The Sales Test This Norwest Partner Gives Founders Before He’ll Invest

Venture capital is entering a phase of deep pragmatism. The era of writing checks based on a flashy slide deck and a vaguely defined AI strategy is closing. Recently, Sean Jacobsohn of Norwest Venture Partners shared some insight into how he evaluates the current landscape, and his approach should be a wake-up call for anyone building in the fintech, HR, or AI sectors. It is no longer enough to have a good product. You need to be able to sell it, and you need to solve problems that the incumbents are too bloated to fix.

The Founder as the First Salesperson

One of the most striking points Jacobsohn makes is his literal test for CEOs. Before he puts money into a company, he needs to see the founder sell. This is something I have seen trip up technical founders repeatedly. They believe the code will speak for itself. It won't. In a crowded market, the CEO is the Chief Sales Officer. If a founder cannot navigate a sales call or articulate a value proposition that makes a customer reach for their wallet, the business is a hobby, not a venture-scale startup.

For builders, this means your ability to communicate is just as vital as your ability to ship features. If you are a technical lead, find a co-founder who can sell, or start practicing. The market is too competitive for shy products. VCs like Jacobsohn are looking for founders who can penetrate the noise of a saturated market through sheer force of will and clear communication.

The AI Accounting Trap

We are seeing a massive rush into AI-driven accounting and finance tools. The promise is total automation: the AI handles the books, and the humans just watch. Jacobsohn is skeptical, and I am too. In the world of finance, precision is everything. A 95% accuracy rate in a LLM-generated summary is great for a blog post, but in accounting, that 5% error rate can lead to audits, fines, or bankruptcy.

The real opportunity here is not in replacing the human accountant, but in providing them with high-fidelity tools that eliminate the grunt work. If you are building in this space, stop pitching "total automation." Instead, pitch "automated accuracy with human-in-the-loop verification." The companies that win will be the ones that recognize the legal and ethical liability of financial data. Don't trust the AI with the final click; trust it with the data entry and the pattern recognition.

Attacking the Giants in HR

The HR tech space is dominated by massive platforms like Workday and Rippling. Trying to build a better core HRIS (Human Resources Information System) is a suicide mission for most early-stage startups. Jacobsohn suggests a smarter route: attack the secondary products. These large platforms are broad but shallow. Their payroll might be great, but their performance management or specialized recruiting tools are often clunky additions designed to check a box.

Builders should look for the features that users hate in these giant suites. If you can build a standalone tool that is ten times better than the "add-on" feature of a major platform, you have a wedge. It is the classic unbundling strategy. Start small, solve a specific pain point that the big players are neglecting, and integrate with them rather than trying to replace them on day one.

Vertical AI and the Death of Generalism

The "wrapper" era of AI is dying. If your startup is just a thin UI over a generic GPT model, you don't have a moat. Jacobsohn points toward vertical AI—models and tools built specifically for a single industry—as the real area of growth. This aligns with what I am seeing across the board. General-purpose AI is becoming a commodity provided by Google, Microsoft, and OpenAI.

The value now lies in the data. If you have access to specialized datasets in legal, construction, or healthcare that the big models haven't chewed on yet, you have a business. Founders need to ask themselves: What do I know about this specific industry that a general LLM doesn't? If the answer is "nothing," you aren't building a company; you're building a feature that someone else will eventually release for free.

The Pragmatic Path Forward

The common thread in this outlook is a return to business fundamentals. The hype is being stripped away. We are looking for founders who can sell, products that solve specific (not general) problems, and a realistic understanding of what AI can and cannot do. For the builder, this is actually good news. It clears out the tourists and the fast-cash chasers. If you are solving a real problem and you can prove that people are willing to pay for it, the capital is still there. But the bar for entry has never been higher.

The Takeaway

• Master the pitch: If you can't sell your vision to a skeptical VC, you won't sell it to a skeptical customer.
• Respect the data: In finance and accounting, accuracy beats speed every time. Don't over-promise on AI automation.
• Find the gaps: Don't fight the incumbents head-on. Find the weak spots in their secondary product offerings and build something better.
• Vertical is the only way: Specialized knowledge and proprietary data are the only real moats left in the AI era.

Read the original at Crunchbase News →