Support software with an AI you'd actually let answer the customer.
Hermes is a customer support platform built around an AI teammate that knows your knowledge base, drafts replies your agents trust, and only sends what you've approved. Live chat, ticketing, and admin — one platform, one shared brain.
Launching soon. Early access opens to waitlist members first.
If you've run a support team, you know this list.
Your knowledge base is rotting
Articles get written once and never updated. The team keeps answering the same questions because nobody has time to maintain the docs.
AI tools you can't trust to send
You've seen what bad auto-replies do to a customer relationship. Most AI support tools give you a binary choice: turn it on and pray, or leave it off.
Chat and tickets in different worlds
Your live chat tool doesn't know what your ticketing tool knows. Agents jump between tabs. Context gets lost. Customers repeat themselves.
Three things Hermes does that other support software doesn't.
AI Auto-Reply you can actually trust
Most AI support tools either send everything or send nothing. Hermes gives you four modes — Disabled, Suggest Only, Draft Only, and Auto-Send — so you decide exactly how much rope the AI gets. Every draft is grounded in your knowledge base, scored for confidence, and gated by policies you set. If the AI isn't sure, it doesn't send. Period.
- Retrieve relevant KB articles
- Score grounding confidence
- Draft reply with citations
- Validate against policy gates
- Send, draft, or escalate
A knowledge base that maintains itself
Hermes watches the tickets your team resolves, clusters similar conversations, and proposes new or updated knowledge base articles for your review. Your KB stops rotting because your AI is the one tending it — and every accepted article makes the next auto-reply more accurate.
- Cluster resolved tickets
- Detect knowledge gaps
- Propose CREATE or UPDATE
- Human approves or archives
- Embeddings refresh automatically
One platform, one brain
Live chat, ticketing, and admin run on the same database with the same AI. When a chat becomes a ticket, nothing is lost. When the AI learns something from one channel, it knows it on the other. Your agents stop tab-switching. Your customers stop repeating themselves.
- Desk live chat for agents
- Resolve ticketing + email
- Command admin + analytics
How Hermes decides whether to reply.
We've watched AI tools embarrass support teams by sending confident-sounding nonsense. Hermes is built to fail safely. Here's exactly what happens when an email comes in.
Searches your knowledge base using keyword overlap and vector similarity. Only relevant articles move forward.
A confidence score measures how well the retrieved knowledge covers the question. Low score, no draft.
The AI writes a reply citing specific articles. Citations are stored with the draft so you can audit any reply later.
The draft passes through policy gates: hallucination checks, tone checks, your own custom rules.
Based on your mode, the reply is sent, saved as a draft for an agent, or held for review.
Disabled
The AI doesn't draft replies at all. Use this while you're getting comfortable, or for sensitive ticket types.
Suggest Only
The AI shows your agents a suggested reply they can accept, edit, or ignore. The customer never sees anything until an agent sends.
Draft Only
The AI writes a complete draft and saves it on the ticket. An agent reviews and sends with one click. Great for high-volume, low-risk queues.
Auto-Send Allowed
For tickets where the AI's confidence clears your threshold and your policies allow it, Hermes sends the reply directly. Everything else falls back to Draft Only.
You can change modes per department, per channel, or globally. You stay in control.
Everything a modern support team needs.
Hermes is a complete platform, not just an AI layer. Here's what's in the box.
Live chat
Real-time agent and customer chat with AI assist on every conversation.
Email ticketing
Full lifecycle from inbound email to resolved, with reply threading and attachments.
AI ticket triage
Every incoming ticket gets a summary, sentiment, intent, and suggested reply on arrival.
Knowledge base
Articles with vector embeddings, CSV import, and AI-assisted maintenance.
Automation engine
If-this-then-that rules with conditions and six action types. No code required.
SLA tracking
First response, next response, and resolution time policies with a real-time evaluator.
CSAT surveys
Tokenized survey emails sent automatically after ticket close. No customer login required.
Reporting & analytics
Date-range reports, trend sparklines, per-agent metrics, and a backlog heat map.
Web forms
Embeddable customer-facing forms with topic-based routing to the right department.
Multi-tenant security
Strict organization isolation, JWT auth, CSRF protection, and full security headers.
Macros & canned replies
Org-wide and personal templates for the responses your team sends every day.
Spam & deduplication
Inbound email signature verification, allow/block/quarantine policies, and dedup.
The details that matter when you're choosing a platform you'll live in every day.
Hermes is built by people who've used a lot of bad support software. We sweat the details that show up at scale.
- Real-time ticket updates over WebSockets, not polling
- Inbound email signature verification (ED25519) on every webhook
- Per-organization OpenAI rate limiting with circuit breaker protection
- Presigned upload pipeline for attachments, S3 or Cloudflare R2
- Idempotent email send with full delivery lifecycle tracking
- Strict tenant isolation enforced at the data layer
Be first in line when Hermes opens to early access.
We're opening early access soon to a small group of teams who want to help shape what Hermes becomes. Join the waitlist and we'll be in touch when there's a spot for you.
You're on the list. We'll email you when early access opens. If you checked the demo box, we'll be in touch sooner.
Why I'm building Hermes.
I spent years on support teams, first as an agent and then as a manager. I watched our knowledge base go stale because nobody had time to update it. I watched agents copy-paste the same answer fifty times a week. I watched leadership get excited about an AI tool, turn it on, and turn it off two weeks later because it was sending replies nobody had approved.
The problem with most AI support tools is that they're built by people who've never taken a ticket. They look impressive in a demo. They fall apart on a Monday at 4pm when the queue is on fire and the AI just sent something embarrassing to your biggest customer.
Hermes is what I wish I'd had. An AI that actually knows our knowledge base. Confidence scores I can tune. Modes I can change per department. A platform that treats my agents as the experts and gives them better tools — instead of trying to replace them.
If you've run a support team, I'd love to talk. Join the waitlist, check the demo box, and tell me what's broken about your current setup. I read every reply.
Questions you probably have.
- We're opening early access soon — likely early in the year. Waitlist members get in first, in the order they joined.
- No. Your customer conversations and knowledge base are yours. We use OpenAI's API with data retention disabled — your data is not used to train any model, ours or theirs.
- Yes. You can paste articles directly, upload a CSV, or push articles into Hermes from an external system using our ingestion API.
- We're still finalizing pricing. Waitlist members will get early access pricing locked in for their first year.
- At launch, email and live chat (with embeddable web forms). We're working on Slack, WhatsApp, and SMS next — the architecture is built to make adding channels straightforward.
- Hermes uses PostgreSQL with strict per-organization isolation. Attachments are stored on AWS S3 or Cloudflare R2. We'll publish a full security and data-handling page before launch.
- Hermes is built by a former support manager who got tired of bad support software. More on that above.