Why we became a studio.

We are going to start with the uncomfortable part, because burying it would defeat the point of writing this at all. Cobalt Glacier published an acquisition thesis. We wrote about why permanent capital is the right structure for software, about how we would pay founders who sold to us, about diligence checklists and LP-grade reporting and cap-table cleanups before close. We had the mechanics of an acquirer worked out in real detail. What we did not have, after all of that thinking, was a single closed acquisition. Every brand under the Cobalt Glacier roof today, Lucidly, AdvisorKit, FlashGTM, RenewalPad, TempoDocs, Attribufi, and RecruitFractional, was conceived and built by us, from a blank page, not bought from a founder who built it first.

That is not a footnote to the story. It is the story. This essay explains what we actually did versus what we said we would do, why the arithmetic underneath company-building changed enough to make that shift rational rather than accidental, what we think a studio owes an operator who joins to co-found a brand with us, and what stays exactly the same as it was under the old thesis.

What we said we would do, and what we actually did

The original thesis was straightforward and, on paper, still sound: bootstrapped B2B SaaS founders in the ten-year range often want liquidity without wanting to hand their company to a private-equity sponsor on a five-year clock or to a strategic that will fold the product into a roadmap and retire the brand. A permanent-capital buyer that intends to hold for decades fills a real gap in that market. We built the underwriting discipline, the payout structures, the reporting cadence, all of it, to be that buyer.

Then we went looking for deals, and the thesis ran into a harder problem than we expected. Good bootstrapped B2B SaaS businesses that fit our box were scarce, priced aggressively by founders who had other options, and slow to close even when the fit looked right. Meanwhile, inside the firm, small teams kept sketching product ideas in categories we understood well: advisory workflows, GTM tooling, contract renewals, technical documentation, marketing attribution, fractional recruiting. Those sketches kept turning into working prototypes faster than any acquisition process was turning into a signed term sheet. We were, without quite admitting it to ourselves at first, building the thing we had set out to buy.

Somewhere in the second year, we stopped pretending the acquisition pipeline was the main event and the internal builds were a side project. It was the other way around. Seven brands existed, all built in-house, all AI-native from day one, and zero acquisitions had closed. Continuing to describe ourselves publicly as a holding company that acquires businesses was, at that point, simply inaccurate. The studio label is the one that matches what we actually do.

Why building started beating buying

The honest driver here is cost, not ideology. Acquiring a proven B2B SaaS business has always meant paying for years of accumulated product-market fit, a working go-to-market motion, and a retained team, usually at a meaningful multiple of revenue. That price makes sense when building an equivalent product from scratch would take a founding team eighteen to thirty-six months and a headcount-heavy engineering org. That was the calculus for most of the last two decades of software.

AI broke that calculus, at least for a meaningful slice of B2B SaaS categories. A three- to six-person team with AI-assisted engineering, AI-assisted support tooling, and AI-native product design can now take a well-scoped vertical SaaS idea from spec to a sellable version in a window measured in months, not years, and can operate the resulting business with a support and ops footprint a fraction of what the same product would have needed five years ago. We wrote about this directly in AI doesn't lower SaaS prices, it widens margins: the same force that compresses build cost also compresses the ongoing cost to run the thing, which is why margins move up rather than down.

Once the cost to build a credible B2B SaaS brand drops far enough, the acquisition premium stops making sense for most of the deals in our original box. Why pay a mid-single-digit multiple of ARR for a business we understand well enough to have underwritten, when a small internal team can build a comparable product, aimed at the same buyer, in a fraction of the time and at a fraction of the cost, and keep full economics and full control of the brand from day one? That is not a universal truth about all of software. It is a truth about the specific, well-scoped, workflow-heavy categories we operate in, and it is the single biggest reason building overtook buying inside this firm.

AI is how we build, and what we build

It is worth being precise about what "AI-native" means here, because the phrase gets used loosely. Inside Cobalt Glacier it means two distinct things. First, AI is how we build: small teams use AI tooling across engineering, design, support, and content to operate at a velocity that used to require a much larger headcount. Second, AI is what we ship: every brand in the portfolio is an AI product at its core, not a legacy workflow tool with a chatbot bolted on. Those two facts compound. A team that builds with AI at high velocity is also the team best positioned to ship an AI-native product, because they are living inside the same toolchain they are selling a version of.

