First year lessons from a solo expertise biz in Data & AI

In a world of AI-generated slop, it’s refreshing to read real human stories.

To contribute my bit to humanity and help me write consistently, I decided to start sharing a monthly reflection. This is in line with old-school blogging, which should help people who find themselves in similar situations to mine.

The focus will be on my career in Data & AI, but it’s not completely detachable from the rest of my life.

To set the scene for future reflections, I’ll start with where I’m at now, and what I’m trying to do. I’ll then summarise some highlights and lessons from the past year, and conclude with the past month.

Now building: Solo expertise business

It took me an embarrassingly long time to understand this, but broadly speaking, I’m building a solo expertise business.

Let’s break this up:

• Solo: I don’t want employees. I may rely on contractors and other outsourced help, but the goal is to grow the business without growing headcount. Despite getting much less attention than VC-funded startups, solo businesses are incredibly common and can be highly profitable. For example, Justin Welsh is on the path to $10M in total revenue with a 90% profit margin (after five years in business), and Jonathan Stark got to a point where he’s making around $400k annually while working about five hours per week.
• Expertise: My accumulated experience and skills are the core source of value. In my case, my experience can be broadly described as being in the areas of software engineering, data science, artificial intelligence, and tech leadership (Data & AI for short). However, all of these are commodities, and I’m aiming to position and market my expertise as unique. This is a key insight – expertise helps generate value, but value capture is a completely different story. Value capture is done via different offerings to specific audiences (aka products), including my consulting services that are targeted at climate & nature tech startups.
• Business: This is a key mindset switch. A year ago, I was thinking of it as doing some freelancing on the side while building a software product. Building an expertise business shifts the focus from the employee mindset of “Yanir is doing some work and bills hourly” to thinking of multiple revenue streams and attempting to maximise value creation and capture while minimising costs (especially the most limited cost – my time).

With this in mind, let’s get into an overview of how I got to this point.

First, ChatGPT’s summary

⚠️ The post ended up being way longer than I wanted it to be. While I don’t advocate for blindly publishing AI-generated slop, AI is amazing for summarisation. Therefore, I asked ChatGPT to summarise the rest of the post. It’s not exactly right, and it lacks personality. However, it does capture the key ideas in case you’re busy or uninterested in going deep into the whole story. Now that I got it out there, I will aim to make future reflections shorter and avoid AI-generated summaries.

1. Leaving Orkestra and Exploring Options (Q2 2023): After leaving Orkestra, the author explored various projects, including revamping a marine biodiversity tool and applying to AI labs. Key lessons included the importance of trust and the realization that foundational AI research was not a preferred path.

2. Shifting Focus and Embracing Consulting (Q3 2023): The author moved away from the idea of building a solo software product to focusing more on consulting, especially in the climate tech space. This period emphasized the value of finding a niche, clear communication of value propositions, and the blurring lines between consulting and product building.

3. Navigating Distractions and Positioning Clearly (Q4 2023): Faced with personal and global challenges, the author learned to manage distractions, position themselves clearly in engagements, and recognize that their identity and ethnicity are inescapable aspects of their work. This period also underscored the limited intersection between true detractors and the target client base.

4. Building Authority and Trust through Outreach (Q1 2024): The start of the new year saw the author becoming more systematic in outreach and lead generation. They committed to consistent publishing, leveraging parasocial relationships to build trust at scale, and reactivating their existing network, which led to increased confidence and new opportunities.

5. Commitment to a Solo Expertise Business (Q2 2024): In this period, the author fully embraced the concept of a solo expertise business with multiple revenue streams. They focused on refining their LinkedIn profile, joining relevant communities, and understanding the importance of long-term relationship building over immediate gains. This commitment marked the true start of their business journey.

These points provide a comprehensive overview of the author’s journey from leaving a startup to fully committing to and refining their solo expertise business in Data & AI.

Q2 2023: Leaving Orkestra and messing around

In April 2023, I left Orkestra – a climate tech startup I joined early 2022. It was a classic first Data-to-AI hire story: I worked on an AI/ML project that they ended up pivoting away from (as startups do). While I could have stayed around to help with engineering on their main product, I decided it was time to leave and figure things out on my own. I still like the Orkestra crew and hope they do well, though.

