Legalfly Chief: A.I. Slop Creates More Work Than It Saves
Set Trending Topics as a preferred source on Google.
A.I. was supposed to save work, and now it is creating more of it. That is the warning from Ruben Miessen, founder and chief executive of the Belgian legal tech start-up Legalfly, in a recent commentary. European start-ups have brought plenty of innovation with generative and agentic A.I., he argues, but they have also flooded the world with junk. He sees the problem most clearly in the legal sector.
Endless Red Markups
Mr. Miessen cites the British author and journalist Jamie Bartlett, who speaks of the burden of “A.I. clog.” Cleaning up the mess means A.I. does not save time but actually creates more work, Mr. Miessen says.
In day-to-day legal work, this shows up in contract negotiations, for example. Many law firms still have no clear rules for using A.I., he says, so some teams rely on general-purpose language models such as ChatGPT or Claude to draft and review texts. The result is documents that come back covered in red markups because A.I. editors delete and replace large sections. According to Mr. Miessen, many tools cannot tell the difference between applying the letter of the law dogmatically and giving ground to move a negotiation forward.
Congress Is Struggling With A.I. Drafts, Too
An example from Washington, which Mr. Miessen alludes to, shows that the problem is real. According to a Politico report, the Office of the Legislative Counsel in the U.S. House of Representatives, which drafts bills for lawmakers, is increasingly busy dealing with flawed A.I.-generated texts. A former adviser said the office spends more time fixing A.I.-drafted legislation than it would take to write it from scratch. Typical errors include faulty citations and confused legal terms.
“Workslop” can also be measured outside the legal world. In a survey of 1,150 U.S. office workers by researchers at the Stanford Social Media Lab and the coaching platform BetterUp, about 40 percent said they had received sloppy A.I.-generated work from colleagues in the past month. Each incident took nearly two hours to fix on average.
What Mr. Miessen Proposes
Mr. Miessen sees the solution in A.I. tools that know when to leave things alone. Instead of rewriting entire contracts, they should follow a clearly defined playbook from the law firm that sets out its positions, policies and limits. The tool would then check for compliance issues, flag deviations and suggest only changes that fit the playbook. He expects users to soon demand A.I. tools that edit with precision, relevance and insight.
Legalfly sells exactly this kind of tool. Founded in Belgium in 2023, the company offers A.I. workflows for contract review, document drafting, due diligence and research. According to the company, it has more than 200 customers in 33 countries and has raised more than €17 million from investors including Notion Capital, Redalpine and Fortino Capital.
Law Firms Keep Buying
Competition in the legal A.I. market is fierce and far better funded. The Swedish start-up Legora is now valued at $8.8 billion and counts the construction group Strabag among its customers. In the United States, Harvey is considered the market leader, and providers such as Luminance from Britain also compete for clients.
The billions investors are pouring into the sector show that companies and law firms expect A.I. to save time despite all the problems. Mr. Miessen’s criticism is therefore aimed at poorly deployed, general-purpose tools. That a vendor of specialized software draws this distinction also serves its own business interests.
