise (e.g., via Nikita).
AI Chat
A conversation enabler — surface timesheet/ATND anomalies, performance trends, behavioral follow-ups, and job-level narratives in real time. Managers trust AI and the info it provides.
Documentation
Mixed opinions on system-generated vs static checklists, but all agreed past trends and deviations should drive recommendations for safety and quality topics.
Visit Summary
Mixed views on when to trigger coaching actions vs growth plans. Managers want automated reminders and progress-tracking notifications after a plan is created.
Takeaway
Managers want AI that automates the diagnostic work — turning raw narratives, codes, and trends into clear, drillable insights they can take straight into a coaching conversation.
© 2025 AT&T Intellectual Property · Proprietary · Internal Use Only