Previously, I have expressed my frustrations at solving novel problems. My biggest complaint was that the cost to educate people is way too high. Government spends so much money to educate people and still the results are not that great. So, what makes you, as a small business, think you can do a lot for your product by educating people unless it’s like stupidly intuitive? It’s really hard to educate people fast enough to trigger high growth. That’s why the fastest-growing ventures are the ones where the pain is clearly understood by people. Because if you are solving a problem and people are not educated about it, what are you going to do?

1. Create a thousands of pieces of content so that they get organic exposure to it
2. Burn thousands of dollars running ads so that you can keep retargeting that until they understand it

Are you really able to do either of those things if you don’t have much capital? And if you’re starting with low capital, are you going to wait for months before your customer acquisition costs are recovered so that you could run another campaign? I read that people need to be retargeted 7+ times before they consider buying. Are you really going to be able to do that in a feasible amount of time?

And then I looked at Grammarly and Microsoft Word. Microsoft spent decades underlining your wrong spellings and grammatical mistakes. They had a special piece of code dedicated to understanding your text and understanding your mistakes and underlining them. For years, people saw how many mistakes they are making. And they also understood that these mistakes are also seen by their superiors, their professors, their teachers, their boss, their clients, their investors. So nobody wants to look stupid. And people used to hunt down these red lines and fix it. So, they were educated about their problem, and they used to fix it also. It was very effective because Microsoft was the authority here. It told you you are wrong. It will show it to everyone that you were wrong.

Then comes Grammarly. They fix your spelling mistakes and recommend better sentences to you. For years, Microsoft spent educating people about it. Grammarly provided the solution for it. Grammarly had an easier time with growth because people understood their product and they also realized they need it. They inherited an already educated audience that was well-educated about how to solve their problems and the consequences of not solving it. And Grammarly basically said that, “Hey, I can do it faster, easier, and better for you.”

Grammarly got to skip the whole investment on customer education around the problem they are solving. They got to spend less money on marketing because they could obviously convert people better into customers. They got away with a lot of pains associated with category creation because of all the education Microsoft has already done for them. So even if someone out there is using Microsoft Word on some island with no internet, when he comes out and finds Grammarly, he is going to love it!

Same with IDEs also because it showed you that your code was wrong, but it didn’t fix it. Then we got AI coding tools that fixed it. I do not want to imply here that the problems we should be focusing on are only the ones where it shows you you are wrong.

These are what I call “red line opportunities.” Instances where a huge incumbent has been highlighting a problem for so long that the customer base is really educated and sick of it. And you swoop in as the person to solve it. The basic conditions are:

1. Incumbent distribution (mass exposure)
2. The user repeatedly encounters friction inside their workflow
3. Visible friction signal (underline, warning, error, alert, rejection, failure)
4. Repeated emotional reinforcement (embarrassment, risk, loss, delay)
5. User already tries to fix it
6. Fix is partial, slow, or cognitively expensive
7. Switching cost is low if improvement is clear

---

IDEs did three things:

1. Made errors visible
2. Made them frequent
3. Made them personal

But they also normalized:

• Debugging rituals
• Stack overflow searching
• Trial-and-error fixing
• Copy-paste adaptation

AI tools didn’t just “fix errors.” They shortened the time to value and effort to value when it comes to fixing. They created instant gratification within workflow. Just like maximizing productivity or perfectionism won’t let you get too much done, same way you can theoretically make a workflow 100x better but people are still people. Adoption and usage is what matters. Let humans be humans and bring efficiencies via their existing behaviors. They would rather shoot their own dog than do something new. If there are ten steps and step 3 involves clicking a button, can you abstract other steps such that they can be done by just clicking the button at step 3? You have to insert yourself at an existing place in the workflow. Adoption comes from compressing value into already accepted step in workflow.

Another example right out of Microsoft Word can be lawyers and their comma. They use Oxford comma, and misplacement of it can result in huge ramifications. It can cause reputational and financial damages. The misrepresentations and misinterpretations that a single Oxford comma can cause are severe enough that they’ve waited hundreds of times.

The first paragraph when you google lawyers and comma

Lawyers and commas are intrinsically linked because precise punctuation, especially the serial (Oxford) comma, is critical in legal writing to prevent ambiguity, with misplacement leading to significant legal disputes and financial costs, as seen in cases where missing commas altered contract interpretations or even determined treason charges. While consistency is key, lawyers use commas for clarity in lists, introductory/interrupting phrases, and to separate clauses, often adhering to firm styles or specific legal interpretations to ensure documents convey exact intent, making comma usage a high-stakes matter. 

So I want to find a lot of red line problems because they are blue ocean and have highly subsidized education. And you don’t need to change the workflow of your users too much.

Red Line Opportunities

Element	Description
Definition	Incumbent identifies a problem (educates users for years) but doesn’t solve it—just shows a warning or error. You build the solution that actually fixes what they’ve been highlighting. The market is pre-educated; they just need someone to stop the pain.
Grammarly	Microsoft Word showed red/green underlines for 40+ years, training billions of users that spelling/grammar matters. But users had to right-click, scan options, and manually fix each error. Grammarly swooped in with one-click fixes, explanations of why it’s wrong, and tone suggestions. Result: $13B valuation by solving what Word only identified.
Excel Error Solver	Excel shows cryptic errors (#REF!, #DIV/0!, circular reference) but makes users Google and manually debug. “Excel REF error” gets 50K searches/month—people need help. An extension that explains errors in plain English, highlights the exact problem cell, and offers one-click fixes would capture frustrated users immediately. Market is massive: 750M Excel users.
Email Safety Net	Gmail warns “no subject line” but doesn’t prevent catastrophic mistakes like sending to wrong recipient or hostile tone. “Sent email to wrong person” gets 15K searches/month despite being rare—because the pain is extreme. A tool that says “You’re sending to john@competitor.com not john@company.com—confirm?” would be worth $50/mo to professionals.
Zoom Reliability Layer	Zoom shows “unstable connection” icon but doesn’t auto-fix anything. Users see the warning, panic, and manually troubleshoot while their meeting falls apart. “Zoom connection unstable” gets 20K searches/month. A tool that pre-tests connections, auto-switches to phone audio when WiFi fails, and keeps local backup recordings solves what Zoom only identifies.
Slack Channel Intelligence	Slack warns “@channel will notify 500 people” but doesn’t prevent sending to wrong channel—a daily mistake for power users. An extension that says “This looks like a question for #support not #random—switch?” or “This was answered 3 days ago in #engineering—see thread?” would save hours of duplicate work and embarrassment.