Super Intelligence (SI) automation catalog
Every template below is a specification we have engineered and run: what it reads, what it produces, how often, and what counts as done. Scheduled agents carry our own company every day. We build the same templates onto your systems, in your departments, under your controls.
Agents in this catalog run on a schedule inside STRAVYNT with a cost cap and an escalation path. Their live status is on the proof page.
Nothing here switches on by itself. Each template lists what you connect. Outreach templates send only through your own mailbox and domain.
Every template carries success criteria. Where the check can be automated, the template is marked machine-verified and the result is logged.
01 · 37 templates
Step engaged inbound leads through the multi-touch nurture sequence and hand off to meeting booking when they reply.
Due nurture steps fire within 24h
Twice-weekly: scrape ICP-matched companies, qualify with Claude, queue approved targets for the Outreach Engine.
Mon + Wed runs both produce ≥ 1 qualified lead
Tuesday/Thursday cold outbound waves on LinkedIn + email with per-prospect personalization from enrichment_data.
Each run sends ≥ 1 personalized message per queued lead
Weekly: pull SEMrush/Search Console data into seo_keyword_rankings and flag movement worth a Content Writer brief.
One run per week
Make sure no conversion on the site goes unnoticed, including the ones written straight from the browser with no server route in the path.
Every pending conversion is either notified or retried on the next run
Three times a week: publish one operating-problem insight from the topic bank, then SEO and role-document fallbacks.
Mon, Wed, and Fri runs each produce ≥ 1 published insight
Twice-weekly: take Content Writer drafts, schedule them across social channels and the newsletter list.
Approved drafts go out within 48h
The cold-email sender. Pull due steps from outreach_sequences, personalize, dispatch via central send gateway.
Due steps fire daily
Mon/Wed/Fri ICP discovery: Serper → Claude qualification → Hunter+Apollo decision-maker lookup → leads.
Each run yields ≥ 1 qualified prospect
Pull inbound Gmail, classify intent, queue AI-drafted replies for operator review.
No mail unprocessed > 24h
Weekly CMO/CBO/Comms/Market-Intel brief: brand/marketing output volume by type (content, social, SEO, newsletter, distribution), inbound lead share, and demand-by-industry (market intelligence), blended into a brand_output_score 0-100 with the top demand segment.
Weekly run via dispatcher
Weekly marketing-org performance rollup (distinct from the exec CMO brief): output volume by type, distribution reach (email sends), and inbound demand the engine produced → marketing_performance_score 0-100. Reports performance, does not author content.
Weekly run via dispatcher
Daily pipeline health: open pipeline by stage, hot leads (score >= 8), proposals awaiting signature, outreach engagement (open rate), and closed revenue (paid invoices, 30d) → sales_pipeline_score 0-100. Surfaces where deals stall.
Daily run via dispatcher
Weekly acquisition-channel performance: lead volume by source and conversion-by-source (proposal signed / deposit paid / won), best converting channel, and dead channels (>=5 leads, 0 conversions) → channel_efficiency_score 0-100.
Weekly run via dispatcher
Daily social calendar: schedule approved Content Writer drafts on LinkedIn/X/Threads in category windows.
Each due piece publishes in window
Daily digest emailed to ops when prospect-finder leaves leads stuck without a decision-maker email.
Email delivered with non-zero list whenever triage queue is non-empty
Re-engage prospects whose proposal has been pending, day 3 / 7 / 14 nudges, then a work_item if still unsigned.
Exactly one follow-up per scheduled day
Grade recent ungraded content drafts with an LLM judge so low-quality drafts never auto-publish.
All recent drafts graded within 24h
Daily bounded EAL acquisition inside the Gmail send window: advance cheap-qualified supply, research, opportunity maps, dynamic copy QA, then send up to the remaining 3/day cap. Does not use the generic Resend cold path.
Runs at 16:00 UTC inside 14:00-21:00 UTC window
Discover editorial sites worth pitching a linkable asset to (SEMrush competitor referring-domain gaps + Serper segment searches), resolve a contact via Hunter, and file them in link_prospects as pending, human approval required before any pitch is sent.
