Among tourists, Belgium has always been famous for its chocolate, beer and Manneken Pis. Among lawyers, however, Belgium is becoming quite famous for its flourishing legal tech scene. Probably the best-known company is Ghent-based Henchman, which markets a software product with the same name.
It’s rare to see the software in action, however, as the company’s website only provides a high-level view of its software, instead inviting customers to book a demo.
In this review, we will therefore take a profound look at how it really works and what can you expect from it.
Henchman initially marketed its software as a “micro-tool” that is focused on doing exactly one thing: giving lawyers quick access to their past clauses. In the course of 2023, Henchman also added a document outlining and document querying tool to “interrogate” your documents with AI (similar to what Microsoft Copilot and many other products, including our own ClauseBuddy do), but the core product still revolves around clause searching.
The product’s central goal is to solve your nagging feeling of having written a clause in the past but being unable to find it. After searching in your emails for several minutes, opening a few old files and perhaps harassing a few colleagues, you give up and write the clause again, probably for the fifth time in your career. (Of course, two weeks later, you stumble over the initial clause and notice that it was slightly better than the new version you had written.)
To solve this problem, Henchman essentially copies your entire history of text documents to its own servers, then crawls through them and automatically splits those documents into clauses, making them searchable through an easy interface.
It is a common misconception that Henchman would search within your existing document management solution (e.g., iManage, NetDocuments, Sharepoint or Dropbox), i.e. that it would use the search facilities of those solutions to extract clauses, but would leave the data within the document management solution.
In a perfect world this would indeed be the preferred approach, but the technical reality is that document management solutions are focused on the document-level and simply do not offer sufficiently advanced technical facilities (so-called “API end-points”) to search documents at the clause-level. Besides, even if document management solutions would offer such clause-level technical facilities, they would probably be too slow for a great user experience.
Henchman is therefore forced to take a copy of all the files in your document management solution and transfer them to its own servers, where it extracts the clauses from each file that looks like a contract, and stores them in a database optimised for text searching. Henchman will then periodically synchronize your document management solution with its own clause-database, so that new files (Word-documents, email attachments, etc.) that get added to your document management solution will automatically become available in Henchman’s clause-database.
The importance of Henchman’s data maximisation approach cannot be underestimated: that nagging feeling of “Where is that [extensive share assignment] clause I’ve written several years ago?” will only be solved if Henchman has access to your entire document set; even just a few missing files would completely undermine your confidence in the product. Furthermore, this also ensures that any new clauses added to your document management system also become automatically retrievable, even if you may have to wait a few days for them to show up.
The beauty of Henchman’s setup is that all of this happens automatically, without any involvement of lawyers. Even though the IT-department will have quite some work to do to establish the technical bridge towards Henchman's servers, and the security department will probably also need to be involved (e.g., to drill a hole in the organisation's firewall to allow for Henchman's document copying, or to exclude some folders that should not be in scope of Henchman's contract crawling), the lawyers will obviously not be bothered by these technicalities, and can simply enjoy the benefits of being able to search in an enormous clause database.
Henchman does not publicly advertise the pricing of its product, instead referring to its sales team. The basic subscription is around 100 EUR per person per month, with a minimum of five users and a contract term of three years. Discounts apply in higher user volumes, while the monthly price per person goes up if you want more frequent synchronisation and additional access controls among the users.
This price is relatively steep, and we have frequently heard lawyers joking that Henchman is “a micro-tool with a macro-price”. However, in all fairness, the price is probably correct when you factor in the significant infrastructure costs that Henchman is facing. For each law firm and inhouse legal team, Henchman will have to set up a dedicated environment at Amazon Web Services (AWS), in which it will host the gigabytes of data copied from its customers’ document management system. The IT-teams of both Henchman and the customer will also need to be intensively involved, because integrations between different IT-environments are not exactly a walk in the park. From this perspective, you can understand why they require at least five users and a long contract term, because the setup costs are significant.
One should also not underestimate the layers of security that Henchman must add to keep its customers’ data safe, considering that Henchman has access to massive amounts of highly sensitive details of its customers’ clients. Henchman strongly emphasises its security, with two certificates in place (ISO 27001 and the US equivalent SOC 2). In the past, and perhaps also still today, it used the services of the world-famous Belgian company Intigriti, which pays white hackers to actively find security holes in a company’s IT-products.
