Skip to content
MERAKYCONSULTING

AI systems and outreach · Netherlands

Built with AI. Verified like it wasn’t.

Two things. An SMS system that reopens conversations with leads who went cold, and production software for businesses that have to run on it.

Every value type‑checked
Not assumed. Proven before the code ever runs.
Tests on every push
Unit and browser tests, gating the deploy.
An outside examiner
Reviewed the charity accounts at a statutory year end.

01  SMS reactivation

A database you already paid for, reopened

AI-led SMS conversations with leads who once raised their hand and went quiet. Starts from a CSV. Your reps only speak to people who replied.

The same lead, four messages apart

Not interested

↓ four messages ↓

Yeah ok. Thursday evening works, after 6

See how it works

02  Systems

Production systems, not prototypes

Specified, built and verified as software a business can run on. You get the repository, the documentation and the pipeline.

Built and running

See the systems

Where the AI is

Build time and run time are different questions

Every system here is written by directing AI agents. That is a decision about how it is built. Whether a model runs inside the finished system is a separate decision, made per project.

Build time — AI throughout

  1. Specification

    Data model, architecture, acceptance criteria

  2. AI agents write the code

    Directed against that specification

  3. Static analysis and tests

    In CI, gating the deploy

  4. Deploy

    With a rollback path

Run time — a decision per project

  1. A transaction arrives

    Bank statement, donation, grant

  2. Deterministic ledger

    No model in this path, by design

  3. Posted and auditable

    Reconstructable after the fact

  4. Statutory accounts

    Reviewed by an outside examiner

Build time and run time are separate decisions, made per project. Some of these systems have models in them by design. In the charity accounts and the disclosure validator there is nothing probabilistic anywhere near the numbers, because a probabilistic answer to whether a filing is valid is worth nothing. Knowing where not to put AI is part of what is being sold.

The claims, and what backs them

One person can ship at this scale
Every system here was delivered by directing AI coding agents against a written specification.
Which is why the checking is not optional
Roughly 34,000 lines at static analysis level 9 of 9, with unit and browser tests in CI, reviewed by an external independent examiner during a statutory year end.
You work with the person who designs and ships it
No account manager, no handoff to juniors, no third party between the brief and the build.

Europe

Built for the regulatory environment it runs in

American operators sell this playbook into Europe and adapt afterwards. Here it is the other way round: the lawful basis for an SMS campaign is established before anything sends, recipients are told they are talking to an AI system, and auditability, retention and data residency are architecture decisions rather than a later review.

None of that is a certification and none of it discharges your own obligations. It is how the work is done, and where your own legal advice is the right answer, I will say so.

Start with a call

Twenty minutes to work out whether there is something here worth building. If there is not, I will tell you.

Book a callpr@merakyconsulting.com