What I've been up to recently

While I generally try to keep SAAS jargon to a minimum, this post is unlikely to be entirely free of it - sorry not sorry

At the start of this year I was in two minds. I enjoyed my work at Attio - the company was full of great people, growing like crazy, and I enjoyed playing my own small part in throwing fuel on the fire there.

On the other hand, I'd always felt drawn to the idea of running my own business. That urge had only intensified recently: the end of my twenties was looming in sight, and I'd still never ventured outside of employment.

I considered a couple of possible options:

• Build something on the side on evenings and weekends: didn't work for me due to some combination of being too focused on my day job + skill issues
• Switch careers into something more potentially challenging and entrepreneurial, like being a FDE (forward deployed engineer) instead of a solutions engineer: evaluated this, but eventually concluded that it would be the wrong call. If my issue is "I like my job but I want to run my own biz" then I should probably just do that, not try and find a job that's 20% more like being a founder.

So that's what I did. So far it's been a blast. There are two kinds of project I'm taking on currently.

CRM consulting

The first is building out new Attio-powered GTM tech stacks for clients. For example this project from the past month

• New VC firm trying to raise a few hundred $m for their first fund
• Can't run pipeline meetings because their existing CRM (Affinity) is a mess
• Need some combination of a functioning CRM + data enrichment workflows
• This requires building an AI workflow that can use a combination of Attio's built-in data enrichment alongside branching and prompts to classify companies in a way that fits their internal framework
• Also requires wiring up Attio to some other enrichment APIs to make their CRM useful / actionable, which involved using Polytomic for reverse ETL

Nothing crazy complex, but also not a plug-and-play - the classification in particular needed a degree of testing and iteration to fit it to the client's specific framework.

This used what Anthropic calls the 'routing' workflow - basically using branching workflows to allow for separation of concerns + specialized prompts. Plan on sharing a a breakdown of this one as I think the pattern is one that can be pretty useful.

I implemented this project with 80x.ai who run the best Attio consultancy for VCs. Working with Dan and Nick has been an amazing crash course in how to operate effectively as a consultant and I've learned a ton from them.

AI workflows

The second is building AI workflows for specific use cases, in a way that is cleanly integrated with the rest of their GTM (including but not limited to their CRM). For example:

• London-based AI startup is growing incredibly fast, GTM people up until midnight answering contact sales emails
• They want to work towards automating replies to contact sales form submissions
• Previous AI solution here had mixed results, so they're wary of automation
• Build solution combining Clay (lead enrichment / scoring) with custom AI workflow for responding to emails
• Scrap initial solution using a vector database in favour of simpler solution (putting docs in context window)
• Curate context from website, iterate on prompts, instruct AI to remove em dashes
• Eventually get to solution their salespeople are very happy with for email draft creation

Early days for this engagement: it's already useful, but I think it can go a lot further - interested to explore the process of taking this from a workflow that's heavily human-in-the-loop towards one that's increasingly autonomous.

I got to work on this project thanks to Alex Sangster who's my go-to person for any question around GTM engineering (and a thoroughly nice bloke)

What's next

There's been a lot of hype recently about the term 'GTM engineering' and while a lot of it is just rebranded RevOps, I see a lot of genuinely interesting opportunity in this space.

The projects I'm working on currently are essentially working closely with growing businesses to help them apply AI to their GTM data. That involves a combination of platform knowledge, data integration, and the ability to apply LLMs in a way that's simple, effective, and not over-engineered, in a landscape that is constantly evolving. As a solo practitioner you are uniquely incentivized to perform that role: you aren't wedded to any particular proprietary framework or LLM. You aren't trying to develop a product that generalizes to every adjacent use case. You're empowered to do things that don't scale in the pursuit of getting results as fast as possible.

I plan to keep applying myself to this space, keep learning, and to share what I learn along the way. If you have a project in mind then drop me a line.