The first weeks
In practice, Henchman is a plugin for Microsoft Word, which presents itself as an easily accessible side-panel in Microsoft Word or (less frequently used) Outlook.
In this side-panel, you get a polished interface that invites you to submit a search query. Assuming you can remember a few good keywords about that famous clause you once wrote, Henchman will quickly present you a list of clauses matching those keywords. You can then browse through that list of clauses in order to find your exact match, and insert the relevant clause into your contract. Alternatively, you can ask GPT to translate or rephrase your clause, which Henchman calls "Smart Replacement" — e.g. to put a certain term in plural, or swap a defined term (e.g., “Customer”) for some other word (e.g., “Buyer”). Before you insert a clause, you can also compare it to the currently selected clause in your MS Word document to see the differences.
We frequently hear from users that the initial weeks with Henchman feel almost magical. Thanks to all the automatic clause splitting being done in the background, lawyers can indeed easily search for clauses without having spent any effort. Compared to the old process of searching in emails or old files, asking colleagues, etc. Henchman drastically speeds up your contract drafting speed. The nagging feeling will therefore vanish in the first weeks.
After the honeymoon
Time and again we have heard that the initial excitement for the product tends to wane after a few weeks. While using a tool such as Henchman remains many times better than using no specialised drafting tools at all, users get tired of crawling through hundreds of search results. Initially, the sheer amount of search results doesn’t bother anyone: wading through many results is more pleasant than finding nothing in old emails or documents. But those feelings apparently change. Sure, Henchman reduces your searching time from 10 minutes to, say, 2 minutes — but is this really the best way to search for your content?
Anyone using these kinds of tools will also question Henchman’s claim that it can boost productivity with one hour per day, per lawyer. Of course, this claim must be limited to legal experts who spend their entire day drafting (only a small minority of lawyers do), but even then we have not yet encountered a lawyer who is so poorly organised as to lose an entire hour searching for contract clauses.
After using a tool like Henchman for a few weeks, lawyers also experience that Henchman’s primary use case — finding that one particular clause you’ve written in the past — does not actually happen that often. In practice, it seems to happen around one or twice per week for frequent drafters, while throughout a week 95% of clause searches are for more mundane clauses. However, being confronted with the feeling of having to rewrite what you’ve already written is emotionally so frustrating that lawyers probably over-estimate its presence.
In its demos, Henchman strongly appeals to these negative emotions, and will therefore focus on this “needle in the haystack” clause problem, which the software handles really well. What Henchman does not focus on, however, are the day-to-day clauses that must be inserted. For those searches, you will notice that Henchman’s focus on keyword search (helped by simple tagging and “favourites”) starts to break down, because keyword searches just don’t work well for searching mundane, day-to-day clauses. For these day-to-day clauses, having some structure (e.g., clauses organised by legal topic) work so much better and so much faster — but this cannot be done automatically and would require effort from the lawyer, so Henchman cannot offer it.
Accordingly, when searching for day-to-day clauses for which no unique keywords can be found, Henchman’s data maximisation approach will result in hundreds if not thousands of results, with a poor signal-to-noise ratio and significant amounts of time lost on reading through lengthy search results.
The reason why keyword searches do not work so well is that contract clauses tend to reuse a small vocabulary with highly popular words (e.g., “shares”, “buyer”, “liability”, “material”, etc.). Even wildly different clauses will use those same words over and over again, so that on average clauses have relatively limited unique keywords. Search companies such as Google solve this problem by relying on the feedback of billions of users worldwide, so that even searches such as “sports scores” or “public holidays” will give good results, because Google can “learn” how good a particular result is, based on whether users come back and click on subsequent search results. Legal tech tools such as Henchman obviously do not have this luxury, as only a small number of users will provide feedback on a particular clause database.