As financial years in Australia run from July to June, I was in no rush to generate more Q2 revenue due to the high tax rate. Instead, I nominally took time off, but actually ended up doing a few different work-y things in addition to a couple of diving adventures:

1. Wrapped up a paid project for revamping Reef Life Survey’s Reef Species of the World.
2. After getting swept up with generative AI excitement and a bit of an existential crisis, I decided to try applying for a job with a top AI lab. To my surprise, they sent me their automated code assessment, which led to me spending more time reflecting on why I didn’t want to proceed with the process than the time I spent preparing for the assessment.
3. Spent some time exploring software product ideas I could build in the energy space, which led me to a year-long volunteering stint with Work on Climate.
4. Reflected on the shifting landscape of data science in an attempt to come to terms with the fact that I always needed to do plenty of engineering to be effective in data science roles. The latter reflection turned into a talk abstract, which led to a conference talk in Q3.

OK, so I’m not great at truly taking time off. But figuring things out was a key goal of Q2 2023.

What did I learn from Q2 2023?

1. I wasn’t ready for this lesson yet, but the Reef Life Survey project was an example of the importance of trust and value pricing. I was basically given a budget to improve the web tools because they trust me after almost a decade of volunteering. Scaling up this trust to enough clients will be key to a sustainable consulting practice.
2. I’m not sufficiently interested in foundational AI research – I prefer applying AI to specific problems. I also wasn’t ready for relocating from Australia, even temporarily for an amazing AI lab job.
3. I was getting close to realising that I don’t have a software startup idea that I’d commit to, but I wasn’t quite there yet.
4. I came to terms with data science drifting away from its engineering roots, and realised I needed to reposition in a way that’d emphasise my engineering skills.

Q3 2023: Shifting away from freelancer and indie hacker mindsets

Going into July 2023, I was aiming to follow a similar path to my 2014-2015 direction: Do a bit of freelancing to generate runway for building a solo software-as-a-service product business (aka an indie hacker product, partly popularised by Start Small, Stay Small). One thing I learned from 2014-2015 was that it’s hard to sustain the effort needed to build a successful product if I don’t care about the problem it solves, so I focused on exploring ideas in the climate space.

As no idea felt compelling, I ended up leaning more towards consulting – especially once I realised that the lines between solo consulting and product building are blurry. Putting more effort into consulting in the climate tech space was meant to serve multiple goals: (1) generate revenue; (2) get exposure to problems that may be addressed by an indie solo product; and (3) improve my marketing and sales skills.

In parallel, I kept an eye on job ads but didn’t hear back about roles I found interesting. Inbound interest in my work was also rather low, partly as a reflection of the job market (not the best timing for me). I was also picky because I didn’t want a full-time job I didn’t care for. However, I was still doing some paid work for Reef Life Survey, and had an inbound lead from a local agency regarding an interesting AI advisory gig. This was more than enough to break even financially, even after going to Bali for a two-week dive trip.

I also pretty much came to terms with being at the intersection of data science and engineering by giving the talk mentioned above: Lessons from Reluctant Data Engineering. My positioning was still pretty commodified, though – I still labelled myself a “full-stack data scientist”.

What did I learn from Q3 2023?

1. Finding a niche that you can serve well is key to building either a solo software product or a solo consultancy.
2. Likewise, communicating the value proposition clearly is needed for both products and services.
3. I started moving away from the framing I had of software products as the only “real” sustainable approach, and towards Jonathan Stark’s business strategy: “Help people you like get what they want.”

Q4 2023: War and hate are powerful distractions

October 2023 started pleasantly enough, but quickly turned into one of the worst times of my life. I’ve been living in Australia since 2009, but I am a Jew from Israel and still have family and friends there. Therefore, I was deeply affected by October 7th and its aftermath.

This had two main elements:

1. I over-consumed news initially. Apart from the immediate worry and uncertainty, this led to secondary trauma (which I only learned about around that time).
2. I was shocked and appalled by the response of politicians and individuals who claim to support tolerance, peace, and non-violence. Locally, this has included politicians from The Australian Greens whom I had previously voted for. Examples abound, but to me the response is epitomised by the slogan “from the river to the sea…” – a call for ethnic cleansing by the same people who generally bend over backwards to avoid hurting the feelings of non-Jewish minorities.

Anyway, it goes without saying that being so shocked and distracted wasn’t good for business. But it pales in comparison to what people in the war zone are going through.

The war has also revealed that I couldn’t work with some people in the climate space due to the tribalism and hate by certain activists. For example, Greta Thunberg was filmed chanting “crush Zionism” – a call that I interpret as advocating for the death or displacement of the seven million Jews that live in Israel. However, by the end of the year I realised that: (1) most activists aren’t building climate tech businesses; and (2) most people understand they don’t know enough about Middle Eastern geopolitics to have strong opinions either way. Therefore, the intersection between true haters and my potential client base is minimal.

