OpenClaw and the New Digital Reality
OpenClaw went from zero to 100,000 GitHub stars in just fourteen days. To put that into perspective: that is faster than most enterprise procurement processes need to schedule a kickoff meeting. The pattern is becoming familiar across large organizations. A tool goes viral on Tuesday, has a community marketplace by Thursday, and by Friday someone in accounting has already forwarded a LinkedIn post to the IT department asking, “Why do we not have this yet?”
Meanwhile, the security team is still reading the README, trying to determine whether giving an AI agent access to the filesystem is technically acceptable or a career-ending mistake.
What OpenClaw Actually Is
If you have never heard of OpenClaw, imagine this: instead of a chatbot that tells you how to do things, you get an AI sidekick that simply goes off and does them. OpenClaw is a small program that runs locally on your own laptop and can take real actions for you. It can read and write files, run commands, browse the web, manage your inbox, or clean up your calendar while you are off making coffee.
You interact with it through messaging apps you already use, like WhatsApp, Telegram, Slack, Teams or iMessage, which makes it feel a bit like texting a coworker who never sleeps and never forgets anything. Because everything stays on your machine unless you explicitly decide otherwise, you maintain control over your data.
Developers can also teach it new tricks through plugins, called skills. The community is already building automations, integrations, and all sorts of enhancements, which is a big reason the project exploded so quickly.
In short: OpenClaw is like giving your computer a pair of hands and a bit of initiative.
The Great Digital Divide
Walking into any modern office and the gap between digital frontrunners and digital traditionalists is impossible to miss.
- On one side are developers using the newest AI tools to negotiate meeting times over email, auto-generate documentation, and manage their calendars while they keep coding.
- On the other side are colleagues storing passwords in an Excel file named Important_Stuff_Final_Final.xlsx, protected with a password that is conveniently written on a Post-it note stuck to their monitor.
While a lobster-themed AI manages someone’s agenda, someone three desks away is still trying to figure out why their pivot table refuses to sort.
Velocity as the New Normal
Adoption curves no longer stretch across months or quarters. They rise almost vertically.
Traditional enterprise software used to move through lengthy evaluations, pilot phases, and gradual rollouts. Today, people are using a tool at scale before the first security advisory is even drafted. The speed is not simply high; it is exponential.
For a company like Proximus, this creates a familiar tension. This is not a startup where developers can install anything they like. There is governance, oversight, and the level of caution expected from a critical infrastructure provider. But the classic rhythm of request, review, and approve simply does not align with technology that refuses to slow down for administrative reality.
By the time the assessment is finished, version 2.0 is out, the community has moved on to a different fork, and the moment has already passed.
Security at Light Speed
For security engineering, this speed is not just a challenge. It is the job.
The attack surface does not simply grow, it mutates.
- Yesterday’s focus was prompt injection.
- Today’s concern is autonomous agents with filesystem access and unvetted community plugins.
- Tomorrow’s threat does not exist yet.
Traditional responses such as blocking, assessing, and controlling assume a static target. But tools like OpenClaw evolve faster than any review cycle. Blocking access at the proxy level feels like trying to stop a river with a tennis racket. You are doing something, but everything still gets wet.
From Gatekeepers to Navigators
This shift is not only technological but organizational.
Security teams are learning that saying no is not a control, it is a transfer of risk to shadow IT. When tools are blocked officially but quietly run on personal laptops or cloud instances, security loses visibility while gaining nothing but the illusion of control.
The better alternative is to accept that developers will experiment, because the productivity gains are real and the barriers to entry are extremely low. The task is to build safe landing zones instead of walls.
Examples include:
- Containerized environments where AI agents can read the repository but not the database configuration files.
- API keys with scopes so tight they could not order a coffee if they tried.
- Network segmentation that treats every AI agent as potentially hostile, which from a security perspective is exactly the right mindset.
The Lobster in the Room
Back to that Excel password manager. That colleague is not stupid. They are working with the tools he understands and have been trained on. The developer delegating their calendar to a crustacean-shaped AI is doing the same thing: solving a problem with the tools available.
Organizations are stuck navigating between two extremes.
- Too much oversight turns the company into a digital museum.
- Too little turns it into a disaster case study.
The ideal balance is controlled descent. Accept that people will experiment and make sure the blast radius is manageable.
The New Reality
OpenClaw will probably be obsolete next quarter, replaced by something with an even stranger mascot. The specific tool is irrelevant. What matters is understanding that technology adoption no longer follows enterprise timelines. It behaves more like a viral spread.
The companies that thrive will not be the ones with the longest evaluation checklists. They will be the ones that learn to secure experimentation, distinguish between risky and foolish (Excel passwords) and risky but manageable (sandboxed AI agents), and accept that in this new velocity, perfect security is the enemy of practical security.
The future belongs to teams that can keep up, ideally without ever storing passwords in Excel again.
A Philosophical Question to Leave the Room Thinking
If technology keeps evolving faster than our policies, habits, and even our intuition can follow, then what does real wisdom look like for an organization: holding tightly to what feels safe, or learning how to stay balanced while everything around it is in motion?
Bram
AI Security Engineer
Proximus Ada