Truth be told, Henchman tries to alleviate this problem, e.g. by allowing users to tag / label results, assign "favourites", filter results by document title and assign clauses to "lists". In its marketing, the product also strongly focuses on the clause counting metric, which — thanks to the fact that it has access to the history of all your clauses — allows lawyers to sort clauses by popularity. While this feature does have its merits, serious drafters will find this a dubious metric — is the most “popular” clause indeed the best one for a given contract? And is it fair to measure “popularity” by simply counting the frequency? For example, if an inhouse lawyer tends to work on deals where the initial documents are drafted by the counterparty, the most “popular” clauses in Henchman’s database will in fact be the completely wrong choice.
In a certain way, the automatic filters and the popularity metric highlight the limitations of Henchman's conceptual approach. A completely automatic approach only brings you this far, and Henchman's automatic filters and popularity metrics feel like stopgap measures. Completely automatic clause extraction has the huge benefit of requiring no initial effort from the side of the lawyer, but almost every legal teams will want more at some point.
Finally, there’s the elephant in the room: security, data protection and confidentiality.
As pointed out above, Henchman’s security efforts are world-class, without any doubt. But the likelihood of being targeted by hackers is completely proportional to how interesting your data is, and Henchman proudly claims to “mine” hundreds of millions of clauses from a staggering number of highly sensitive documents from its customers.
No matter how tight its security, Henchman is therefore playing a dangerous game, because no money in the world can shield you from the interests of motivated hackers. Microsoft, for example, is known for spending billions on security every year, but also suffers around one major hack on average every year; similarly, even companies whose entire existence is around security become a target if the stakes are high enough.
But the hacking problem is not the biggest downside of the data maximisation approach. An even bigger problem is its customers’ compliance with data protection legislation (such as the GDPR) and confidentiality obligations. The data maximisation approach — collecting all contracts so you can find that one needle in the haystack of clauses — completely goes against the GDPR’s data minimisation requirement. Every GDPR-expert will refer to countless instances where authorities have taken a firm stance against the idea of “let’s collect everything and hope there’s something useful in there”, which is exactly what Henchman’s data maximisation and “needle in the haystack” approach is all about.
Similarly, it is a standard requirement in most non-disclosure agreements and deontological rules to strictly limit access to confidential data, typically expressed as disclosure on a “need-to know basis”. Tools such as Henchman for which ease of use and maximised search results are essential undermine this requirement by presenting random snippets of text to users searching for interesting clauses.
Customers often think that, from a compliance perspective, using products such as Henchman is similar to using a case management system, which stores gigabytes of similar confidential data. The difference is that exhaustively keeping track of client files in a case management system is necessary to get work done, and often even a legal requirement in its own right (e.g., for law firms). Conversely, good standalone search tools are nice-to-haves, but not a necessity and therefore do not benefit from the lenience that the GDPR affords to necessary tools. Also, as any GDPR expert will tell you, the intention of the user makes a huge difference when assessing whether a certain kind of data processing is lawful. At this point, Henchman’s data maximisation goal and promise that lawyers don’t have to spend any effort will of course completely backfire, once data protection authorities will investigate this market.
Henchman is aware of this problem, offering an automatic search-and-replace option to remove customers names. More recently, it also offers the possibility to involve GPT to anonymise a specific clause. But both features are stopgap measures, as they only remove some confidential data. It is also questionable how popular these features for users who are attracted to the completely automatic nature of the product.
Henchman's founders express in interviews that — compared to the US — their European target audience is significantly lagging behind in terms of purchasing, due to these data concerns. It should therefore not come as a surprise why Henchman so strongly emphasizes security on its website, and only briefly mentions GDPR as part of its security page.
In all fairness, we also heard stories from compliance-driven law firms who use Henchman, but deliberately limit the data to clean templates and manually cleaned documents. Of course compliance is a non-issue then, but one can wonder whether Henchman is then still a good fit, as you work against the product’s philosophy and pay a premium price for using an automatic tool in manual mode. The popularity metric will obviously not work in such setup (as only a fraction of your data is uploaded), so you'll have to rely on suboptimal automatic filtering and manual tagging to bring some structure in your clauses.
Comparison with other approaches
To better understand the advantages and drawbacks of Henchman’s completely automatic approach, it is interesting to compare it to the approach taken by the “Truffle Hunt” module of ClauseBuddy.