Concurrently, the AI advisory work with the agency didn’t turn out well. We had different ideas on the direction of the product, I didn’t want to increase my commitment and help them build it, and so we parted ways.

With whatever energy I could summon, I managed to start positioning myself as a fractional chief Data & AI officer, and began reaching out to prospects more systematically. However, things were slowing down with the end of the year, so nothing came of it at that point.

Looking at my notes and messages, at that point I still didn’t fully let go of going down the software product path. This wasn’t particularly fruitful, as time was in short supply, and I was also exploring a startup idea with a former colleague. You really can’t do it all.

What did I learn from Q4 2023?

1. I can’t escape core elements of my identity and ethnicity. This applies in any work scenario, but is especially salient in a solo business.
2. Advancements in technology and education still haven’t countered human tribalism, including among the well-educated.
3. I need to manage my exposure to news and other distractions. Indeed, by the end of 2023, I was down to about one news check a week – enough to remain informed without getting too sucked in. I also strictly reduced my exposure to algorithmic feeds (especially LinkedIn – I’m not on Twitter, TikTok, or Instagram, and Facebook is boring).
4. When negotiating engagements, I need to position myself clearly. With the agency, they hoped I’d be “a pair of hands” rather than an advisor (despite how it was pitched to me). This led to disappointments on both sides.
5. Most people in Australia don’t care about Middle Eastern geopolitics. While some local politicians and news publishers use it to fire up their base, it has little to do with what’s happening here. This is hard to fathom if you’re over-exposed to the news and algorithmic outrage machines.

Q1 2024: Time off, some work, and the business of authority

Coming into the new year, it was time to move on from the mess that was Q4 – at least to the extent that the political reality would allow me. Indeed, the war has led to one added time constraint, as I went on a family trip to the US for a couple of weeks in January. I had also committed to a Reef Life Survey dive trip in March, which meant I took four weeks off in Q1. This wasn’t ideal given the stage I was at with the business, but not terrible either, given that one of my goals as a soloist is time flexibility.

Early in the year, I still wasn’t fully committed to the business. I had even considered going full-time for the right role and applied to a few jobs. However, I didn’t proceed when I heard back, which meant that it was overall mostly wasted time.

On the non-wasted-time front, I started becoming more systematic with my outreach and lead generation. Back in Q4, I started with cold rather than warm outreach, with lacklustre results. In Q1, I got more into reactivating my existing network, and experimented with building an outreach system inspired by Atomic Habits. This worked much better. Among other things, it led to project work, fresh LinkedIn recommendations, increased confidence, and various leads. It was also kinda fun, though I’m still figuring out how to best do it efficiently.

When it comes to efficiency, one thing I resisted up until Q1 was publishing more. However, following Jonathan Stark and equipped with lessons learned from The Business of Authority podcast (co-hosted with Rochelle Moulton), I realised that publishing is the path to parasocial relationships. In short, it’s about building trust at scale with people who know much more about me than I know about them. To this end, I committed to posting here weekly when I’m not on holiday – what you’re reading now is a direct result of that! I also started getting more active on LinkedIn, albeit somewhat sporadically.

What did I learn from Q1 2024?

1. Letting fear get in the way of commitment to one path is silly. The job applications mostly served to allay my concern that I’m becoming unemployable, but they were a time sink. They also meant I wasn’t positioning well – to get a job you need to fit into a commodity box, which is the opposite of the positioning needed for a high-value expert.
2. Building trust and relationships is invaluable. One of the projects I did was with Orkestra – my former employer. It took 15 minutes to seal the deal with a fixed price. This is another instance of the Q2 2023 lesson from working with Reef Life Survey – scaling up to multiple clients with whom I have mutual trust is the key to high-value engagements.
3. Nurturing all your relationships can be helpful to your business. Everyone is busy, and this should go beyond transactional activities, but Consulting Success mentions network reactivation as a key strategy for a reason. It actually works, as the people who care about you want to see you succeed – but they can’t help you if you don’t keep in touch and say what you’re up to. Shocking, I know.
4. Publishing consistently is key to trust at scale. In the words of a recent post by Stark, his marketing strategy is: “help people you like get what they want for free at scale”. Yes, it’s kinda cringe, and I don’t absolutely love everything I post, but I see it working on me and for me. And again, this post is a part of it – it may help people I like get to their solo biz or startup goals faster than me.
5. Prioritising time off is good for the soul. In my view, there’s little point in going down this business path if it doesn’t eventually lead to work that is more flexible, impactful, profitable, and satisfying than what I could achieve as an employee. This is also why I chose this path over cofounding a high-growth startup – I don’t want to overwork for at least 5-10 years in the hope of a future reward.

