Comparison
Customer support with Hermes
Helpdesk, ticket triage and knowledge-base answers run with or by the agent.
Three posts touch support, with little evidence of results. Craig Hewitt’s first automated scan of support tickets produced 6 help-doc and 2 marketing suggestions; deciding how to validate and act on them is still open. He says support and dev suit agents because they are tightly coupled, data-rich and fairly deterministic. In a Hacker News thread, a founder who replaced their helpdesk says their company replaced its helpdesk with Hermes, using an escalation chain from per-customer sessions to Hermes to humans, while a Hacker News commenter questions this, arguing that records of prior contact and the surrounding infrastructure are what a helpdesk provides. A lawyer builds a Q&A knowledge base from sent email, with a human tag gate before answers go public. No volumes, resolution rates or customer outcomes appear.
Key points
- Mine support tickets for help-doc and content gaps — 1 post. (Craig Hewitt · Day 4 of agentifying my business ¶ L7)
- Escalate in layers: customer-facing agent, then Hermes, then humans — 1 post. (a founder who replaced their helpdesk · We replaced our helpdesk with Hermes… ¶ L6)
- Keep a human gate between private drafts and public answers — 1 post. (an attorney who runs a local business · My Hermes setup (Notion knowledge base)… ¶ L80)
- A helpdesk is infrastructure and records, not only a text interface — 1 post. (a Hacker News commenter · We replaced our helpdesk with Hermes… ¶ L9)
Where posters disagree
Can an agent replace a helpdesk product? ⚠️
- The company replaced its helpdesk with Hermes, which holds long-term business memory (a founder who replaced their helpdesk · We replaced our helpdesk with Hermes… ¶ L4)
- A helpdesk’s value is the surrounding infrastructure and contact history, which a chat agent alone lacks (a Hacker News commenter · We replaced our helpdesk with Hermes… ¶ L9)
Frameworks
- Hermes-fronted support escalation chain — Hermes gathers the customer’s details and starts a per-customer Pi session on a local model; that agent can escalate to Hermes, which can escalate to the humans. (a founder who replaced their helpdesk · We replaced our helpdesk with Hermes… ¶ L6)
- Customer messages the company
- Hermes gathers relevant customer details
- Pi session on local Gemma 4 talks to the customer
- Pi agent escalates to Hermes
- Hermes escalates to the team
- Private-to-public Q&A pipeline with human tag gate — Hermes backfills and nightly-updates a private Q&A database from sent email; the poster reviews and tags rows, and a separate cron sweeps tagged rows into the public subscriber-facing database that the chatbot reads. (an attorney who runs a local business · My Hermes setup (Notion knowledge base)… ¶ L80)
- Private Notion database: question, answer, category columns
- Hermes backfills from past email threads and checks the sent folder each night
- Hermes leaves category blank; the poster reviews, revises and tags
- A cron sweeps tagged Q&A into the public-facing database
Tactics
- Scan new support tickets for help-doc and blog opportunities — An automated scan flags tickets that could improve or create help docs, or update or create blog posts on the website. (Craig Hewitt · Day 4 of agentifying my business ¶ L7)
- Let the Pi session fetch customer data through a tool call instead of having Hermes hold it. (a founder who replaced their helpdesk · We replaced our helpdesk with Hermes… ¶ L21)
- Update a Q&A database nightly from the sent folder — A nightly check of sent mail for threads where a client asked a substantive question and the poster answered; Hermes leaves the category blank for the poster to tag. (an attorney who runs a local business · My Hermes setup (Notion knowledge base)… ¶ L77)
Stories & examples
- First automated support-ticket scan: The first scan produced 6 help-doc suggestions and 2 marketing suggestions; the open problem is the process for validating and acting on them. Lesson: The author sees the validation and action process as the next challenge. (Craig Hewitt · Day 4 of agentifying my business ¶ L10)
- Hermes proposes a subscription Q&A chatbot business: Noticing clients ask the same questions, Hermes suggested a Notion knowledge base with a chatbot giving first-pass answers (web and SMS) for subscribers, backfilled from four years of email. The poster says Hermes has been building it with little input beyond approving commands in Telegram; the poster expected to have the sites live over a weekend. Lesson: Per the poster, Hermes can propose and build a product from patterns in the owner’s own work. (an attorney who runs a local business · My Hermes setup (Notion knowledge base)… ¶ L51)
Numbers
- Help-doc suggestions from the first ticket scan: 6 suggestions (self-reported) (Craig Hewitt · Day 4 of agentifying my business ¶ L10)
- Marketing suggestions from the first ticket scan: 2 suggestions (self-reported) (Craig Hewitt · Day 4 of agentifying my business ¶ L10)
Caveats & warnings
- The process for validating and acting on the suggestions is not yet decided. (Craig Hewitt · Day 4 of agentifying my business ¶ L10)
- The author’s company replaced its helpdesk with Hermes, which keeps long term memory about the business (self-reported, no volumes or results). (a founder who replaced their helpdesk · We replaced our helpdesk with Hermes… ¶ L4)
- Unanswered in the thread. (unnamed commenter · We replaced our helpdesk with Hermes… ¶ L23)
How much to trust this
- Evidence: 3 posts, 5 independent speakers (repeat posters counted once).