Files up to 10 qualified prospects per run
Enqueue operator-approved link prospects into the backlink_outreach email sequence (value-first pitches for our linkable assets) with a hard 5-pitches-per-day cap, and sync reply state back onto the prospect record.
Never exceeds 5 new pitches per UTC day
Weekly crawl of every recorded backlink placement: confirm the source page still links to stravynt.com (or the legacy expertailabs.ai), mark placements lost after 3 consecutive misses, and report the live referring-domain count.
Every recorded placement re-verified at least weekly
Weekly live-web brand sweep per org: brand mentions and referring domains, review surfaces scanned for negative-sentiment keywords (work_item on hits), and competitor domains ranked on our category query.
Every active org swept weekly
Weekly partner-pipeline build from warm signals the platform already earned: backlink prospects who replied, sites already linking to us, and agency/consultancy-shaped leads. Files one approval-gated work_item per candidate domain; never sends outreach itself.
Every warm signal becomes exactly one candidate
Captures an inbound lead, enriches it, scores it, writes it to the CRM, and notifies the owner in one pass.
Every submission reaches the CRM within a minute
Pulls keyword positions for the tracked set, stores the history, and flags drops worth a human look.
Positions recorded every day for every tracked keyword
Watches named competitors for pricing, positioning, and content changes and summarizes what moved.
Digest delivered weekly with sources cited
Moves a content brief through drafting, review, and publishing with a human approval before anything goes live.
Nothing publishes without an approval record
Steps engaged leads through a multi-touch sequence and hands off to booking when they reply.
No touch sent after a reply or unsubscribe
Re-scores open leads from fit and engagement signals so the team works the right ones first.
Scores refresh on schedule with the reasons recorded
Drafts long-form content from a brief in the house voice and routes it for review.
Draft matches the brief and passes review
Personalizes and sends approved outreach steps through your own mailbox, with suppression and reply handling built in.
Nothing sends to a suppressed or unverified address
Finds companies that match the ideal customer profile and queues them for qualification.
Every queued prospect carries its match reasons
Reads inbound replies, classifies intent, and routes interested, not-now, and unsubscribe correctly.
Unsubscribes honored immediately
Offers times, books the meeting, and confirms both sides without a scheduling thread.
Meeting on both calendars with no double booking
Assembles a proposal from the discovery notes and approved pricing, ready for a human to review and send.
Every figure traces to approved pricing
Follows up on open proposals on a set cadence and stops the moment there is a reply or a signature.
No follow-up after a reply or a signed document
02 · 19 templates
Monthly: pull usage, compute overage, create Stripe invoices, write the invoice_runs ledger.
Invoice generated for every billable org on the 1st
Run the monthly book-close summary after billing-run finishes: aggregate invoices, dunning events, and AI spend, write a monthly_close_summary outcome per org, and flag anomalies for human review.
One run on 3rd of month at 09:30 UTC
Compile a weekly audit-evidence rollup per org so any compliance review can answer "did the platform do what it said it did" without manual data pulling.
One run every Friday at 15:00 UTC
Compute weekly FP&A variance per org against the 4-week trailing baseline so the operator catches cost spikes, run-rate drops, and pipeline slowdowns the same week they happen.
One run every Monday at 11:30 UTC
Weekly sweep of ai_invocation_logs (canonical AI cost source) to surface cost outliers (>3σ above mean), duplicate AI invocations (same agent/model/token signature within 10-minute windows), and cost-cap breaches, so wasted AI spend gets caught and fixed.
One run every Thursday at 14:00 UTC
Monthly per-workflow COGS attribution: AI cost per agent, cost per successful run, and month-over-month growth, so unit economics are visible per agent and margin alerts fire on uneconomic flows.
One run on 5th of month at 10:00 UTC
Daily AR-aging snapshot per org: open invoice value bucketed by age, DSO computed from trailing 90-day revenue, past-due ratio, oldest open invoice, so the operator never goes more than 24h without knowing the cash picture.
One run every day at 04:30 UTC
Weekly deal-funnel analysis: stage-to-stage conversion ratios across the Loop 1 ladder, median funnel velocity, average deal size delta, and a simple LTV proxy, so the deal pipeline gets quantitative reviews not just dashboards.