Q2 2024: The magic of commitment and focus

On return from my Reef Life Survey trip in April 2024, I had a choice to make: I could interview for a well-paying role with a big tech company (following interest from a recruiter), or I could keep going with the biz. Big tech interviews are a bit of a pain – I know I can pass them, but I also know they require a significant time investment to pass. The problem was – once again – the potential for distraction.

Following some solid leads in early April, I decided to stick with the biz. Unfortunately, many of those leads ended up fizzling or going super-slowly through Q2. I’m targeting startups with my consulting to work closely with quick decision makers, but sometimes the best decision for them is to focus on other areas than on using my expertise. Oh well.

Another major shift in my thinking happened in Q2, after taking Justin Welsh’s LinkedIn Operating System course. I initially dismissed Welsh as a LinkedIn influencer bro, but he grew on me – especially as his content just makes sense. The shift was fully “getting” the idea of a solo expertise business with multiple revenue streams. As with other shifts, it was partly experiential. I paid something like US$100-150 for his course on a whim, which is how he designed it. Given that he’s sold the course to over 20,000 people, you can figure out yourself how valuable such “whims” are to a soloist.

Now, I’m not going to start selling “how to LinkedIn” courses, but coming from the world of building complicated software and data solutions, I have a silly bias against finding “low tech” ways of providing value. And that is the thing that took embarrassingly long to get to: “helping people you like get what they want” can take many forms. It can be a high-touch fractional Chief Data & AI Officer engagement with a climate tech startup, or it can simply be pointing people who ask me about solo consulting to this post. And it can be anything in between – even if it doesn’t perfectly align with my high-touch mission of helping climate tech startups.

For example, given my new mindset of low-touch monetisation, I decided to look for an appropriate affiliate for my established-remote company list. This list gets 100-200 unique visitors on a normal day, with some mad spikes (e.g., almost 10,000 visitors two weeks ago). A quick search led me to Remote Rocketship, so I signed up and added the affiliate link to the list and to my long and popular rant post on automated coding assessments. To date, those links have generated a bit over US$100, which is far from life-changing. But given the low level of effort required to add the links, the effective hourly rate on that revenue stream is going to keep rising.

Another key change I implemented in Q2 was committing to the solo expertise business more wholeheartedly. In a sense, I consider April 2024 to be the true start of the business, after a year of messing around. A key part of this commitment was revamping my LinkedIn profile and sharpening my positioning. In the spirit of Stark’s laser-focused positioning statement, this is more or less where I stand:

I’m a Data & AI expert who helps climate & nature tech startups with shipping data-intensive solutions (advanced analytics / AI/ML). Unlike my competitors, I have over a decade of experience working with startups & scaleups across data science, engineering, and tech leadership roles – including close collaborations with product & marketing teams, and a track record of respectfully communicating inconvenient truths to CEOs and other authority figures.

Following Welsh’s advice, I see my LinkedIn profile as my top-of-funnel landing page, so it looks like this now:

I post regularly on LinkedIn as well, and I’m seeing results: I get inbound interest from people who understand what I’m offering – even if they don’t exactly fit my target market.

Further, I remember that there’s a world outside LinkedIn and online content. I’ve joined and made connections through Climate Salad, mentoring with EnergyLab and GrowthMentor, and good old in-person meetups.

Overall, Q2 2024 was a bit slow when it came to client work, but there were leads galore, and some of them became clients later.

What did I learn from Q2 2024?

1. Selling to Carol works. In other words, committing my focus to a specific ideal client profile (ICP) has yielded the promised results – it’s easier to know who I should speak to, and prospects understand how I can help them.
2. I’m in the business of farming, not whaling. Just because my ICPs get what I’m about and want to work with me doesn’t mean they’re ready to buy right now. Leads that I started talking to in 2023 only came close to signing contracts towards the end of Q2. This means that I need to maintain enough concurrent relationships (sowing seeds, watering, etc.) for the consulting revenue stream to be viable. Hunting whales is less fun, leads to feast-famine cycles, and is illegal in most countries for good reasons.
3. Long-term, I should figure out ways to help people I like at every price point. To take another unsustainable marine analogy, it’s a bit like fishing down the food web. High-touch consulting is only accessible to large fish, and I can only do little of it due to time constraints. Low-touch info products and affiliate partnerships are accessible to a broader audience. I don’t know what it’ll look like for me, but I’ve come a long way from seeing high-effort software products as The One True Path.

