Autonomous lifecycle marketing
Your next campaign won't need a marketer.
Hermes finds high-intent cohorts, writes tailored outbound, and moves every campaign through review and delivery without a manual ops loop.
See cohort discovery, draft generation, and approval flow live.
# proactive-agent › hypothesis
SELECT user_id, last_order_date
FROM orders
WHERE last_order_date < NOW() - INTERVAL '90 days'
AND total_lifetime_value > 500
✓ 312 users matched — submitting for review
email-agent › draft
To: maya@northstarcommerce.com
Subject: A smarter reorder flow for your spring cohort
Hi Maya,
Customers who reordered hydration bundles in February are now returning through organic search. We drafted a win-back sequence around that behavior.
Your VIP segment drove 847 repeat orders last quarter, so the message leans on replenishment timing instead of discounts.
queue › pipeline status
proactive-agent
campaign-dispatch
hyperpersonalize
email-delivery
Finds the cohorts you didn't know to look for.
The agent scans your customer database schema, formulates SQL hypotheses, and surfaces high-value segments before you even think to ask. Runs daily. Each trigger goes into a review queue before anything ships.
[08:02:14] Scanning schema...
[08:02:15] Tables found: orders, users, subscriptions, events
[08:02:16] Hypothesis: high-LTV users dormant >90 days
[08:02:17] SELECT user_id, last_order_date
FROM orders
WHERE last_order_date < NOW() - INTERVAL '90 days'
AND lifetime_value > 500
✓ 312 users matched → submitting for review
→ Trigger queued. Awaiting human approval.
One email per person. Not per segment.
The email agent reads each user's full history — orders, behavior, lifecycle stage — and writes a message that feels like it came from someone who actually knows them. Not a template with a first name swapped in.
To: sarah@acmecorp.com
We noticed you've been away
Hi Sarah, it's been 94 days since your last order of the Pro Plan toolkit. Given your team processed 847 reports last quarter, here's what's new that fits your exact workflow.
To: james@startup.io
Ready to upgrade? Here's why now.
Hi James, you've hit the free tier limit 3 times this month. Based on your usage — mainly batch exports and API calls — the Growth plan would save your team ~6 hours a week.
To: priya@enterprise.com
Your renewal is coming up — let's talk
Hi Priya, your annual contract renews in 28 days. Your team's adoption jumped 40% since onboarding. We'd love to expand the seat count before renewal at your current rate.
proactive-agent
Scans schema, generates hypotheses
campaign-dispatch
Fans out one job per matched user
hyperpersonalize
Drafts individual email per user
email-delivery
Sends via Resend, handles retries
Trigger → dispatch → personalize → deliver.
A BullMQ-powered pipeline fans out one job per matched user, runs personalization in parallel, and handles retries, cooldowns, and failures automatically. Review mode or fully autonomous — your choice.
It learns your brand. Over time.
The agent builds persistent org-level memory across every campaign — brand voice guidelines, what drove opens, anomalies in cohort behavior. Each new run is informed by every previous one.
Tone guideline
Avoid urgency language in subject lines. This org's audience responds better to curiosity-led hooks. Open rate lifts 12% when we omit 'limited time'.
Cohort insight
Users who churned after month 3 share a pattern: no API usage in first 14 days. Re-engagement emails citing specific features they never used outperform generic 'we miss you' by 3.1×.
Delivery anomaly detected
Campaign #14 saw 22% lower open rate in enterprise tier. Likely cause: subject lines over 60 chars get clipped on Outlook mobile. Adjusted max subject length to 52 chars.
Up and running in minutes.
Four steps from signup to your first autonomous campaign.
Sign up and create your org
Instant setup, no forms, no sales call. Your org workspace is ready in seconds.
Connect your customer database
Paste a read-only connection string. Encrypted with AES-256 immediately — we never store raw credentials.
Review your first AI-discovered cohort
The agent scans your schema and surfaces its first trigger within minutes. You approve or reject before anything moves.
Launch your first campaign
Approve the drafts, hit send, watch the pipeline run in real time. The agent remembers what worked.
Built for teams that care about trust.
Security and transparency aren't features — they're defaults.
Your data stays yours
Queries run on your infrastructure. Raw customer data never hits our servers — only aggregate results and schema metadata.
AES-256 encrypted credentials
Connection URLs are encrypted at rest with AES-256-GCM before we store them. We cannot read your connection strings.
Human-in-the-loop by default
Every discovered cohort and drafted email passes through your review queue before sending. Full autonomous mode is opt-in.
Full audit trail
Every job run, agent decision, and send is logged. You always know what happened, when, and why — query the full history anytime.
Frequently asked questions