One run every Tuesday at 12:00 UTC
Daily composite risk score per org (0-100) blending operational risk (manual-intervention surface), reliability risk (alert incidents/100 runs), cost risk (vendor outliers > 2σ vs 30d baseline), and coverage risk (silent agents). Severity bucketed healthy/watch/elevated/acute. n8n-dispatched.
Daily run via dispatcher
Weekly reconciliation between automation_logs (run counts) and outcome_events (business outcomes written) per workflow. Catches phantom-green workflows (healthy in dashboard but zero customer-facing outcome events) and ghost outcomes (events written without a backing run). n8n-dispatched.
Weekly run via dispatcher
Daily fraud signal scan: dunning ratio (steps / paid invoices, 30d), rapid retry clusters (same invoice > 3 dunning steps in 48h), payment-API call-volume spikes (> 3× prior baseline), and payment-adjacent workflow failure rate. n8n-dispatched.
Daily run via dispatcher
Weekly controls operating evidence: confirms audit/quality/fraud/cyber/risk/internal-audit agents each wrote at least one outcome in the last 8 days, and snapshots third-party exposure (active integrations, open approvals, failing vendors). Surfaces missing-control gaps before an external auditor would. n8n-dispatched.
Weekly run via dispatcher
Weekly economics of the AI workforce: per-agent monthly actual cost (api_cost_tracking, 30d, attributed by service hint), monthly budget (costCapUsd × expected monthly runs derived from cron schedule), utilization %, status (over-budget >80% / on-track / underutilized <10% with 0 runs). Org rollup: workforce_payroll_30d_usd + average_utilization_pct + over/underutilized lists.
Weekly run via dispatcher
Weekly CFO/CRO/CGO financial posture: paid revenue (invoice_runs), AI spend by service (api_cost_tracking), cost-per-outcome, margin proxy %, and AR risk (dunning events) over trailing 30d. Reports money, does not forecast.
Weekly run via dispatcher
Weekly finance-org money picture: revenue, AI spend, margin proxy, collections risk, revenue momentum → financial_health_score 0-100. Does not forecast.
Weekly run via dispatcher
Weekly tax + forensic finance posture: tax filings, fraud signals, expense-audit anomalies, payment workflow reliability → compliance_finance_score 0-100.
Weekly run via dispatcher
Daily AP/payroll posture: outbound spend velocity (api_cost proxy), AP queue, honest payroll-not-connected flag → ap_payroll_health_score 0-100. Lights up with ERP/payroll when connected.
Daily run via dispatcher
Daily: chase failed payments, Stripe retries + payment-method-update emails + tenant pause per policy.
Every failed invoice retried within 24h
Tracks spend against budget lines and raises an alert before a line is exceeded, not after.
Every overrun is flagged before it lands
03 · 14 templates
Compute per-tenant health scores so customer-success can intervene before churn.
One score per active tenant per day
Weekly: fire the right CS playbook per tenant lifecycle stage.
Each active tenant gets the right playbook weekly
Daily lifecycle dispatcher: welcome / first-value / setup-check + monthly value report + promoter referral.
No tenant skipped a lifecycle milestone
Daily queue-health + SLA snapshot per org from support_tickets: open volume by priority, aged-open buckets, median first-response + resolution times, and breach counts on high/urgent tickets open > 4h.
One run every day at 08:00 UTC
Every-2-hour sweep of open support_tickets that lack AI classification. Applies heuristic category + priority recommendation + human-escalation flag, stamps conversation_context, and writes a ticket_classified outcome per ticket. Upgrade path: swap heuristic for callAiSafely() when AI budget surfaces are confirmed clean.
One run every 2 hours
Daily onboarding velocity report from onboarding_configs: step distribution, stuck-org detection (>7 days no movement, not yet completed), time-to-completion stats, completion rate trend. Writes a per-org onboarding_stuck outcome for each stuck org so the operator UI surfaces a nudge CTA.
One run every day at 13:00 UTC
Every-4-hour technical triage: cross-references newly-opened support_tickets against alert_incidents within the same 24h window for the same org. Tickets correlated with an open/recurring platform incident get an ai_root_cause hint stamped on conversation_context so the support team sees it inline.
