AI Marketing: The Full Feature List
An always-on email and SMS outreach system that imports your leads, personalizes every touch, learns what works from real reply and open data, and quietly retires anything that doesn't. Winning templates get amplified, the hottest prospects move to the top of the queue, and every cross-cutting change goes through your review before it rolls out to your whole list.
Below is the complete list of what the system does, including the universal features (learning, background automation, admin chat) that work the same across other system types.
Every prospect is one record regardless of where they came in (CSV import, form submission, a team member adding them manually) and stays that way. Lifecycle tracking makes sure messaging fits the relationship, so a customer never gets pitched again and a past customer gets a win-back instead of a cold intro.
Bulk CSV Import
One header row (email, phone, fname, lname, feed, source, assignedTo) and the import routes each lead to the right team member, the right feed, and the right source. Dedupes against email and phone so the same contact is never loaded twice.
Lifecycle Stages
Lead, engaged, interested, customer, past_customer. Each stage drives different messaging, so the outreach agent knows to skip current customers and run a different angle for past customers. The relationship stays appropriate at every touch.
Team Ownership And Routing
Each contact has an assignedTo field that sets the Reply-To on outgoing email. Replies land in the right inbox, not a shared black hole. No more "who answers this?"
Full Contact History
Every send, reply, open, bounce, unsub, and note is appended in order on the contact record. Look up anyone by email or phone and see the whole relationship, from first touch to last, in plain language.
The outreach agent sends personalized messages on a cadence you set, using your templates, your signature, your booking link, and your own domains. SMS runs alongside email from the same prospect record, so a contact can get an email today and a follow-up text next week with the conversation intact on both sides.
Personalized Every Time
Full variable substitution on every template: first name, company, paragraph 1, paragraph 2, CTA URL, CTA text, closing, signature, unsub URL, company address. No generic mail-merge gaps, no placeholder tokens leaking into a send.
Multiple Email Domains
Configure one or more SMTP domains, each with its own signature and reply routing. Outreach picks the right sending domain per contact based on who owns them. Sender reputation splits across domains by design.
SMS On Three Providers
Twilio, Ahoi, and Volt all supported out of the box. Pick the one your business already uses, drop in the API key, and SMS lands in the same queue as email. Same personalization variables, same send history.
Send Window Controls
Default 8am to 6pm UTC, configurable per deployment. Outreach skips its own sends outside the window so a message never lands at midnight. Holidays auto-pause on top of that (see the Deliverability section).
Every template is tested against every other template in the same slot, so send volume flows toward what actually works. Cadence and touch-phase variation both drive from real data, not from a scheduled guess.
Balanced A/B Variants
When multiple templates fit the same slot, the variant with the lowest send count gets the next send. Tests stay fair, the data builds fast, and the losers retire themselves once the difference is clear.
Sequence Cadence
Configurable intervals between touches, default 3, 4, 7, 14, 21, 30, 45 days. Tune it to your audience, or leave the defaults alone, the cadence is the same either way, tested across your whole list.
Max-Touches Guardrail
Default 7. After that the contact moves cold instead of burning out the relationship with a ninth or tenth send. Cold contacts can be reactivated later through a re-import or a direct reply.
Touch-Phase Variation
Early touches introduce, mid touches share value (case studies, resources, insights), late touches shift angle or offer the booking link directly. A tired opener never hits a prospect twice, and the last-chance message reads differently from the first.
The outreach agent processes replies first on every cycle, so hot responses get immediate follow-up while the rest of the queue waits. Classification decides what the system handles automatically and what comes to you.
IMAP Inbox Polling
The agent reads the mailboxes you configure, matches each incoming message to an existing prospect by email, and appends the message to their history. New inbound messages from unknown senders create new contacts on the inbound-lead feed automatically.
Auto-Classified Replies
INTERESTED routes to the next step with an optional booking link, NOT INTERESTED gets a polite decline, QUESTION gets an answer from product knowledge, BOUNCE auto-suppresses, UNSUB auto-suppresses, REFERRAL creates a new contact, LEGAL or dispute messages flag for your review without sending a reply.
Referrals Become New Contacts
When a reply mentions forwarding to a colleague or passing you along, the agent pulls the colleague's first name, last name, and email out of the message and adds them to the list. The original referrer gets a thank-you, and the colleague enters at the right feed.
Engagement-Score Priority
Opens, clicks, and replies feed a recency-weighted engagement score on every contact. The hottest prospects move to the top of the follow-up queue automatically, so the next cycle spends its time on the contacts most likely to respond.
Every number is broken out by template, feed, and product. You see what's working, not just an aggregate. The learning loop uses the same numbers you see, so when it proposes a change, the evidence is already on the page.
Reply Rate Per Template And Feed
Count of sends, count of replies, reply rate percent, all broken out by template, feed, and product. The winning templates stand out immediately, and so do the ones that need to be retired.
Open Rate With Tracking Pixel
With the optional tracking pixel deployed to your domain, opens attribute to the specific send that was opened. Unique-open rate per template tells you whether the subject line is pulling its weight, separate from whether the body converts.
Bounce Rate Per Sending Domain
Catches deliverability problems early. A rising bounce rate on one domain is the kind of signal you want to act on before the reputation damage compounds. Counts of sends and bounces are per-domain so you know exactly which one is drifting.
Daily Activity Stats
Rolling 7-day view of sends, unsubs, and bounces. The coordinator uses the same data to spot drops and surges without you checking. Also feeds the best-hours-per-feed analysis so you see when each segment is actually reading mail.
