How vibe coding is creating a new 'rat race' in law
An ongoing short piece on vibe-coding, the 'legal builder' clout and the future of talent politics in the legal market.
As a programmer, I love vibe coding. Over the summer break, I was able to smash out my to-do list of new features and workflows to my Global AI Regulation Tracker (more on that in a separate post!). I tripled my output, and finished a month’s job in a few days.
So this post is not a critique of vibe coding. This is more about the potential impact of vibe coding on talent politics - i.e. exploring the dynamics of how employers compete for talent and the ‘zeitgeist’ behind how employees approach their career choices and identity. It’s a different angle compared to the typical question of looking at how vibe-coding affects the future of code, SaaS, etc.
To connect vibe-coding and talent politics is quite an ambitious and niche exercise. I admittedly don’t have an HR background or qualification. So I’ll stick to my lane as close as possible and only focus on the legal market based on my own personal observations being within this market.
This will be a live bullet-point piece which I’ll continuously update as I further refine my thoughts. Perhaps ‘talent politics’ and ‘rat race’ are not the right words, so feel free to tell me what’s a better term once you’ve reviewed my thoughts!
1️⃣ Vibe-coded apps become 'vanity projects’ for building resumes…but only in the short term. The recent flurry of vibe-coded apps is very impressive and inspiring. I expect this trend will continue strongly but eventually run into an over-saturation problem as more and more lawyers (from graduates to seniors) realise this is new table stakes to stay relevant and establish a professional brand. You’ll notice this (perhaps in 6 - 12 months’ time) when you start seeing more resume-padding ‘vanity projects’ (e.g. a glut of video demos only but no usable prototype / repo provided) and the copycatting of ‘how to vibe-code’ playbooks and courses, rather than actual usable solutions. Once the ‘low hanging fruit’ use cases (e.g. doc review, research etc) oversaturate with generic solutions, the bar is raised and the ‘differentiation game’ moves into practice area specific (e.g. real estate v. finance v. employment use cases, etc), regional-specific (West v East), and organisational-specific solutions (law firm v in-house solutions). The cycles will continue for a long time, but I think the ‘vanity’ appeal will dilute as each cycle normalises vibe-coding into “just another skill on the resume” (rather than a unique personal feat). Hence why I think the vanity side of vibe-coding will be relatively short-lived. But it’s useful to track these cycles as they help you predict the zeitgeist of legal builder talent (see point below).
2️⃣ An inflation of the ‘legal builder’. The idea of legal builder (e.g. legal engineers, coding lawyers, etc) is still quite rare and refreshing in the market, but that appeal is slowly diluting as vibe-coding drops the technical barrier and new hordes of legal talent (especially post Gen Z) pivot towards that angle. Universities, law firms and other legal organisations will start to promote their legal builder cohort as the face on enrolment/recruitment materials, reinforcing some kind of ‘star status’ around them (which attracts more people). There’s indeed a ‘best of both worlds’ appeal to all of this, as the legal industry finally has an outlet for STEM-minded talent that mixes the thrill of Big Tech / Silicon Valley culture with the stability and clout of the law. It’s why every year I hear more and more legal builder aspirations among law students and graduates. It’ll get to a point where being a legal builder no longer holds as much clout in the market compared to today, and they will start to rethink their niche/brand direction in 3-5 years time. That leads to my next point.
3️⃣ How to stand out? Everyone can now build, but not everyone can scale. I think that’s the insight that the ‘pros’ are grasping right now. As the ‘amateurs’ push out open-source prototypes and no-code agentic workflows, the ‘pros’ are getting their hands dirty with cloud infrastructure, serverless functions, cybersecurity, payment gateways, distribution channels, etc to bring their creations into a scalable product. I know from experience these things are less ‘vibe-codable’, and require deep systemic thinking rather than writing code (i.e. you have to really understand the tech stack and how data flows throughout the stack, which is not so easy to capture in a written context for Q&A with an AI chatbot). This is one area where the ‘pros’ can stand out and remain relevant. I wouldn’t be surprised if there are already people thinking two steps ahead, and quietly building a marketplace platform that tracks, catalogs and hosts vibe-coded solutions (i.e. the whole “be the pond, not the fish” analogy).
4️⃣ As product dilutes, thought leadership becomes more competitive. Flowing from points 1-3, the legal builder talent pool may start to feel burnt out from the ‘rat race’ of building and scaling the best products just to stay relevant. So the relevance game starts to ironically shift towards more traditional frontier, being academic thought leadership. The idea of a “legal builder & thinker” is incredibly powerful - i.e. someone who can deliver both on the theoretical and practical fronts. Simply put, thought leadership adds gravitas and credibility to your voice, and helps you sell your product (if any). There could be an interesting pattern where the legal builders start to write on (and ‘compete’ for ownership of) academic topics with governance professionals, academics, researchers, legislative writers (and vice versa, with the latter group also building their own products to compete with the legal builders). At the same time, you may also see existing private practice lawyers hone down on their advisory expertise, and really make their knowledge as non-productisable and indispensible as possible that it cannot be touched by vibe-coded solutions. This is healthy for a producing a more vibrant legal talent pool, but will also likely generate some ‘politics’ (or even toxic sub-cultures) as participants feel threatened by others “crossing in their territory”, “stepping on their toes”, “eating their lunches”, “wannabe gurus”, etc!
5️⃣ The ‘employer certainty’ and ‘employee loyalty’ equation is shifting. Legal practitioner careers broadly diverge across three destinations: private practice, in-house or legal tech. While this is not new, vibe-coding (plus the FOMO feeling from seeing massive fundraising rounds, founder glory and employee-share schemes at legal tech companies) has really boosted the appeal of the legal tech path. Because of so much optionality, the legal talent churn will start to mirror the tech talent churn where people jump roles every 2-5 years. This creates uncertainty for employers but also creates a more diversified talent ecosystem to draw upon. Traditional “restraint of trade” and “employee IP is employer IP” clauses in employment contracts will start to be seen as red flags rather than boilerplate language for the super ambitious and creative builder talent who have entrepreneurial ambitions. Whichever employer learns to let go of 100% certainty might ironically end up with the best and most loyal talent.
Views my own.
Want more?
Global AI Regulation Tracker (an interactive world map that tracks AI regulations around the world).
Note2Map (a platform to build and launch your own interactive world map tracker).
Agree, Ray, especially on your point about scaling. We continue to dedicate real effort on this.
Separately, I’m concerned about lawyers putting client data into vibe-coded apps. Most lawyers have no visibility into whether the libraries being used are secure or potentially compromised. When building an app this way, an LLM could easily pull in a dependency from a non trustworthy source without anyone realising it.
Proper governance and controls are essential when building these kinds of products. I’m planning to write more about this separately.
Hope you’re well.
Really interesting takes. I hope that when the AI hype slows down, there will be a gradual convergence on the solutions that actually work and create value and less therefore less focus on largely speculative use cases that do not bear any real fruit.