What started as a side project born from personal pain has become a venture-backed, fast-growing business that’s helping their customers manage self-employment tax as they would salaried job.
James didn’t set out to build a global fintech—he just wanted to help people facing the same headaches as he was. With competition now hot on their heels, they Hnry team knows the only way to win is to keep delighting their customers. Watch, read or listen to find out what that means in reality 👉
Skim the highlights from my chat with James:
“You're always going to have competitors, right? You're never going to have a market to yourself. But the only thing that you can rely on is two things, which is serving your customers better than anyone else and innovating faster than your competitors. [Our competitors] are all struggling with the regulatory side of things of how do they build legitimacy or how can they actually work with regulators to make sure that they're aligned with that. We're at the other end going, all right, how do we, at scale, deploy features to our customers that will really surprise and delight them and not charge them any more money for it?”
“I do look at the copy cats that are coming out and I think they're going to have a rude awakening. First financial year end where they're actually on the hook for filing people's taxes and they're actually going to get people coming back to them going, you did this wrong. It does not add up. Someone else has fact checked you and it turns out you are incorrect or I've suddenly found out that you've pushed all of my financial information into chat GPT and so now all of my personal finances have become a training for a learning model, I don't like that.”
“When you're building an enduring business, as we've done that gradually over years, we're not trying to shortcut that process by copying and just going, we'll just follow the same playbook. Because the bit that you can see of the playbook is the public facing bit. And the bit that's hidden is all the hard work and effort that has come to this stage to actually build a system that is correct and that works and that actually customers really, really like. And you can't copy that. You can't copy the trust that customers give you from having used the service for years. And you can't copy you know, and instantly scale a business, you know, in the same way as you can, like, build a sustainable business over time.”
A16z published a think piece last month, drawing attention to the rise of search recommendations from LLMs. We see it as an evolution that complements existing SEO tactics. However, there’s a knack to optimising for language models, not just page rank or keywords. Dual optimisation is required - 1) for the end user, and 2) for the LLM model itself. Each model has its idiosyncrasies but all of them are capable of encoding a “perception” of your brand from virtually all datasources - social sentiment, press, owned content, third party reviews.
The way people search is changing. Queries are becoming longer (averaging 23 words, up from 4), sessions are deeper (averaging 6 minutes), and responses are personalised, multi-source syntheses that vary by context and source. Unlike traditional search, LLMs can remember, reason, and respond. For marketers, this means traditional SEO's emphasis on precision and repetition is less effective. Generative engines prioritise content that is well-organised, easy to parse, and rich in meaning, rather than just keywords. Formatting tips like using "in summary" phrases or bullet points can help LLMs extract and reproduce your content effectively.
It's no longer just about getting clicks; it's about "reference rates" – how often your brand or content is cited or used as a source in model-generated answers. GEO requires optimising for what the model chooses to reference, fundamentally changing how brand visibility and performance are defined and measured. ChatGPT is already driving referral traffic to tens of thousands of domains, indicating the value of outbound clicks from LLM interfaces.
“In a world where AI is the front door to commerce and discovery, the question for marketers is: Will the model remember you?”
Today Halter announced it’s $100m Series D round, valuing the company at $100B. Founded in 2016, Halter now has thousands of customers across New Zealand, Australia and the US, with new cattle ranches and dairy farms going live daily. Halter employs over 200 people, is headquartered in Auckland (New Zealand), plus a US office in Boulder, Colorado.
The round was led by BOND, a global technology investment firm, with new investment from NewView and continued support from early investors Bessemer Venture Partners, DCVC, Blackbird, Icehouse Ventures and Promus Ventures.
The capital will boost their expansion into the US “rancher” market, that’s facing issues around labour shortages and scrutiny on sustainability.
Read, watch and grow in under 5 minutes a fortnight.
A newsletter designed by Atlas Digital.
