Wait... Are We Consultants?
I was chatting with Aviv, our CTO, the other day about project plans for the upcoming quarter. We got into a deep discussion when he asked me, “So are we operating more like a consulting business? Or a SaaS business?”
It’s a fair question. For those who don’t work with us: we have both a platform team and client-facing product engineers. The platform team operates more like a SaaS business, building universal tools and processes used across the company. The product engineers work more like consultant, using those platform tools to assemble custom solutions curated for each firm.
So which one are we?
The Engineer's Bias
Consultants get a bad rep, especially in technology circles. A quick Google search will bring you content like this:
I used to agree with him. I took a lot of pride in building software that resonated with a wide audience. I owned my products for years, learning from user feedback to fix bugs and broken workflows. Consultants, on the other hand, built bespoke solutions and left before getting to the hard part: making their product great.
But my opinion relied on the assumption that quality and customizability were inherently antithetical. I thought it was impossible to build well-designed products in a custom way. But now I think it’s not only possible, but integral, in building a strong business in the age of AI.
The Consulting Model’s Moment
Last summer, MIT published research report that made waves in the AI community1. Despite $30-40 billion in enterprise GenAI investment, the study found that 95% of organizations are getting zero return.
The report calls it “the learning gap.” Real enterprise workflows are nuanced and complex, and static off-the-shelf solutions can’t effectively optimize them. On the other hand, organizations were significantly more successful at adopting AI when they hired external vendors to build custom tools that integrated into their existing workflows2.
So it turns out the key to unlocking AI's potential at your firm isn't getting everyone a ChatGPT subscription and hoping for the best. It's bringing in someone who can actually understand your business and build for it.
SaaS Constraints Are Crumbling
Enterprise SaaS has been moving toward consulting for years, even before AI.
Solutions and sales engineers were the first step. Companies like Salesforce and Snowflake built configurable products, which internal experts set up for each client. Forward-deployed engineers and deployment strategists took this further with the idea of a core platform that’s customized by client-facing engineers.
This progression happened because companies needed customization but building truly custom software was prohibitively expensive. Engineering resources and time were hard constraints. So companies compromised—they built flexible platforms that could be configured, not truly custom solutions.
AI changes that equation entirely. Now engineers can have an army of agents building in parallel, 10X-ing their output. The constraint isn’t engineering capacity anymore—it’s having the right tools and frameworks to direct that capacity effectively.
Where We Fit
If you’ve been following us for a while, you might remember a time when our website advertized a core platform that could ship with custom configurations, Palantir-style. But as we spoke to more funds and built more AI, we noticed two things:
Clients need more custom solutions than we originally expected
New developments in AI (agents, MCP, skills, etc) can build net-new software radically faster
We took those insights to heart and made a bolder bet.
Instead of investing in a core platform, we pivoted to invest in two things. The right tools—an arsenal of capabilities our platform team builds and maintains. And the right knowledge—a team with the relevant experience to understand your business and design elegant solutions.
Which brings us to the current paradigm. Our platform team builds core capabilities. Our product engineers compose those capabilities into a completely custom platform for each client. There’s no need for any standardization in the middle.
Growing Pains
Adopting this model is harder than it sounds, especially when you’re building the tools at the same time you’re using them. The platform team need to build resilient tools that work across multiple clients. Product engineers need to move fast and deliver solutions for their client.
Those needs often conflict, which can be challenging and uncomfortable. Sometimes we ship platform features that only one client needs, betting (sometimes incorrectly) that others will need it soon. Sometimes product engineers build workarounds when the relevant tool isn’t ready, creating diverging designs and technical debt.
But I see these challenges as a privilege, not a problem. Most companies are still trying to force-fit themselves into the SaaS or consulting or forward-deployed buckets. And I feel like we’re building the muscle memory for what comes next. Whatever it ends up being called.
This Substack post provides interesting tips on how companies can improve AI adoption.