Alec Harmon

• Run omnichannel support and success as a solo department head for a platform of about 15,000 monthly active users; selected as the CX leader to transition with the product through its spin-off and acquisition.
• Own the full customer relationship for a seat-based, month-to-month platform, from demo and sale through onboarding, support, expansion and offboarding.
• Track churn, retention, and seat expansion and contraction by account; negotiate discounts for large seat additions and to retain unhappy customers.
• Measured trial-to-paid conversion and 90-day retention: about a third of trials convert to paid, and over 90% of those that add a card stay through 90 days. The finding moved the focus from post-payment churn to getting trials to add a card.
• Led the support team's migration from Salesforce to Zendesk, and later from Zendesk to Intercom.
• Run AI-assisted ticket triage: a Grok Bot agent reviews the support queue every 10 minutes and leaves an internal note on each ticket with the likely issue, using the support MCP server to spot wrong regional URLs (such as a UK user on the .com site), mismatched login and notification emails, and people in more than one organization. It removes spam automatically only at 95%+ confidence.
• Own Intercom Fin as the AI support agent, tuned to an 89% resolution rate and 88% deflection rate (January to September), which lets one person cover the user base.
• Rank top contact drivers from conversation data and target the highest-volume reasons first to raise AI resolution and reduce human workload.
• Review Fin's answers, especially low-CSAT conversations, plus random spot checks; fix issues by updating the knowledge base or writing custom rules, and update Fin's instructions at least monthly. Use Intercom Operator to improve weak conversations.
• Set up Intercom Copilot and train it alongside Fin, and test changes in Intercom's AI simulation tool while tuning (the same approach used with Zendesk AI before Fin).
• Own the internal and external knowledge bases and update articles with every release so Fin answers from current information.
• Designed Fin's escalation rules: hand off when troubleshooting is exhausted and for all billing and cancellation topics, with easy access to a human and no stonewalling. Also wrote Fin's tone, handoff and escalation rules and its Procedures, including for HIPAA-sensitive conversations. Live on chat.
• Standing voice of the customer in the weekly Product meeting (Level 10): use conversation data to help shape product decisions across the full customer journey. Example: through an email-delivery tool built for Fin, found that transactional emails (password resets, invitations, billing) wrongly offered an opt-out; took it to engineering, who fixed it.
• Set customer success platform priorities every quarter in EOS Rocks alongside the COO, CEO and head of engineering.
• Designed and shipped the AI systems listed under Selected Technical Builds: a customer success platform, onboarding and retention automation, an MCP server, and Fin tool middleware.
• Strategic data liaison to the C-suite: write SQL against the production read replica, and run billing operations, refund logic and revenue queries in Stripe Sigma and Maxio.
• Use Pendo and PostHog to advocate for UX fixes, and run weekly bug-triage standups with engineering (Jira, Linear, browser DevTools and network logs).
• Train customers on the full product in success meetings, including Atlas AI, the company's AI chief-of-staff product, and report bugs found in production.

• Directed global omnichannel support across three business lines: General Support, Implementer Support and Tech Support (EOS One).
• Built the support and success processes for EOS One from its first months.
• Deployed AI-driven self-service tools and workflows that cut first response time by 55%.
• Partnered with engineering on streamlined resolution paths that reduced resolution times by up to 85%.
• Built employee onboarding with custom questions and scoring.
• Developed a team lead who later became the support department manager after EOS One was spun off.

• Built and directed global support operations for 30,000 accounts, scaling offshore BPO partnerships and internal teams in India and the Philippines.
• Led 40+ technical support professionals at one time: about 15 direct reports, plus 3 BPO team leads who each managed a team of about 10 offshore agents.
• Served as a Zendesk administrator: helped build and review triggers, macros, routing, views, automation and reporting, and built Zendesk Explore dashboards on the metrics a support leadership team tracks.
• Planned coverage in a call-center environment using Erlang C calculations; ran agent quality reviews alongside the QA analyst.
• Taught new-hire training topics as a new-hire team lead. Coached team members into team lead roles (at least four Weave team leads started on the team), continuing after they left the team.
• Launched Weave's first dedicated email response team (3 agents), cutting open tickets 68% in 30 days and ending the overtime "nights of tickets." Rolled the model into Tier 1 through team-lead-run round-robin assignment.
• Redesigned bug-triage workflows by moving customer communication on escalated bugs from Tier 1 to Tier 2, freeing Tier 1 for solvable issues.
• Served as the support champion for the 10DLC (texting compliance) rollout, working with product, engineering, success and onboarding, and joining vendor meetings with Bandwidth.

• Directed omnichannel fulfillment across multiple locations, managing 35+ people, planning staffing and coverage, and building division-wide leadership training programs.

Customer success platform (EOS One)

Internal customer journey and health platform on Vercel and Neon Postgres. Seventeen scheduled syncs join Stripe, PostHog, Intercom, Metabase (production app data) and feedback forms into one view per organization: MRR, seats, usage, NPS, CSAT and health. Multi-user, with per-person logins and permissions. 844 automated checks guard the program rules, permissions and data masking.

90-day customer success program

Trials enroll hourly. Each email is chosen from what that customer has and hasn't done yet, not from fixed dates, and Claude drafts it from the account's real activity. Every draft is human-reviewed before sending. Automatic sending is built, with per-org controls, and held back until the drafts are tuned. Live since September 2026.

Program analysis and measurement

Pulled the data from Metabase, defined what each field meant, and worked with Claude to compile and analyze it. Found that activation was split (some trials finished all four early milestones, many finished none) and that orgs holding a team meeting in the first 30 days converted at nearly twice the rate. Built a Program Results report that compares every email against a baseline replayed over the 13 weeks of trials before launch, and a weekly leadership Scorecard pushed to PostHog.

Retention monitoring

Health scoring weights survey scores by the person's role and how recent the score is. Flags declining usage, pending cancellations and NPS detractors. Claude drafts outreach from usage, NPS and recent Intercom conversations. Includes a phone-friendly queue for the CEO's outreach to cancelling accounts.

MCP servers for the platform

Two remote MCP servers, a success side and a support side, with OAuth 2.1 sign-in tied to company logins, permission rules, write actions behind three separate checks, customer emails masked in every answer, and every tool call audited. The support side also queries Metabase (a read replica of the product database), Clerk (login) and SendGrid. Works with Claude, ChatGPT and Grok. Available to all 8 employees, with adoption just starting.

Intercom Fin tool middleware

Gives Fin scoped, audited access to SendGrid to check email delivery status and clear bounces and blocks. Resolves after-hours tickets in minutes for a solo team with worldwide customers. Least-privilege API key and a full action log.

Customer feedback routing

Sends Intercom CSAT, PostHog NPS and form feedback to Roam, Google Sheets and the platform, and shows a teammate's call status in Roam.

Survey outreach automation

Gemini and Google Apps Script drafted personalized follow-ups to survey responses into a review queue. It was the foundation for the platform's outreach.

Tier Drop (prototype)

Multi-tenant AI knowledge-base platform on Next.js, Supabase and Gemini. Experts record a process on video, Gemini generates the article, and support agents use it through a Chrome side panel and Slack. Validated with prospects but not taken to market.

B.S. Business Management

Utah Valley University · Orem, UT

2020