Similar to Henchman, Truffle Hunt also copies your original documents into a separate server environment, crawls those documents, automatically splits them into clauses and makes them available in an easy interface. The difference with Henchman is that Truffle Hunt invites lawyers to make a subselection of their documents, instead of copying their entire stack of documents. Lawyers still have the possibility to upload thousands of documents, so nothing is stopping them from copying their entire inventory, but they are actively discouraged to do so.
This difference may seem small at the surface, but exposes a fundamental conceptual difference between both approaches. Data maximisation products such as Henchman initially copy as much data as possible, requiring no effort during the upload phase, but instead shift the effort to the search phase, as the user must spend more time reading and filtering. Knowledge-oriented tools such as Truffle Hunt instead invite you to spend pre-filtering efforts during the upload phase, but require significantly less effort during the search phase.
Compared to Henchman, Truffle Hunt has the downside that lawyers have to spend some effort during the upload phase: while selecting and uploading files is fast and easy, it is not fully automatic. Henchman will also be clearly better than Truffle Hunt at “needle in the haystack” searches, as it may be the case that you did not upload that one old Word-file from a deal of 8 years ago in which that one particular clause happens to be present that you’re now desperately searching for.
However, Truffle Hunt’s approach also has immediate advantages. First, it is not necessary to involve your IT department, because the uploads can be done by the lawyers themselves; data security concerns are minimal because of the upfront selection; and compliance is much more manageable. Second, the amount of "noise” in the search results will be much lower, as the number of documents is lower and you will not have to wade through hundreds or thousands of clauses extracted from irrelevant documents (such as documents received from clients and counterparties that just happened to be present in your document management system).
Third, ClauseBuddy encourages lawyers to primarily treat Truffle Hunt as an intermediate step, by offering tools to copy the truly interesting clauses from Truffle Hunt into a curated clause library. Such clause libraries contain a team’s best clauses and are typically enriched with various kinds of metadata, so that searches such as “give me a short copyright assignment clause that protects the author as much as possible” can be performed in seconds as they do not require you to read hundreds of results or use filters that rely on questionable metrics.
Henchman offers an automatic clause extraction tool that guarantees to save time compared to the traditional “Word + Outlook + document management system” setup. The price is steep, but for legal teams who spend significant time drafting contracts the return on investment should be clear.
The tool has several sweet-spots where its automatic data maximisation approach really shines. Chief among them are small teams of senior lawyers (typically corporate or commercial law) who are aware of each other’s drafting history, can quickly judge a clause on its legal merits, and frequently require “needle in the haystack” searches in their collective drafting history.
Henchman is also a very good match for legal teams who want to spend as little effort as possible on the setup of their drafting tool, e.g. because their primary reason for acquiring it is to mention some legaltech tool in client pitches.
Third, several users see to really like using Henchman as a replacement for the poor search tools of their outdated document management system.
Conversely, Henchman is not a good fit for legal teams for whom compliance is a concern, due to the data copying and data maximisation issues.
It is also not a good fit for legal teams who want to leap ahead in terms of knowledge deployment, e.g. to increase quality and offer better guidance to junior lawyers. After all, the product focuses on maintaining existing contract drafting processes, optimizing them without addressing the underlying chaos that may exist. Several legal teams therefore expressed their reluctance to give junior lawyers access to Henchman, fearing that the wrong clauses will get selected, due to the tool's abundant search results, over-reliance on a clause's popularity, and lack of context/guidance for individual clauses.
Finally, due to its “micro-tool” approach, Henchman will be only a partial fit for legal teams who are looking to seriously advance their drafting workflow. Contract clause searching is crucial of course, but efficient legal drafting is about so much more, e.g. automation of full documents, document reviewing, proper styling, proofreading, definition list checks, playbooks, internal drafting notes, self-service platforms for business users and clients, and so on. Also note that Henchman is completely focused on contract clauses, ignoring others types of legal drafting, such as memos and legal briefs. Accordingly, while Henchman's completely automatic approach will have a strong initial appeal, legal teams will quickly want more options, more features and more control than automatic clause extraction tools can offer.