One run every 4 hours
Weekly CX program health: support ticket flow, resolution rate, customer-health outcome volume → cx_health_score 0-100.
Weekly run via dispatcher
Daily multi-channel contact-center ops: ticket volume by source (chat/phone/email), open backlog, AI-classification coverage → channel_efficiency_score 0-100.
Daily run via dispatcher
Weekly technical support + escalations: technical/urgent open tickets, SLA breach signal, incident correlations → tech_support_score 0-100. Builds on support-team-lead + client-support-engineer crons.
Weekly run via dispatcher
Weekly self-service health: knowledge/community output volume + how-to ticket deflection proxy → self_service_score 0-100. Reads dedicated KB CMS when connected.
Weekly run via dispatcher
Answers common questions from an approved knowledge base and escalates anything it is not sure about.
Escalation on low confidence, never a guess
Sorts client email by urgency and topic and routes it to the right owner with a suggested reply.
Urgent messages reach a person within the agreed window
Drafts the weekly client update from what actually shipped and what is next, for a human to approve.
Every claim in the update traces to a logged item
04 · 13 templates
Aggregate daily run counts, success rates, and AI spend into platform_rollups for Outcomes + Proof dashboards.
One row per (org, day)
Monthly: generate the outcomes report (runs, hours saved, AI spend, lead pipeline) and email the operator on the 1st.
Email delivered with provider_message_id by 1st @ 11:00 UTC
Monday digest: deltas vs last week across every agent, surfaced as a single email + dashboard banner.
Email delivered every Monday by 09:00 UTC
Weekly executive product portfolio health: synthesizes product_delivery_weekly + product outcomes + backlog → portfolio_health_score 0-100.
Weekly run via dispatcher
Weekly operational-efficiency proxy for ESG: cost-per-outcome, process reliability, outcome throughput → efficiency_posture_score 0-100. Honest ESG-not-connected flag until wired.
Weekly run via dispatcher
Friday case-study harvester: pull win-rate + outcome deltas per tenant, draft a publishable case study when thresholds hit.
At least one draft per month if any tenant crosses thresholds
Daily synthesis: read platform state, propose the most leveraged operator actions, file high-priority work_items, and email the operator a one-pager.
One run per day
Generate four retention-grade artifacts per client at the start of each month, monthly report, case study draft, renewal argument, upsell recommendation, so operators always have a ready proof-of-value packet.
Four artifacts generated per active org
Surface pricing, capacity, focus, retention, and growth-lever proposals every Monday based on 8-week trends so operators steer the platform from data, not gut.
One run per Monday
Make every role in the canonical 212-role catalog produce real work: each run, the roles with the thinnest libraries build their next operating asset (SOP, playbook, framework, policy) so all 212 seats have a real body of work before customer volume arrives.
Produces or refreshes up to 10 deliverables per run
Operate the 199 distinct corporate roles as the company unit of autonomy: snapshot role state, probe representative seats, and write verifier-controlled receipts. Does not send prospect email or invent work.
Snapshots all 199 distinct roles
Compiles the numbers leadership asks for into one morning brief, with sources beside each figure.
Brief delivered before the workday starts, every figure sourced
One daily read-out of what every automation did, what it cost, and what needs a decision.
Delivered daily with every agent accounted for
05 · 2 templates
Turns a signed agreement into accounts, folders, kickoff tasks, and a welcome sequence.
Every signed client has a complete kickoff checklist the same day
Requests, tracks, and securely stores the access a new engagement needs, and nudges until it is complete.
Secrets stored encrypted, never in email
06 · 2 templates
Weekly enterprise-risk leadership rollup: synthesizes risk_score_daily, fraud_signals, cyber_risk outcomes → enterprise_risk_score 0-100 (higher = safer).
Weekly run via dispatcher
Weekly audit leadership assurance: synthesizes reconciliation, SOX controls, phantom-green findings → audit_assurance_score 0-100.
Weekly run via dispatcher
Start with one department
A strategy call maps your systems to this catalog and comes back with a build order, not a sales deck.
