Run a support inbox with a scheduled coding agent that replies when confident and forwards the rest
A solo developer describes an hourly support agent on a spare computer that classifies emails, replies to what it can answer from an instruction file, and forwards everything else to the founder.
Evidence: The source shows its work: steps, screenshots, code or data you can inspect.
The business problem
A solo founder supporting an open-source product loses time to repetitive installation and configuration questions but cannot afford a support team.
What was tried
A dedicated always-on Linux machine runs a coding agent from an hourly cron job. A single instruction file holds the agent's role, triage workflow, product description, a table of common issues and answers, escalation rules and reply tone. The agent has its own support address, classifies each unread email (installation, feature, configuration, bug, business or spam), replies when confident, and otherwise forwards it to the founder with a one-sentence reason. Business, press, partnership and legal emails are always forwarded, spam is marked read, and counts are logged.
What was reported (not reported)
No results are reported. The author expects 70 to 80 percent of emails could be automated and says response time should move from hours or days to within the hour, but these are expectations, not measurements.
Limitations
There are no volumes, accuracy or escalation rates, and the underlying model and the full instruction file are not shown. The author calls the project experimental and the free email service is in alpha, which is a dependency risk. Output quality depends on the instruction file, a confident wrong answer is an acknowledged risk, and spam or volume attacks are flagged but not mitigated. The author advises a conservative answer boundary and auditing sent replies early.
What you need
A machine that stays on, a coding agent CLI, a separate support email address with API access, and a written instruction file. Cost is estimated at a few cents per hourly run, with about an afternoon of setup.
Sources
- Dr. Tali Režun (Substack) ↗ Firsthand write-up, published June 19, 2026
Source published: June 19, 2026. Last reviewed here: October 11, 2026. Spot a mistake? Tell us.
Related workflows
Roll out OpenClaw for inbox triage, meeting notes and CRM updates in phases with approvals and isolation
An infrastructure vendor's guide sets out business uses for OpenClaw and a phased rollout: pilot a low-risk workflow, add guardrails and approvals, then scale once value and safety are shown.
Measure true resolution, not just deflection, for an AI support agent with an escalation policy
A small open-source evaluation shows a naive support agent deflecting every ticket while resolving about half, and a policy-gated agent that deflects fewer but resolves what it handles.
Draft help-centre articles automatically from resolved tickets and keep a human in charge of publishing
A merged helpdesk change that, when a ticket is resolved, asks an AI whether other customers are likely to ask the same thing and writes a draft article for managers to review if none exists.
Draft help-centre-cited support replies and propose refunds that a person must approve
An open-source support workbench classifies tickets into three tiers, drafts replies that cite help-centre articles, reads billing data in read-only test mode, and only proposes refunds that need a named approver.