Production AI, built and operated by us.
Maciej builds production AI across agents, RAG, data pipelines, content systems, and custom applications at Enterprise Bot. Hendry builds and operates 7+ production AI marketing engines, publicly documented across 105+ curated iterations. The apps below are the systems we run.
The systems we've built.
Each app solves a specific marketing or engineering problem we hit in production. Some are open. Others require access because they run against live data or client environments.
growthsetting builds and operates production AI marketing systems.
This page is the apps library. Every tile above is a live system we use to run growthsetting or to prove a capability we can apply to future client work. Operator log at hendry.ai.
The proof is three layers: engineering, operational, and live.
Each founder brings a distinct track record. The apps on this page are the third layer.
Maciej builds production AI.
Across agents, RAG, data pipelines, content systems, and custom applications. Enterprise Bot today. Google Assistant before that. 8 AI systems built under growthsetting.
Hendry builds, operates, and leads.
7+ production AI marketing engines inside an 11-engine framework. 75+ extracted principles, each traced to a specific failure and fix. Publicly documented.
The apps above are live systems.
Every tile on this page runs against a real backend. Gated apps connect to production data or client environments. Open apps are safe to poke at. Nothing here is a mock, a demo, or a screenshot.
growthsetting is pre-revenue. The first external AI engagement under the brand is the next proof point. It will be named here when it ships.
Every app runs against a real backend.
Gated apps connect to production pipelines or data we do not expose publicly. Open apps have their trust boundary at the UI layer and are safe to poke at. The operator-log methodology behind the marketing engines lives at hendry.ai/ai-marketing/operator-logs.
For agencies pitching AI marketing outcomes.
If you need live examples to point at in a client pitch, the apps here are those examples. If you are a startup shipping a marketing AI feature, the same engineering patterns apply.
Why this might not work. If an off-the-shelf SaaS tool solves your problem, use the SaaS tool. If your team has no operator to own the system after it ships, fix the process before adding AI.
Common questions.
Terms, defined.
- Context engineering
- The discipline of designing the information layer all AI systems share: brand voice, ICP definitions, validation rules, orchestration. Context engineering determines output quality. Tools change quarterly. Context architecture compounds. This is Hendry’s core discipline, with published methodology at hendry.ai.
- Evidence-based validation
- A validation pattern where LLMs must output work product such as lists, counts, or quotes as proof of compliance, rather than answering a binary pass/fail prompt. Evidence-based checks have been measured at 97% accuracy compared to 16% for binary prompts.
- Three-proof rule
- The growthsetting credibility framework. Three distinct proof layers are kept separate in any claim. Maciej’s engineering track record at Enterprise Bot and Google Assistant. Hendry’s operational track record across 7+ production marketing engines and 105+ curated iterations. The live apps on this page. growthsetting is pre-revenue; the first external AI engagement under the brand is the next proof point.
- Operator log
- A publicly documented record of production AI operations, curated to extract reusable principles. Hendry’s operator log at hendry.ai/ai-marketing/operator-logs covers 105+ curated iterations and 75+ extracted principles across 7+ production engines.
Start with the apps. Talk to us about the rest.
Register to request access to the gated apps. If you are pitching a client or scoping an AI project, sign in first and we will pick up the conversation from there.