Selected Work
Case Studies
Production systems shipped for clinics, agencies, and B2B service teams — the architecture, the rollout, and the numbers.
100%
Call Visibility
0
Manual Reviews
Eliminating Operational Blind Spots
Front-desk performance is one of the biggest levers on clinic revenue, but manually auditing thousands of inbound calls is slow, expensive, and inconsistent between reviewers. The client needed a central system capable of analysing patient interactions without human involvement — turning raw call data into measurable growth levers across all locations.
Multi-Tenant Call Intelligence Infrastructure
Built a large-scale multi-tenant call-intelligence platform serving 20+ dental practices. The system ingests every inbound call directly from GoHighLevel into a real-time dashboard powered by 16+ data tables and 22+ API modules. Call transcription and scoring are handled end-to-end by AI using Google Gemini, with no manual step between call receipt and scored output.
Precision Scoring & Automated Reporting
Every patient interaction is automatically evaluated against an 8-step scoring framework, surfacing critical metrics including booking rate, call volume, missed vs. answered ratios, and human vs. AI handling comparisons. The result: 100% call monitoring with zero manual reviews. Daily performance reports are delivered automatically across three integrated channels — SMS, email, and an internal dashboard.
Answered vs Missed
Booking Rate
| Period | Calls | Booked | Rate |
|---|---|---|---|
| Week 1 | 412 | 168 | 41% |
| Week 2 | 389 | 171 | 44% |
| Week 3 | 455 | 210 | 46% |
Gemini 8-Step Scoring
4+ Hrs → 5 Mins
Report Generation Time
~16 Hrs
Saved Per Month
A Reporting Bottleneck That Compounded Every Month
Every end-of-month, the ops team spent four-plus hours manually pulling GoHighLevel data, reconciling it across spreadsheets, and formatting client-facing performance reports. With client count growing, this was a bottleneck that scaled with headcount — not with automation. Something had to change.
A Fully Automated ETL Pipeline
Built a multi-stage n8n workflow that triggers automatically on the last business day of each month. It pulls all relevant performance data from GoHighLevel via API, normalises and aggregates it across clients, then populates a templated report format — no spreadsheet touching required. Error handling and retry logic ensures the pipeline never silently fails.
16 Hours Returned to the Team Every Month
Report generation now completes in under 5 minutes from trigger to delivery. The team no longer allocates a half-day to end-of-month reporting. That time has been redirected to client strategy work. The pipeline has run without manual intervention for every reporting cycle since deployment.
40%+
Tickets Auto-Resolved
0
Human Touch Required
Support Volume Outpacing Headcount
The support team was drowning in repetitive, low-complexity tickets — password resets, status checks, standard FAQ responses — that consumed the same time as genuinely complex issues. Ticket backlog was growing and first-response times were slipping, despite the team working flat out.
Rule-Based Classification with Smart Escalation
Built a webhook-driven intake pipeline that intercepts every incoming ticket before it reaches the human queue. A layered rule engine classifies tickets by type, urgency, and account tier. Standard-pattern tickets are resolved automatically with templated responses. Ambiguous or high-priority tickets are escalated to ClickUp with full context — channel, account history, and a suggested resolution path — so the agent can act immediately.
40% of Volume Handled Without a Human
Within the first full month, over 40% of all incoming tickets were resolved end-to-end by the automation, with zero escalation needed. Human agents now only see tickets that genuinely require judgment. Average first-response time on escalated tickets dropped because agents are no longer context-switching from low-complexity work.
New
In Progress
Resolved
90%
Time Reduction
< 2 Min
Time-to-Draft
Sales Reps Buried in Email Triage
Inbound leads were arriving across multiple channels — contact forms, cold replies, referral emails — and landing in a shared inbox that no one owned cleanly. SDRs were spending the first hour of every day sorting, categorising, and drafting initial responses. High-intent leads sat unread while low-priority emails got actioned first simply because they arrived sooner.
GPT-Powered Routing Straight Into Slack
Built an n8n workflow that monitors the inbound inbox in real time. Each email is passed to GPT-4 for intent classification — demo request, pricing inquiry, partnership, or noise. High-intent leads are immediately routed to the correct Slack channel with a structured summary and an AI-generated reply draft the SDR can send in one click. Low-priority emails are logged and batched.
90% Faster Response, Zero Inbox Archaeology
Time-to-first-draft dropped from 45+ minutes to under 2 minutes. SDRs no longer open the inbox to triage — they open Slack and act. High-intent leads are now contacted faster than competitors who are still doing this manually. The system handles volume spikes without adding headcount.
0
Missed Calls
24/7
Coverage
After-Hours Revenue Walking Out the Door
Across 5 clinic locations, a significant portion of incoming calls arrived outside front-desk hours — lunch breaks, evenings, and weekends. These calls went to voicemail, and a large fraction of patients did not leave a message or call back. Each missed call was a missed booking. Staff were also spending disproportionate time on routine scheduling calls during peak hours, reducing availability for in-clinic patients.
Aria: A Voice Agent with Full CRM Access
Deployed an AI voice agent — named Aria — across all 5 clinic phone lines. Aria handles natural language conversations end-to-end: she looks up patient records in GoHighLevel, checks practitioner availability in real time, books new appointments, and processes reschedule and cancellation requests. The agent understands clinic-specific terminology, appointment types, and patient preferences. Escalation to a human is available on request.
Zero Missed Calls Across All Locations
Since deployment, every call across all 5 locations is answered — regardless of time. Aria independently handles the majority of routine scheduling calls, freeing front-desk staff to focus on in-clinic patient experience. The clinics have recovered bookings that would previously have been lost to after-hours voicemail, and staff report a measurable reduction in phone-related context switching during busy periods.
Hours → Minutes
Onboarding Time
100%
Consistent Delivery
Manual Onboarding Was a Hidden Tax on Growth
Every new client triggered the same sequence: create a Drive folder structure, copy template documents, populate them with client details, set up the CRM record, and send access links. Done manually, this took 2–4 hours per client and was error-prone — wrong names in documents, missing folders, CRM fields left blank. As the agency scaled, onboarding became a recurring ops tax that delayed the start of actual client work.
A Single Trigger, Fully Automated Workspace
Built an n8n automation triggered the moment a new client contract is signed. The workflow provisions a structured Google Drive folder hierarchy using the Drive API, copies and populates all standard document templates with client-specific data, creates the CRM record with all intake fields pre-filled, and sends the client a branded welcome email with access links. All steps complete in sequence with validation checks between each stage.
Consistent, Instant Delivery at Any Scale
New clients now receive a fully configured workspace within minutes of contract signature — no ops team involvement required. Document accuracy is 100% because population is automated from a single source of truth. The agency can onboard multiple clients in the same day without the process becoming a bottleneck, and the team can focus on strategy from day one rather than admin.