What a studio owes an operator who co-founds with us

We are not claiming we build everything alone, and we should not. There are categories where the right product idea needs a domain expert in the room from day one, someone who has run the workflow we are trying to replace and knows exactly where the sharp edges are. For those situations, Cobalt Glacier occasionally co-founds a brand with an outside operator rather than building it entirely in-house.

This is where the acquisition-era mechanics we published earlier still matter, just aimed at a different moment in the relationship. The payout structures we described in how permanent-capital holdcos pay founders assumed a founder selling us a business they already built. The underlying logic, cash sized to remove financial pressure, meaningful equity that lets the operator participate in the compounding they create, and a role that does not force them out of the building, translates directly to a co-founding relationship. The operator is not selling us a company; they are joining us to build one. But the question they are really asking, "what do I actually get, and do I stay in the seat," is the same question, and it deserves the same level of specificity in the answer.

What we owe a co-founding operator, concretely: real equity in the brand, not a token grant; a clear decision-making split between what they own operationally and what the studio owns structurally (capital allocation, shared services, brand-level reporting); AI engineering and product-design capacity from day one so the domain expertise does not sit behind a six-month build queue; and an honest conversation, up front, about what happens at each of the outcomes we plan for, hold, sell, recap, or list. An operator who joins to co-found a brand should never discover the terms of that conversation for the first time three years in.

What genuinely does not change

It would be easy to read the studio pivot as a wholesale reinvention, and it is not. Three things carry over from the original thesis unchanged, because they were never actually about being an acquirer. They were about how we intend to run capital and businesses over a long horizon, and that intention did not move when the mechanism for building brands did.

• No fund clock. We are still not underwriting to a five-year exit window. A brand we build gets the same patient runway a brand we might have acquired would have gotten. Nothing about building in-house changes the math on why time is an advantage rather than a constraint.
• Outcome-driven per brand, not portfolio-wide. Hold, sell, recap, or list is still a decision made brand by brand, based on what is right for that business and its operators, not a calendar the whole studio marches to in lockstep.
• Operators run brands. The commitment we made in operator continuity is the real asset holds just as much for a brand we built as it would for one we bought. The person running the day-to-day of a Cobalt Glacier brand, whether they founded it inside the studio or co-founded it with us, is the person best positioned to keep running it for as long as the job is being done well.

The reporting discipline we described in our LP-grade reporting piece also carries forward unchanged in spirit: we still think the absence of an external forcing function, no LPs, no fund clock, means we owe ourselves and anyone watching a rigorous, voluntary standard of transparency about how each brand is actually performing. That standard does not care whether the brand under the microscope was bought or built. It applies the same way either way.

Why we are telling you this instead of quietly editing the website

We could have rewritten the acquisition-era essays and pretended the holdco chapter never happened. We are choosing not to do that. Those essays are still on this site, with an editor's note at the top of each one pointing back here, because the reasoning in them was not wrong, it was just aimed at a mechanism we ended up not using at scale. Founders and operators evaluating us deserve to see the actual history, including the part where the plan changed, rather than a version of the record that was cleaned up after the fact. A studio that is honest about its own pivot is a better bet for an operator thinking about co-founding a brand with us than one that is not.

What this means if you are evaluating Cobalt Glacier today

If you came to this site expecting a holding company evaluating your business for acquisition, the update is simple: that is very unlikely to be the conversation. We are not actively running an acquisition pipeline, and we do not expect to be the right home for a company you have already built and want to sell outright. If you are an operator with deep domain expertise in a B2B workflow, who has been thinking about building a company around it and would rather do that with an AI-native product team and patient capital behind you than raise a venture round and hire a headcount-heavy team, that is exactly the conversation we want to have.

We stopped describing ourselves as a company that buys good software businesses because we were not, in practice, a company that bought good software businesses. We were a studio that builds them. Saying so plainly is not a rebrand. It is a correction.

Read more about how we structure that work on the Lucidly case study, our first brand built entirely in-house under this model, or take a look at how Cobalt Glacier is organized today. If you are the operator we described above, the next step is a conversation, not a pitch deck: come talk with us about co-founding a brand and we will tell you plainly whether the fit is real.