Now booking projects for Q3 2026. Limited availability.
AI SYSTEMS STUDIO

Everyone automates.Most automate the wrong things.

I build custom production software with embedded intelligence: systems that classify, match and decide, with your team in control. Only where it creates measurable value.

See selected work

Limited engagements · Punta Cana and Europe

process audit running
manual workload
Order matching14h/wk
repetitivevalue leak
Invoice approval11h/wk
repetitivevalue leak
Lead enrichment8h/wk
repetitivevalue leak
find leak
92
automation fit
route by confidence
system after build
Automate
fast lane
Review
exceptions
Measure
ROI live
47h
saved/wk
€147k
recovered
0.3%
errors
Scroll

THE PROBLEM

Software is cheap. Knowing what is worth building is rare.

With AI tools anyone can spin up a website, wire an automation or ship an MVP. The barrier collapsed.

Businesses do not gain value from more automations. They get burned by them. Flows nobody can debug. AI agents that work in demos and break in production. Solutions built by people who do not understand the business they are meant to serve.

The shortage is no longer builders. It is people who know exactly what is worth building and who stay involved until it genuinely works.

This is what most automation projects look like in production.

Paul Den Hartogh
Founder, Den Hartogh Solutions
Builds and operates own products
Works Punta Cana ↔ Europe

WHO YOU'RE WORKING WITH

I'm Paul. I diagnose, design and build the system myself.

No account managers, no handoffs to a dev team you never meet. The person who maps your process is the same person who ships the code and stays until it works.

I built SalesLobe, an AI reply system running at multiple outbound agencies. I built the product matcher that cut IB Data's manual matching by ~85%. Production systems, running today, operated by the teams that own them.

Every diagnostic conversation on this site lands directly with me.

MY APPROACH

Five principles that shape every engagement.

01Strategic

Find what matters

Most businesses don't need more automation. They need the right automation. I map where your time, money, and attention actually leak, then tell you what's worth building, what isn't, and what I'd build first.

process-analysis.dhs
6 leaks
Process Analysis
DHS audit · Q2 2024 · 30-day window
47h
wasted / week
€147k
annual recovery
6
processes found
Operations
Order matching
HIGH14h
Finance
Invoice approval
HIGH11h
Sales
Lead enrichment
MED8h
Support
Ticket routing
MED7h
Operations
Shipment updates
LOW4h
HR
Contract review
LOW3h
Based on process interviews + system logsFull report →
02Honest

Think first, build second

If automation won't add real value, I say so. Sometimes the right answer is a documented process, not a system. I won't build something just because I can.

scoping-decision.dhs
Q4 2024
3
automate now
2
skip for now
€147k
total upside
Automate
Order matching
14h/wk+€82k/yr
Invoice approval
11h/wk+€44k/yr
Lead enrichment
8h/wk+€21k/yr
Not yet, document as SOP first
Client onboarding flow
Needs human judgment on edge cases
Proposal review
Variable too high, document as SOP first
Reviewed with client · Oct 2024View plan →
03Transparent

You stay in control

No black boxes. Every system I build is meant to be understood and operated by your team. You see what happens, adjust without calling me, and own the result.

system-config.dhs
LIVE
Automation rules
Auto-approve matches above threshold
Flag low-confidence for human review
Send replies without review
Email digest on daily summary
Confidence threshold87%
0%100%
Audit log · changed by client
2m agoclientThreshold raised 80% → 87%
1h agosystemRule added: vendor fallback
YesterdayclientConfig exported
04Measurable

Measurable value

Time saved. Errors prevented. Revenue lifted. We agree on what success looks like before I start, and make it visible after with dashboards that don't require me to read them.

impact-dashboard.dhs
Last 8 weeks
Hours saved
47h
/ week
+34%
Error rate
0.3%
of all orders
−89%
Annual impact
€147k
est.
Hours saved / week
47h
Week 1Week 8
Baseline agreed before project startFull report →
05Ongoing

