Technology
I rarely start from zero.
Building my own products and client software continuously has produced a growing library of technology, intelligence and software foundations that can be used again.
So I do not have to reinvent everything for each product, and more attention can go to what is genuinely unique about your company or product.
What sits in the library
Six areas, in plain language.
The larger titles are written for founders. The small labels underneath are for people who want to recognise the technology. You can understand this section without reading a single label.
Thinking and understanding
Systems that read, compare, summarise and assess information, with the relevant context of your company included.
AI agents · knowledge · prompts
Researching
Gathering information, data and sources, and giving them structure.
research · search · clustering · data processing
Getting work done
Software that carries out next steps on its own, within clear boundaries.
workflows · background jobs · notifications · actions
Working together
Several digital parts with different roles, able to review each other's work.
orchestration · specialist roles · review flows
Product foundation
The reusable base serious software needs, from signing in to permissions and teams.
auth · users · workspaces · permissions · multi-tenant
Connecting and charging
Connecting software to the systems you already use, and making commercial functions possible.
integrations · payments · subscriptions · credits
Reusable technology speeds things up, but it is not a ready made product. Every part is adapted to the company, the process and the people who will work with it.
Digital coworkers
Software that does not only wait for someone to press a button.
A digital coworker is smart software that can understand, prepare, carry out or check parts of the work within clear boundaries. Not a human, not a chatbot with a new label, and not endlessly autonomous.
Understand
What is going on?
Use context
Which knowledge and information is relevant?
Act
Which step is the system allowed to take?
Check
Is the result good, or should a person look at it?
The goal is not to imitate a person. The goal is to give software a clear role in which real work gets done, with human control wherever that remains necessary.
Classic automation
Example: form submitted, send email.
Digital coworker
More context, more steps, with a person where that matters.
What such a coworker can do
What a digital role can look like.
Which role makes sense depends on the work that keeps piling up in a company. These are examples of tasks this technology makes possible.
Digital intake assistant
Reads a request, collects missing information and prepares the next step.
Digital researcher
Gathers information from different sources and brings the insights together clearly.
Digital content specialist
Works from a brief, company knowledge and quality rules towards a first draft.
Digital quality reviewer
Checks output against fixed criteria before anything is published or sent on.
Digital account assistant
Keeps information up to date, flags changes and prepares actions.
Digital product specialist
Combines customer questions, data and business rules into prepared advice.
Nuance
The question is not who you replace.
The more interesting question is which work people no longer have to do by hand, so there is room for judgement, creativity and personal contact.
A good system can take feedback, new company context and stored preferences into account, so the next run fits better. How far that goes depends on the product and the control you want to keep.
Under the bonnet
Several parts. One working system.
What feels like one coworker from the outside is, inside, an interplay of roles, knowledge and checkpoints.
- Letting several specialists work together
AI orchestration
Letting different AI roles work together, under control, inside one process.
- Using company knowledge
Knowledge / context
Letting the system use relevant information, rules and context from the company.
- Deciding what happens
Workflows
Setting out which steps may run, and when.
- Human control
Human in the loop
Letting a person review or decide where that is needed.
- Connecting to existing software
Integrations
Letting the system work with tools and systems already in use.
The toolset is secondary. The value sits in what gets built with it.
From question to result
Proof
I build this myself because I want to know what is actually possible.
The library does not come from theory, but from my own products and client software. What works can be used again. What does not, dies early.
Where the technology was built and tested.
A short selection of products where parts of the library run in practice.
Influde AI
A link-building co-pilot for planning campaigns, creating content and managing placements from one workspace.
Built and tested here: AGENCY WORKFLOWS · AI ORKESTRATIE · MULTI-TENANT
SearchBalloon
Visual search-world intelligence that reveals hidden demand, market clusters and content opportunities.
Built and tested here: Visuele search-world intelligence die verborgen vraag, marktclusters en contentkansen visueel zichtbaar maakt.
Fotospots
A community-driven map for discovering and sharing locations for landscape photography.
Honest
Not every problem needs AI.
Sometimes a simpler system is better. Sometimes the process or the positioning has to change first. Technology follows the question, not the other way around.
Next step
Curious what of this is useful in your company?
You do not need to know which technology you need. Start with the problem or the opportunity.
