How to Pick a Startup Idea – 1984 Ventures - Founders Handbook

Every industry is being transformed by AI. As a founder, this means there have never been more opportunities, since nearly every sector is now in play. Yet it's not trivial to find an idea worth pursuing. Below I lay out my thoughts on how to find a good startup idea, including some non-obvious examples from our portfolio in the hope they spark some thoughts.

1) Ignore market size

I meet with many founders who focus on market size when evaluating opportunities (e.g. customer service is a huge market!). I don't believe this is a helpful filter to find an idea. Large markets are obvious, attract competition (100x more in the AI era) and already have incumbents. Unless there is a clear innovator's dilemma, there is a real risk of being crushed by an incumbent's distribution advantage.

2) Pick a niche or offline market

Instead, I would advise founders to focus on an esoteric or niche market with old school competitors. This includes offline or services industries e.g. insurance claim processing or loan servicing. The reasons are threefold. First, these markets are larger than they appear because when you automate labor, you command much higher ACVs than pure software (i.e. selling work, not software taps into labor budgets, not software budgets). Second, choosing something hidden from plain sight or more esoteric shields you from competition. And third, I have seen firsthand how founders who effectively solve their customer's problem get access to more problems and budgets over time, building a layer cake that expands their market size.

A great way to look for these markets is to think about jobs that your family and friends do. Perhaps a parent works as an architect. Or a friend is a conveyancing lawyer. Or an uncle runs a travel agency. These types of sectors are now interesting places to build a startup.

3) Find a manual workflow and then automate it

The typical advice is to find a hair on fire problem that customers are desperate to pay for. This makes sense, but in the age of AI is limiting. Instead, you should look for tedious, repetitive and manual workflows/jobs/roles and design a solution that automates it. (Labor costs are often the largest cost item for services businesses, so labor automation is very much a solution to a hair on fire problem).

Some examples of job or task automation could be:

• Automating the drafting of marketing proposals within a particular industry (e.g. architecture)
• Using web scraping and LLMs to aggregate and synthesize court filings. This could be sold as a data product to lawyers
• Automating calls in contact center-like industries (e.g. mortgage servicers)
• Programmatically managing marketing campaigns, thus building an AI marketing agency

But… while it is tempting to think of automation through the lens of cost cutting, there are better ways to position your product.

4) Focus on revenue generation, not just time or cost savings

Many founders think of automation as being something that saves time and cost. But in my experience, unless the efficiency gain is 10x, they don't get adopted by customers. If the efficiency gain is not visceral, you aren't thinking big enough.

Instead, the better framing is to think of AI as tool that can make customers more money.

This is best brought to life with an example. Imagine a recruiting agency makes 20 placements a year and makes $20K per hire. But an AI tool comes along and automates all their first round interviews (cost saving), allowing them to make each hire faster (time saving) and thus undertake 30 placements a year (a revenue expansion). This type of efficiency gain is available across a wide swath of industries, but tends to resonate most strongly with customers when you help them grow their topline.

Examples of revenue generating value propositions

Below, I include some examples of revenue generating value propositions from our portfolio. The purpose is to show how founders targeting services industries (e.g. debt collection, auto repair, medicine) have been able to position efficiency gains as something more valuable than time and cost savings.

• Collectwise uses AI to automate debt collection, a $35B market in the US. As a 6 month old startup, they've already 2x'd the collections rates of industry incumbents. These incumbents are 100+ year old companies that are unlikely to modernize their approach and tech stack, while Collectwise's solution will only improve over time.
• Apriora is an AI interviewer, sold to recruiting firms to help them automate interviews. Automating initial screens drastically changes the cost structure of a recruiting firm. It allows them to place candidates faster (while keeping costs constant), and thus undertake more placements per year.
• HouseRx enables specialty doctors to dispense medication directly through their practices. This creates a new revenue stream while significantly improving patient experience and outcomes.
• Exacare is an AI powered CRM for the senior care industry. CRMs typically do not solve hair on fire problems. However, Exacare's AI functionality allows facilities to review 250+ page patient referral docs in 2 minutes (down from 45), quickly determining if a patient gets admitted. Facilities are bottlenecked by time spent on administrative processes, so Exacare's product is revenue generative, cost reductive and frees up valuable time to focus on patients.
• Cline is an AI coding assistant. It turns tasks that took developers hours into minutes. Whilst this is a time saving value proposition, developer time is highly valuable and the efficiency gains are so stark that it easily meets the bar for radically better than the status quo.
• Revv uses AI to help auto shops automate vehicle research. Research automation is typically a time saving value proposition, but in this case it's directly tied to revenue generation. This is because Revv focuses on vehicle sensor repair procedures which are complex (i.e. take lots of time to research) and lucrative (generating $1K+ of high margin revenue per vehicle). Time saved doing research allows for more vehicles to be repaired and allows shops to make more money.

Revv's primary pitch is to make body shops an additional $1,250 per job