Built to evolve

Systems that live and improve. We measure, learn, adapt. An engagement doesn't end at delivery. It ends when the system is genuinely better than the day it shipped.

system-changelog.dhs
LIVE
v1.4
current version
6mo
in production
+31%
accuracy gain
May 2025
v1.4
Improvement
Model confidence +4% after 200 new labels
Mar 2025
v1.3
Rule
Vendor fallback for ambiguous SKUs added
Feb 2025
v1.2
Fix
Edge case: multi-line product descriptions
Jan 2025
v1.0
Release
Initial deploy · 94% exact match rate
Nov 2024
Baseline
Manual process recorded: 14h/wk
Next review: Jun 2025Full log →
HOW I ACTUALLY BUILD

Real systems. Not flows.

Every engagement results in a tailored application your team can understand, operate and evolve without me.

Human in the loop

AI proposes. Your team decides on exceptions. Full override and audit trail. Never blindly autonomous where it matters.

Scoring + fallback

Exact rules first. Then confidence scoring. LLM only as a smart fallback. Always transparent and adjustable.

Self-service rules

Your team adjusts matching, thresholds and routing themselves. No deploy needed. Everything logged and reversible.

Full observability

Every decision logged. Dashboards your team actually uses. You never need to call me to understand what happened.

These patterns power the IB Data product matcher and SalesLobe. Production-grade. Team-owned.

HOW WE WORK

From problem analysis to working system.

One clear path: first we find the real problem, then we agree on exactly what to build, then I build it. No scope surprises, no dependency on me afterwards.

01
Diagnostic
10 minutes · Free

The chat on this site or a short call. I map how your business actually runs and where time, money and attention leak. You get a concrete map of what to automate and what to skip, whether we work together or not.

02
Scope
A few days · Fixed proposal

I translate the diagnosis into a precise plan: what the system does, what it deliberately doesn't, what each part delivers, and a fixed price range. You know exactly what you're buying before anything is built.

03
Build
4–8 weeks

Focused build with a working demo every week. You see progress in your own data, not slide decks. Course corrections happen during the build, not after.

04
Handover
Yours, fully

You get the repository, the documentation and the dashboards. Your team adjusts rules and thresholds without deploys and without calling me. You own the system.

05
Evolve
Ongoing · Optional

We measure the results against what we agreed. Systems that learn need tuning. I stay available for that, but the system never depends on it.

PRICE INDICATION

Diagnostic
Free
Chat or call, concrete map included
Focused build
$8k – $15k
One core process, 4–6 weeks
Larger system
$15k – $25k
Multiple processes, 6–8 weeks

Exact pricing always follows the scope. Fixed range, no hourly surprises.

SELECTED WORK

Selected work.

01 / 03
01Own Product · B2B SaaS · 2024–2026

Built to govern the AI reaching your prospects and clients.

SalesLobe is governance-first AI reply management. I built the layer most "AI sales tools" skipped: the one where a human stays in control of every AI-written word.

Live at multiple agencies · Learns from every correction
SalesLobe: AI reply triage inbox with classification confidence and reasoning

Where I Work

Two time zones. One way of working.

Home base shifts between Punta Cana and Arnhem: Caribbean sun in the morning, European clients by afternoon. Most work happens remotely. When a whiteboard session or face-to-face kickoff makes the difference, I show up.

Paul Den Hartogh
Paul Den Hartogh
Founder · Den Hartogh Solutions

START THE ASSESSMENT

Not sure where AI would actually move the needle?

10 minutes. No sales pitch. I ask how your business truly runs, what eats time, what frustrates the team, and what slips through. By the end you have a concrete map: what I would build as a custom application, what I would not touch, what each part would deliver, and a realistic price range. Then you can build it yourself, have me build it, or do nothing. All three are valid outcomes.

Engagements range from focused 4 to 8 week builds to longer partnerships. I work with a limited number of clients at a time.

TAKING NEW ENGAGEMENTS
10-minute diagnostic chat
No pitch. I ask how your business actually runs, find where value leaks, and give you a concrete map with pricing.
paul@denhartogh.solutions