Prototype Development
HLB & Lovable: Prototype‑Driven AI Innovation
Vibe coding is changing how professional services firms build, buy and deliver technology. As AI lowers the barriers to software development, firms can create bespoke tools faster and more economically, but the shift also raises important questions around skills, pricing, governance and where professional development remains essential.
In the last few years, vibe coding has transformed how businesses deploy technology. Unlike low-code or no-code tools, which usually give a visual interface with predefined building blocks, vibe coding is more conversational. It enables organisations and individuals who aren’t skilled developers to build software by describing what’s required in natural language and letting AI generate most of the code before refining it through prompts.
One of the platforms driving this shift is Lovable. A software creation platform, Lovable allows users to describe what they want to build in plain English, and the platform generates the underlying interface, database, logic and code, allowing even non-technical users to create and refine working software through conversation rather than traditional programming.
The scale of Lovable’s growth is a sign of the vibe coding shift from a novelty approach to a serious consideration for firms. To date, 70 million projects have been built on the Lovable platform, and employees at nearly two-thirds of the Fortune 500 build with the tool. Enterprise customers include Adidas, Deutsche Telekom, Nvidia, Workday, and Zendesk. In February 2025, Lovable became a tech unicorn and was recently valued at $13.3bn, with 900M monthly visitors.
Alexander Kollmann, EMEA Hackathon Lead, Lovable
Alexander Kollmann, EMEA Hackathon Lead, Lovable: "In under two years, Lovable has grown from an idea into a platform used by millions of people around the world to bring their ideas to life. What's changed more recently is what happens after something gets built - for a growing share of our users, Lovable isn't just where they build, it's where they run the business or product that comes out of it. That's the shift we're leaning into: giving the person closest to a problem, whether a first-time founder or someone inside a large company, the ability to both build the solution and operate it."
The build-vs-buy reframe
Historically, firms bought software because creating even a relatively simple bespoke tool required developers, significant upfront investment and ongoing technical support. But AI coding tools, and platforms like Lovable, have lowered that barrier considerably. Software spend is now being assessed not just against competing vendors, but against the cost of creating a highly specific internal tool. That puts pressure on expensive solutions and software with lots of functionality a firm does not use.
Professional services firms are now starting to see the benefits of building their own tools. HLB CBS Group in South Africa built ‘Clarity’ on Lovable, an internal AI platform with usage-based pricing that avoids per-seat licenses. It gives staff a secure alternative to unapproved AI tools, along with agents built around the firm's own audit methodology. These include a working paper builder, a financial statements reviewer and a research tool that answers only from the accounting standards loaded into it.
An approach like this has a number of benefits beyond cost-savings. For example, it reduces the risk of shadow AI use, whereby staff use their own AI tools, which the firm has no oversight of, and the firm retains control over what information is stored and for how long. It also means that the firm decides when the platform needs updating according to feedback, rather than being reliant on a third party.
The impact on skills
This approach to technology has a significant impact on the skills that firms need to hire for and develop. The skills to understand the business problem, map the workflow, identify users and requirements, and decide what should actually be built, will be more important than the expertise to do the building itself. However, understanding what information can be put into an AI tool, where data is stored, access permissions, authentication, privacy requirements and the risks of third-party integrations will still be required.
As build times fall, vibe coding will likely change how accountancy and advisory firms scope and deliver client work, enabling faster turnaround on bespoke tools, more customised deliverables and less dependency on third-party software timelines. That could make engagements more iterative and practical, allowing clients to test and refine solutions earlier and helping advisers move more quickly from identifying a problem to implementing a workable solution.
Of course, when the cost of building falls, firms have to become much clearer about what clients are actually paying for. Increasingly, it won't be the hours required to produce something; it will be the expertise to identify the right problem, create the right solution and make sure it delivers a useful business outcome. Over time, this could broaden what counts as a professional-services deliverable, with clients receiving not just advice, reports or spreadsheets, but lightweight software tailored to a specific business problem.
Alexander Kollmann, EMEA Hackathon, Lovable commented: "The 'vibe coding' label captures the first moment - describing an idea and watching it come to life - but we're already past that. The real shift isn't a new way to code, it's a new answer to who's allowed to build at all. Advisors and accountants who understand a client's problem better than any developer can now build and run the tools themselves, without waiting on a technical backlog. That doesn't shrink the role of technical teams, it elevates it, from writing every line of code to owning the architecture, judgment, and quality control behind what gets built."
Limits and risks
Nevertheless, the rise of vibe coding highlights the need for firms to have a robust technology strategy with clear guardrails in place.
Firms need to be aware that AI-generated code can contain vulnerabilities; if staff don’t fully understand the underlying codebase, fixing bugs, applying security patches or handing the system over to another developer can become difficult. Although the speed of vibe coding can be a major advantage, it can also encourage organisations to move from prototype to production without the testing, documentation and ongoing maintenance that conventional software development would normally require.
It’s important for firms to assess the risks associated with the tool(s) in question and to help mitigate these risks with policies that ensure the appropriate development and deployment of AI. These frameworks will also reassure clients that the use of AI platforms is properly managed.
HLB’s partnership with Lovable is helping HLB’s member firms to broach these questions and support their innovation. HLB’s recent Technology & Innovation Symposium brought together colleagues from over 17 countries within its global network to focus on developing practical applications of AI. Within the event, Lovable hosted a hackathon with teams challenged to turn real-world professional services problems into working prototypes. The hackathon gave participants the opportunity to experiment directly with the technology to innovate using AI even though they’re not technologists by trade, and was a live example of the network engaging with this shift deliberately and getting hands-on with cutting-edge technology.
Vibe coding is unlikely to replace professional software development, but it is changing who can build, how quickly ideas can be tested and where firms choose to spend their technology budgets. For accountancy and advisory firms, the opportunity lies in using that flexibility to solve problems faster and deliver more tailored services, while retaining the governance, security and technical oversight needed to use it responsibly. The firms that benefit most will be those that treat vibe coding not simply as a shortcut to cheaper software, but as a new capability that sits alongside professional judgement, domain expertise and robust technology controls.
Main video credit: nano-stocker/Shutterstock.com