Finding AI Disruption Opportunities using Deep Research
Here's a simple technique to find specific companies that might be vulnerable to AI-first disruption using Deep Research.
Everyone is talking about AI disrupting everything, but how do you identify specific AI-first opportunities as an entrepreneur?
In this post, I will share a technique for finding specific AI-first opportunities using OpenAI Deep Research AI agent. This is not a get-rich-quick scheme, and you’ll certainly have to do your own heavy lifting as an entrepreneur. However, I believe that Deep Research can considerably accelerate the process.
The process has four steps:
Formulate your thesis
Define the shape of the solution
Use Deep Research to find specific opportunities
Iterate
Let’s dive in!
Formulate a thesis
The starting point is formulating a thesis around why you believe AI will be a considerable disruptor. There isn’t a single answer here; there are countless possibilities. What you’re looking for is just one angle that aligns with how you AI disruption.
Here’s my example. I believe that AI will create opportunities to compete with small established businesses with a solid product-market fit (PMF) and a predictable cash flow through a 10x higher efficiency or 10x lower cost.
Today, many businesses with a great PMF and predictable cash flows are not attractive to invest it or compete with because of their small scale. A private equity firm might buy a business with £50m in revenues and £5m in profits, aiming to get it to £100m and £20m, respectively. But the same approach doesn’t work on a small scale because the numbers don’t add up. I believe that AI tech will change that.
Consider a fictional example of a background checking firm with 100 employees, £10M in revenues and £1M in yearly profits. When an HR team at a big company needs to verify a candidate’s CV, they send it to this firm. Their team calls previous employers to confirm the accuracy of data and looks up university archives to confirm education credentials. It’s a lot of manual work and the market isn’t going to explode.
It has a strong PMF, giving it reliable cash flows. However, it’s too small and too niche to attract investment. It’s doing its work in a largely manual way.
So an AI-first disruptive opportunity for an entrepreneur would be to build the same service that automates all checks and operates at a fraction of the cost using AI tech from the likes of ElevenLabs. Since their clients don’t care who does the job as long as it’s done well, they’ll likely switch to a cheaper and faster product.
Why am I focusing specifically on small businesses here? I think they’ll be easier to disrupt because big companies have more money and expertise to invest in such tech and, often, bigger moats (more complex products, long-term complex contracts, regulation, etc.) Small companies might be more vulnerable to such disruption.
Why am I focusing on companies with strong PMF? To eliminate market risk. Building stuff is easy, but knowing what people will pay money for is hard. So we should look for examples of strong PMF and weak defences.
Once you have a disruption thesis in mind,1 how do you find specific targets? Let’s outline the solution.
Define a shape of a solution
Now that we have a thesis, let’s outline what we are looking for:
an SME in the UK with revenues under £15M
strong PMF
predictable cash flows
human-intensive value creation (this is key!)
operating for at least 10 years
value creation that could be 10x improved with AI
But given how many companies that fit this description exist, how do you find them? Especially given that many of them will be operating in small niches that you never heard about?
Let’s as ChatGPT Deep Research.
Use Deep Research to find specific companies
We can use Deep Research to find specific companies. Here is a prompt I would use, based on our thesis and our desired target companies.
### Objective:
Identify at least 10 UK-based SMEs suitable for AI-driven disruption based on the following thesis and criteria.
**Disruption Thesis:**
AI technologies enable entrepreneurs to disrupt small, established businesses with strong product-market fit (PMF) and predictable cash flows by significantly increasing efficiency (up to 10x) or reducing costs (up to 10x). Small businesses, lacking financial resources and technical capabilities to invest in AI, are particularly vulnerable.
Focusing on SMEs with strong PMF removes market risk, as these businesses already have reliable revenue streams and stable customer bases.
### Criteria for SME selection:
- Based in the UK
- Annual revenue under £15 million
- Operational history of at least 10 years
- Demonstrated strong PMF (consistent revenue and customer retention)
- Predictable cash flows
- Highly human-intensive processes (e.g., background verification, manual data entry, transcription, customer support)
- Clear potential for significant efficiency gains or cost reductions (at least 10x) through AI automation
### Example scenario:
A UK-based SME conducting background checks with ~100 employees, ~£10M revenue, and ~£1M profit annually. Its manual verification processes are costly and inefficient. AI-driven automation (e.g., employing ElevenLabs technology) could dramatically lower costs and increase speed, making it vulnerable to disruption.
### Desired Output:
Identify and detail at least 10 SMEs that meet the above criteria, including:
- Company Name
- Industry
- Brief service/operation description
- Estimated annual revenue
- Year established
- Description of human-intensive processes
- Assessment of potential AI-driven improvements (highlighting 10x efficiency or cost gains)
### Next Steps:
Briefly assess feasibility and strategic advantages of targeting these identified SMEs for AI-based disruption initiatives.
Here’s what I get if I run Deep Research on this prompt. It’s too long to copy-paste here, so check the link to see the depth of the analysis. It’s a list of companies that are ripe for AI disruption in line with my thesis and an explanation why each of them is on the list. Here’s a very short summary:
1. Verifile – Background screening service; highly manual verification process, AI can automate document checks and flag issues.
2. Accuro – Medical & legal transcription; AI speech-to-text can replace human transcriptionists.
3. Answer4u – Telephone answering service; AI chatbots/voicebots can replace human operators.
4. Time Etc – Virtual assistant service; AI automation can handle repetitive admin tasks, reducing human effort.
5. Hilton-Baird Collection Services – Debt collection & credit control; AI can automate payment reminders and risk assessment.
6. Dataplan Payroll – Payroll processing; AI can automate payslip generation, tax filing, and compliance checks.
7. Wolfestone – Translation & localization; AI-powered translation can replace human translators for many tasks.
8. Eagle Couriers – Last-mile delivery service; AI can optimize route planning and customer service automation.
9. The Telemarketing Company (TTMC) – B2B telemarketing; AI voice agents and predictive dialing can improve efficiency.
10. Face for Business – Virtual reception & live chat; AI chatbots and voice AI can handle customer inquiries.
Iterate
Reading this report, I notice that several companies aren’t exactly what I was looking for.2 This is a chance to update my thesis and my prompt and run it again.3 But even as it is, I have already learned a lot from it about companies ripe for AI-first disruption in various industries I didn’t think about before.4
So, after you get your first report, iterate by improving the prompt based on what you learned. After a few iterations, you’ll get a far better version. More importantly, it will stimulate your thinking and give you new ideas.
Conclusion
Deep Research may seem like an expensive AI agent ($200/month for 100 reports a month or $20/month for 10 reports a month), but it can really speed up the research process if you’re trying to identify AI-first disruption opportunities.
The process outlined above can be adapted to different scenarios, e.g. looking for opportunities to apply AI inside your own business. What’s important here is that we’re learning how to use AI agents as a thinking partners, doing fast research for us and expanding our thinking process as if we were surrounded by an eager team of MBA students 24/7, happy to do as much work as you ask them to do.
Found this post helpful? Forward it to your colleagues!
Exercise for the reader: give your thesis to AI, explain what you’re doing to do with it and ask to improve it. Then proceed with the rest of the process.
For example, I’m sure that debt collection (#5) is more than just payment reminders. Same for #8 — courier delivery.
It’s an exercise for the reader, otherwise this post will be twice as long.
The fictional example of CV-checking service that I used above? I learned that such thing exists (and how many companies use it) from Deep Research looking for such companies. I wasn’t even thinking about this problem before.