Every interaction (send, reply, open, bounce, unsub, suppression, resolution) is logged as a structured outcome. The learning agent compares outcomes against your goals and the memory bank, then generates improvement ideas grounded in what actually worked. A skeptical step reviews each idea against your rules before anything ships.
Creative Step
Reviews recent outcomes, flags, and memory coverage for patterns worth acting on. Groups similar patterns so one action can clear many (five contacts asking about pricing is one knowledge gap, not five).
Skeptical Step
Verifies each proposed change against your rules and the actual evidence. Ideas without real data behind them never reach you, and ideas that would override a rule you set get rejected before they go anywhere.
Auto-Apply Or Queue For Review
Single-template edits, archiving an underperforming variant, and memory updates auto-apply so wins compound fast. Cross-cutting changes to cadence, brand voice, compliance text, audience strategy, or outreach strategy become suggestions for your approval first.
Suggestion Review Through Chat
Master surfaces pending suggestions with priority, category, effort, and risk. Approve, reject, or mark implemented with one command. Pending items older than 7 days get a nudge so nothing quietly rots, and anything past 30 days expires on its own.
Deliverability features are first-class. Bounces and unsubs are handled automatically, footers meet CAN-SPAM, geographic restrictions hold at send time, and holidays auto-pause the whole system so the domain never sends on a day its intended audience is not working.
Auto-Suppress Bounces And Unsubs
Bounce notifications get parsed and the failed recipient gets suppressed. Reply-based opt-out keywords detect unsubscribes and suppress too. Counts by reason feed the analytics so you know whether a rise in suppression is policy-driven or a content problem.
One-Click Unsubscribe
Every email carries a signed per-contact unsub token. The unsub page confirms and suppresses without requiring an inbox reply. CAN-SPAM-compliant footer with your company name, mailing address, and signature injected on every send.
Geographic Blocklist
Two-letter ISO country codes enforced at send time. Contacts in blocked countries simply do not receive sends, so strict opt-in jurisdictions stay clean. Add or remove countries through config without editing code.
Holiday Auto-Pause
Dates set in config. The system pauses itself on matching days so the domain is not sending on a day its intended audience is off. The pause is automatic and lifts on its own the next morning.
Natural-Language Commands
"What is happening," "show pending suggestions," "pause the system," "look up contact@example.com," "show the pending duplicates" all work as plain requests. The chat agent runs the right tool and returns the answer in plain language.
Per-Contact Summary
Ask about any contact and master pulls the full history, the current lifecycle stage, the engagement score, recent touches, and any outstanding flags. Useful before sending a manual reply or resolving a ticket by hand.
Duplicate Detection And Merge
Maintenance spots contacts linked by related email or related phone and writes them to a duplicates report. Master walks you through merges one pair at a time, preserving history and summing counters, so the combined record keeps every touch from both originals.
No-Code Configuration
Products, brand voice, email signature, unsub URL, booking URL, outreach rules, compliance rules, geo-blocklist, holidays, sending window, sequence intervals, max touches, and feeds are all set through the chat agent or the Config tab. No file editing required.
Four rules the background agents cannot override. Everything else is driven by your config, so the system stays tuned to your business without the learning loop overstepping.
Your Rules Take Precedence
Any rule you set as a human-source reward ("never pitch on Sundays," "never send to competitors," "mention pricing by the third touch") overrides AI judgment. The AI cannot change your rules, only you can. The skeptical step rejects any idea that would conflict.
Cross-Cutting Changes Through Review
Template strategy shifts, cadence changes, compliance text edits, audience strategy pivots all become suggestions for your approval before the system rolls them out. Single-template tweaks auto-apply; anything that reshapes how the system reaches your whole list waits on you.
Permanent Suppression Respected
Unsubs and blacklist entries are permanent. A re-imported CSV never re-subscribes someone who opted out. Bounces and spam classifications can reactivate only when the contact explicitly returns, so hard-won trust with your sending infrastructure is never spent twice.
Full Pause Control
One command stops outreach sends, the chatbot, the learning agent, and every scheduled job at the next tick. Master keeps working so you can review and adjust. Resume restarts everything.
What Is Actually Running
The system runs as a set of background agents on your server, each on its own timer. All of them respect the system-wide pause flag and skip cycles silently when paused.
Master agent handles the chat interface (5-second tick). This is the one you talk to.
Outreach agent runs every 15 minutes: processes responses first, then scheduled followups, then new contacts. Sends via SMTP for email and the configured provider for SMS.
Learning agent runs every 15 minutes through a three-step cycle: creative ideas grounded in outcomes data, skeptical review against your rules, execute (auto-apply single-template wins, queue cross-cutting changes as suggestions for your approval).
Coding agent runs every 15 minutes through plan, build, review, fix for templates, landing pages, and utility scripts.
Coordinator ticks every 5 minutes: curates background agent logs, surfaces stale pending suggestions, extracts patterns from outcomes for the learning agent, notices deliverability drifts.
Scheduler handles daily maintenance (engagement score recompute, cold detection, orphan cleanup, metrics refresh), holiday auto-pause, and any custom daily cleanup scripts you add.
Memory bank runs continuously as an embedding service so every reply classification, every learning pass, and every master lookup has instant semantic search across experience entries, memories, and product knowledge.
All data stays on your deployment. Your prospects, your templates, your reply data, your learning outcomes. Nothing is shared with any external service the system itself does not call (the SMTP server you configure, the SMS provider you configure, the IMAP mailboxes you configure).