Autonomous AI Marketing Operator

namtool is an autonomous AI marketing operator for entrepreneurs and small business owners — currently in soft launch. Users work with Marko, the AI inside namtool, who owns the marketing loop so their attention stays on the product. Underneath: a deep multi-phase agentic pipeline that orchestrates multiple LLMs and generative models, rooted in real performance data.

Product

Entrepreneurs and small business owners need their attention on the product, not on weekly marketing judgement calls — or on briefing an agency that never fully learns the business. namtool removes that trade-off: Marko (the AI users interact with) runs strategy, creative, campaigns and measurement in the background.

~10-minute onboarding from a URL and/or Instagram — research and scraping build brand profile, styling, voice, competitors and USP. The user stays as in control as they want: Marko asks to confirm among different styles, directions and preferences, stays on-brand, and learns from those choices.

Ongoing work is minutes a week: Marko emails when the weekly strategy is ready to review; a few clicks and the week is set. Want to go deeper? A dashboard shows the data, and Marko explains the why behind every decision when you care to look.

Strategy sits on a code-backed goal model — Goals (Awareness, Win customers, Grow revenue) → Objectives → weekly Initiatives, with Key Results as scoreboard and diagnostics on one deterministic driver tree (Reach → Engaged visits → Customers → Revenue). Initiatives stay agile; the scoreboard does not churn.

Behind that: multi-model orchestration, a config-driven model registry, a long observable pipeline with parallel inference and AI-shaped branching, critic agents with retries loop-backs, and deterministic guardrails that keep AI grounded in measured outcomes and safe bounds.

Key Features

• Marko — the AI users interact with — owns strategy, creative, campaigns, measurement and weekly course-correction
• Built for entrepreneurs and small business owners — marketing in the background
• ~10-minute onboarding from URL and/or Instagram — brand profile, styling, voice, competitors, USP and angles
• User-controlled autonomy — directions and preferences; on-brand; remembers and learns
• Minutes a week — email when review is ready; few clicks to approve the week
• Optional depth — dashboard plus Marko’s explanation of why behind each decision
• Structured goal model — Goals → Objectives → weekly Initiatives; KRs + deterministic driver tree
• Multi-model agentic orchestration — best LLM / image / video model per task, across providers
• Provider-agnostic, config-driven model registry — swap models without rewriting business logic
• Deep multi-phase production pipeline — dozens of discrete, independently observable steps
• AI rooted in real data — measurement and analysis drive decisions, not generation alone
• Deterministic-first hybrid design — code owns guarantees; AI owns judgement
• Conditional, AI-shaped control flow — pipeline branches mid-run from earlier model outputs
• Parallelised inference — concurrent AI calls with dependency-aware scheduling
• Multi-agent critic workflow — critics review other agents’ output; retries, re-passes, suggested fixes
• Deterministic guardrails — spend and brand constraints enforced in code
• Closed-loop optimisation — adaptive targets from observed performance
• Retries and fallbacks — timeouts and weak outputs take alternate paths
• Full-stack architecture — orchestration, providers, deterministic layer, observability, production runtime
• Stack — Next.js · TypeScript · Postgres + Drizzle · Supabase · Render · Cloudflare · Meta Graph API · GA4 · WooCommerce · Stripe
• Production bar — designed to run unattended, not a single-model prototype