July 2024: Leads blooming and a path to fractional stability

Going into the new financial year, the opportunity cost of all my messing around felt more salient. I’m fortunate to be in a stable financial situation, so the net value of my assets has increased despite lower-than-desired income from the business. However, I’d be in a better position if I had worked full-time over the past year.

Given the opportunity cost, and following my understanding of the marketing I need to do for a sustainable business, I gave that big tech opportunity I mentioned a second thought. Then I looked at Glassdoor and gave it a third thought: Nope, if I go full time there, it’ll likely lead to too much drama that’ll cause me to drop the business.

However, in the spirit of sustainable revenue generation, I started exploring a few options for fractional (aka part-time) work that’d give me some peace of mind while I generate a more stable pipeline of clients that match my positioning. I hope to start contracting with one of them in August.

Beyond that, I signed a contract to deliver an AI/ML audit package, which I alluded to last week. Getting to that point took forever for various reasons, but it’s the sort of work I want to get better at: I aim for flexibility and creating high value. As I’ve been indoctrinated by Stark, I know that the path there can’t be through selling my hours (Hourly Billing Is Nuts!). Productised consulting is one path to creating value and charging independently of hours spent – but to get to a product I need to repeatedly sell the same service. One step at a time.

On the marketing front, my LinkedIn posts and newsletter haven’t had amazing traction, but some have sparked interesting conversations. I’ve also had the chance to help people I like for free at scale by giving a webinar on the first Data-to-AI hire to Climate Salad startups. The positive response was heartwarming, and it also led to some interest from prospective consulting clients. However, these are just more seeds in the business farm – gotta keep the expectations low and focus on long-term value.

What did I learn from July 2024?

1. I still have a lot to learn on publishing content I like. The initial step was putting stuff out there consistently. The next steps are to consistently make it better. For example, I recorded myself in preparation for the webinar and guested on a couple of podcasts, but my online speaking leaves a lot to be desired (at least by me).
2. My profile seems sufficiently clear to my ICPs, and my buyers lean to the technical side. I had a few conversations with marketers who don’t fully get what I do – but that’s fine. It’s countered by positive feedback from ICPs (climate & nature tech startup founders), e.g., one reached out and said I have the best AI profile he’s seen. I certainly don’t have the best AI profile I have seen – but I’m not my ICP. It’s all about getting the message across to the right audience, and I will keep tweaking it as I go.
3. Persistence, rather than obstinacy, is key to success in the long term. This is inspired by a recent Paul Graham essay. The gist of it is that persistence includes constantly learning and deliberately changing how I do things, whereas obstinacy is the wrong kind of stubbornness – beating one’s head against a wall. For example, I’ve long considered the idea of selling advisory calls via paid ads. In July, I reached out to marketers via GrowthMentor to assess the viability of the idea, and they both advised against it. The obstinate path is insisting on paid ads (it’s tempting to be able to turn $1 into $2), while the persistent path is doing what’s worked for other solo expertise businesses: farming relationships, content marketing, and delivering value to clients and prospects. It’s not fast or easy, but it should work with enough flexible persistence.

Onwards to monthly reflections

When I started writing this post, I felt a bit wary for two reasons.

First, my business is still a work in progress and may ultimately fail, so I’m a bit uncomfortable with sharing some of its inner workings with the world. However, I’m learning to push through such discomfort since I see the benefits of authentically sharing what I’m learning as it happens. I believe it’s in line with helping people I like, including my current ICPs of climate tech startup founders – because some lessons are transferable across different business types.

Second, I knew that to set the scene for monthly reflections, I’d have to summarise my path since last April, which would result in a post that’s too long that may still not cover everything. Indeed, it took too long to write the post – I aim to start and finish writing my weekly posts on Monday mornings, but it’s late afternoon now. And indeed, I feel like I missed a lot of lessons I’d like to share. However, if I stick to doing these monthly from now on, they’ll be shorter to write and read, and I’m sure some lessons will recur along the way.

In the meantime, if you have any questions or thoughts on my reflections, I’d love to hear from you.