The Future of Drafting

The year is 203X, the senior partner jumps off her hoverboard through the open office window of the client. En route, she had fired various instructions into her virtual assistant who had assimilated them to prepare and proofread the perfect contract, requested mere minutes ago. It mattered not that the firm handled these kinds of transactions infrequently, the assistant represented the sum of all human knowledge, and slavishly refined and corrected its outputs with the power of a thousand scribes. Confidently the output is beamed over to her waiting contact, all before settling in for a coffee as key talking points flash up on her smart glasses curated from a mix real-time market scanning and salient nuances of her client’s current deal.

Some slight embellishment aside, the drafting dream feels closer today than ever – deeply trustworthy, instantly accessible and effortless outputs seem like a future heralded by generative AI. So what is stopping this, and why is this so enticing in the current tech landscape?

Document Automation is hardly new, and it is storied. For every great success story there are hushed whispers of abandoned projects. The simple truth is, that while a well delivered and maintained project gets very close to the dream (solutions are now using AI to populate questionnaires from emails and documents), getting there requires a lot. You need the right people to create and maintain it, the chosen solution needs to balance your drafting complexity against ease of set up, and everyone needs to agree to use it in order to get a decent return on investment.

Similarly AI has much promise in the drafting world, but is not there yet for the initial stages. People are trying of course, and they should continue to do so. For simpler documents there are examples out there, but an AI generated NDA offers little to the bottom line nor the innovative vision. More complex documents, well, the technology today produces highly varying results. Much is still needed to get a passable output, and between the ever-present potential for hallucinations and the underlying lack of true understanding, the outputs arguably require more review than more traditional methods.

Some firms felt that while they were getting close, they’d created a new beast in the complexity of the prompts being close to the length of the contracts. Others felt that they were spending more time reviewing outputs, and using more costly labour to do so. The tech is moving fast too, which offers improvements, but that presents its own challenges, moving to a new or updated model may in and of itself present costs and rework.

So what should you do today, wait on a dream of unclear timescales, or embrace what may soon be outmoded? It is always best to go back to fundamentals and ask yourself how long will taking on today’s tech take to return on investment. If you cannot see a project paying for itself within the space of a few years, then it probably wouldn’t have been worthwhile absent of future tech developments anyway. Consider your risk profile, do we need to move faster, or do we want to ensure solid foundations. If you’ve attempted document automation before and it failed, have an honest reflection on why, was it the technology, or were there other barriers? AI will not miraculously solve change management issues.

AI is definitely making its mark on drafting, and no doubt in the future it will do so with greater depth and in more places. The ability to generate automated reviews of other party paper against your gold standard for negotiation seems magical compared to ten-year-old tech. The capability for flexible data extraction at scale is running rings around machine learning based solutions of yesteryear. As with any tool, it will excel at certain jobs, and fit others less.

The mythic future for drafting described in my opener isn’t here yet but today’s AI solutions can offer great gains in post initial draft elements – summarising content, suggesting alternate language and comparison against playbooks. And well-implemented document automation conversely still represents the gold standard in initial drafting for consistency, reliability and overall speed. Sadly there are no major developments on the hoverboard front.

William Sumners

Chief Operating Officer

echo.legal