I find where your support operation loses time and money, then fix it with the right mix of process, automation, integrations and AI. Results are measured in the numbers leadership already watches: cost per contact, service levels, customer satisfaction and backlog.
For support and customer experience teams of 20 to 200 people, in house or outsourced.
Volume grows, so you hire. Costs grow with it, service levels still slip, and nobody can say exactly where the time goes.
Customers ask the same questions again and again because the cause, in the product, a policy or billing, never gets fixed.
Agents copy details between the helpdesk, CRM (customer relationship management), billing and spreadsheets on every ticket.
Service level agreement (SLA) breaches show up at the end of the week, in reports someone built by hand.
Self-service, macros, chatbots or AI features were bought, but nobody owns whether they actually take work off the team.
I pick each fix by what it will do to your numbers. Often that includes AI; sometimes a process change or an integration does more.
Fix the causes of repeat contacts, and give customers answers they can find on their own.
Remove copy-and-paste work between the helpdesk, CRM, billing and your other systems.
Put AI on the contacts it handles well: answering common questions, drafting replies, summarising tickets. Tested before it goes live, with clean handoffs to your people.
Live SLA, backlog and cost-per-contact views, so problems show up the same day instead of the end of the week.
Most consultants know either the operation or the technology. I spent years on the operations floor before I became an engineer, so I start from your service levels, staffing and cost, not from a tool.
Every engagement is fixed scope and fixed fee. The diagnostic fee is credited against the sprint if you continue within 30 days.
I analyse 90 days of your tickets, workflows and reports, and show you where time and money go and what to fix first.
I fix the biggest drivers from the diagnostic and measure the result against your baseline.
Products, policies and volumes change. I keep the operation tuned as they do.
Founding prices apply to my first three clients, in exchange for a short case study and a reference. Prices in US dollars.
Operations results show I understand the business you run. Engineering results show I can build what it needs.
See how I work before you hire me: a full sample diagnostic run on 26,872 public support tickets, with the code, the editable rules and a record of every decision. Also an LLMOps reference platform (personal project) and more projects on the portfolio.
Zendesk, Intercom, Freshdesk, Gorgias and Salesforce Service Cloud. If you use something else, ask; the method works on any system that can export ticket history.
Wherever it pays off. AI is often one of the biggest levers, for example answering common questions, drafting replies or summarising tickets for agents. I set it up alongside the process, content and integration fixes it depends on, test it before customers see it, and measure it like any other change.
The goal is to handle growth without headcount growing at the same rate, and to give your people better work. Most teams redeploy time rather than cut roles.
Not for the diagnostic. For integrations, for example checking an order or updating an account from the helpdesk, I need an engineering contact and API access, and I build them myself.
A ticket export and your current reports for the diagnostic, plus read access to your helpdesk. For the sprint, admin access to the helpdesk and help centre, with your approval on changes.
I work from exports or read-only access, remove personal details before analysis wherever possible, sign a non-disclosure agreement (NDA) on request, and delete your data when the engagement ends.
Not before I have seen your data. After the diagnostic, I agree specific targets with you for the sprint and report against your baseline every week.
Book a 20-minute call, or email me your monthly ticket volume and helpdesk. I reply within 48 hours and tell you honestly whether I can help.