"Focus on one thing." That's the advice every founder gets, from every investor, in every era. It's not bad advice. It's been right enough times that it became the default. If you're building a single SaaS product, the focus model usually works. You go deep, you find product-market fit, you scale, you exit. Single product, single team, single thesis.

The reason it works is that building a product used to be extremely expensive. You needed a big team, a long time, a lot of capital. You couldn't afford to build two products. The opportunity cost of dividing attention was huge.

That cost equation just changed.

What AI did to the cost of building products

If you're building software in 2026 and you're being honest about what AI tooling can do, the cost of building product N has fallen by something like 5 to 10x for a competent technical team. Not the cost of distribution. Not the cost of customer support. Not the cost of regulatory compliance. The cost of actually writing the software.

This is a real, measurable thing. A two-person team with modern AI tooling can ship a working v1 of a product in weeks that would have taken a five-person team three months in 2022. Multi-modal models read documents. Coding agents write boilerplate. Test generation is automated. The boring 60 percent of any project that used to consume most of the engineering hours is now mostly handled.

What this means strategically is that the marginal cost of an additional product is way down, and the marginal value of focus (in the strictly-one-product sense) is way down too. You can build more, faster.

This is the thing that makes a holding company model viable for a small team. You wouldn't try to build five products at once in 2015. You can try it in 2026, and a few teams (including us) are.

What we actually built

To make this concrete, here's the current GIS portfolio:

  • Cloak. Chrome extension and API for humanizing AI writing.
  • Gravity Realty Group. AI-driven commercial real estate brokerage.
  • Titan01. CRE opportunity signal aggregation.
  • ClawHouse. AI agents predicting Polymarket.
  • Station CRM. AI-augmented sales pipeline tool.
  • Basic CRM. The next-generation, AI-native CRM, in active development.

That's six products across four categories (AI writing, CRE, AI research, sales software) from one company. The conventional advice says we should pick one and kill the rest. We disagree, and the reason is structural.

The structural advantages of the holding company shape

Shared infrastructure is real leverage

Every one of our products needs the same things. Authentication. Billing. Customer support. Analytics. Email infrastructure. Deployment pipelines. Document storage. Vector databases. Most of those are now built once and reused across the portfolio. The marginal cost of standing up product N is much lower than starting from scratch.

This compounds. By product four we had a backbone that meant product five took half the time. By product six we're shipping more deeply integrated experiences (like Cloak's signed-license API) because the foundations were already there.

Knowledge transfer across categories

The lessons we learned building Titan01 (how to extract structured data from terrible PDFs at scale) directly informed our work on Basic CRM (how to extract structured data from email and voice). The patterns we developed for Cloak's text-scoring model are showing up in our internal tooling for Gravity's listing analysis. None of this would have happened in single-product mode because the products wouldn't exist.

Risk diversification at the portfolio level

Some of our products are going to work. Some are not. We know this going in. The portfolio model lets us take big swings on speculative products (like ClawHouse, which is openly experimental) while still operating commercially viable businesses (like Gravity). A single-product startup has to bet everything on one thesis. We don't.

This is conservative in one sense and aggressive in another. We're conservative about company survival (the failure of any one product doesn't kill us). We're aggressive about exploration (we can pursue weirder ideas because they aren't existential).

Cross-product distribution

Customers of one product often become customers of another. The brokers using Titan01 are increasingly using Gravity. The writers using Cloak are talking to us about CRM features. The portfolio creates a top-of-funnel for itself in a way a single product can't.

The honest tradeoffs

This isn't a free lunch. Holding company structure has real downsides we've felt.

Fundraising is harder

Venture capital is built around the single-product narrative. "What's the one thing you're building? What's the TAM? What's the go-to-market motion?" The honest answer for a holding company is "six things, six TAMs, six motions." That's an unusual pitch and not every investor is set up to evaluate it.

The workaround is to fundraise at the operating company level when we need to. Each product can stand on its own commercially. Some may eventually take outside capital. The holding company itself stays self-funded.

Focus is genuinely harder

The classic single-product advice exists for a reason. When you're working on six things, you're sometimes working on five. The discipline of saying "this is what we ship this quarter and nothing else gets attention" is harder when there are multiple urgent things across multiple products. We've had to develop explicit operating rhythms to keep this from becoming a problem.

Brand dilution

"What does General Intelligence Systems do?" has a longer answer than most companies. This is a marketing problem. We've leaned into it by framing the company as a holding company explicitly, with the products as named brands underneath, but it does mean we don't get the simple-pitch benefits of a single-product company.

When this model fits and when it doesn't

The holding company model is not right for everyone. It works when:

  • The founders are operators who can hire and trust strong sub-leaders
  • The products are in adjacent enough categories that infrastructure is genuinely shared
  • The capital structure is flexible enough that you're not dependent on traditional venture rounds
  • The team is technically strong enough to actually ship multiple products

It doesn't work when:

  • You're a first-time founder still learning what "ready to ship" looks like
  • You're chasing a single huge market where focus would let you win
  • Your product needs deep regulatory work that's category-specific
  • You're allergic to context switching

Why we think this matters more broadly

The reason we wrote this post isn't to argue everyone should structure their company this way. The reason is that the cost-of-building math has changed permanently, and we don't think the implications have been fully metabolized yet.

You're going to see a lot more multi-product companies in the next five years. Some will be holding companies like ours. Some will be platform plays. Some will be venture studios. The common factor is that AI has made it cheaper to build, which has made it rational to build more, which has made the focused-single-product default less obviously correct.

If you're thinking about company structure right now, the question worth asking isn't "what's the conventional wisdom?" It's "what does the conventional wisdom assume that's no longer true?" In our case, it was the assumption that building product N would cost roughly what building product 1 cost. That hasn't been true for two years. We bet our company on that, and so far the bet is working.

For more on the technical side of how we think about building, see AI-Native vs AI-Powered. For the operating philosophy on specific product categories, our writing on CRM design and AI in commercial real estate covers the deeper detail.