ExpertsFMThe Right Way To Write With AI
The Startup Ideas Podcast · September 21, 2026
5 Selected Lessons:
framework · 3 heuristic · 2
Point-of-view moat for SaaS
Source
The Startup Ideas Podcast — The Right Way To Write With AI (September 21, 2026) · at 3:03 ▸
When this applies
when building any product in a category where features alone can be copied
The takeaway
A moat comes from owning a point of view that drives product decisions, not from the features themselves.
Why it works
If the product encodes a specific methodology or point of view, competitors can't just copy features — they must first understand and adopt the underlying philosophy, which keeps them structurally behind the originator of that thinking.
One level up
Durable advantage comes from being the source of the thinking others must constantly reference, not from the artifacts they can copy.
Where it stops holding
This stops working when the product truly has no distinct methodology or opinion behind it, making it pure commodity functionality.
Detail
procedure:
- Identify a distinct methodology or point of view you believe in for your category.
- Build the product to execute and reinforce that specific method, not just generic tasks.
- Force every new feature decision to reference back to the point of view, so copying requires copying the thinking, not just the feature.
benchmarks: {}
rationale: Competitors copying surface features will always lag because they lack the underlying point of view driving decisions.
In their words
There's a ton of moat in SaaS and product and writing and marketing if you do have a point of view.
Timely vs. timeless content spectrum
Source
The Startup Ideas Podcast — The Right Way To Write With AI (September 21, 2026) · at 49:50 ▸
When this applies
when deciding what kind of content, product, or messaging to invest time in and what payoff timeline to expect
The takeaway
Every piece of content sits on a timely-timeless spectrum, and your effort and expectations should be matched to which end you're targeting.
Why it works
Content tied to a specific moment (news, trend, feature launch) gets a short burst of high relevance that collapses once the moment passes, whereas content built on durable, no-date-dependent ideas keeps compounding value because it can be reposted/reused indefinitely.
One level up
Set your effort investment and success expectations according to the actual decay rate of the asset you're creating, not a generic standard.
Where it stops holding
This stops working when the goal is explicitly short-term capitalization (e.g., a product launch moment) where timeliness is the entire point and longevity is irrelevant.
Detail
procedure:
- Before creating, ask where the piece sits on the timely-timeless spectrum.
- Match your time/effort investment to the expected decay: don't spend two weeks on something inherently 24-hour timely.
- Avoid embedding timestamp-dependent phrases ('just launched', 'in 2026') in content meant to be reused long-term.
- Set distribution expectations accordingly — a timeless piece won't get viral short-term traction like a timely meme, but will keep paying off for years.
benchmarks: {}
rationale: Timely content has high short-term/low long-term value; timeless content has low short-term/high long-term value — mismatched expectations cause frustration.
In their words
Timely content has very high short-term value and very low long-term value. Timeless content has very low short-term value and very high long-term value.
Voice as a bundle of word choice and sentence structure
Source
The Startup Ideas Podcast — The Right Way To Write With AI (September 21, 2026) · at 52:21 ▸
When this applies
when someone feels a piece of delegated or AI-generated output 'doesn't sound like them' but can't say why
The takeaway
Deconstruct 'voice' into concrete variables (word choice, sentence structure) so small, targeted edits can fix a piece rather than requiring a full rewrite.
Why it works
Perceived 'voice' is not a mystical whole but the sum of discrete, swappable variables (specific word choices, sentence length, structure); fixing one or two of these variables can flip the perception of the whole piece from 'not me' to 'exactly me.'
One level up
Intangible-seeming qualities (voice, taste, feel) are usually decomposable into a small set of concrete, adjustable variables.
Where it stops holding
This stops working when the mismatch is about substance/content (wrong facts, wrong argument) rather than style — in that case a wording tweak won't fix it.
Detail
procedure:
- When someone says content 'doesn't sound like them,' diagnose which specific variable is off: word choice or sentence structure/rhythm.
- Make a minimal, targeted change (e.g., swap one adjective, add attribution phrasing) rather than rewriting the whole piece.
- Verify the person's style preference (e.g., fact/citation-heavy vs. opinion-heavy) and align to that.
benchmarks: {}
rationale: In Coles's case, changing roughly 1% of the piece — adding two or three citation-style phrases — fully resolved the client's complaint.
In their words
I didn't rewrite the piece. All I did was go back in and add in like two or three sentences that just said, according to the New York Times or according to Harvard Business Review. I changed 1% of it.
Original thinking requires years of narrow focus, not volume
Source
The Startup Ideas Podcast — The Right Way To Write With AI (September 21, 2026) · at 43:38 ▸
When this applies
when anyone believes breadth of output, rather than depth of focus, is the path to a genuinely novel insight
The takeaway
Depth and duration of focus on a narrow topic, not volume of output or breadth of topics, is what produces original thinking.
Why it works
Original insight emerges from crystallizing a worldview on a narrow topic over a long period; spreading effort across many topics prevents the depth needed to reach anything beyond recombination of existing commodity ideas.
One level up
Breadth-seeking strategies trade away the depth required for true differentiation.
Where it stops holding
This stops working when the goal is distribution/reach rather than originality — breadth can still serve awareness goals even if it doesn't produce original ideas.
Detail
rule: Just because you personally typed a sentence doesn't mean you own or originated it — most writing is recombination of open-source ideas.
threshold: Original thinking appears only after years of crystallizing thought in a narrow topic, sometimes distilled into as few as six words.
rationale: Wanting to 'write about everything' deprives you of the depth needed to ever say something truly original.
In their words
Original thinking comes as the result of crystallizing the way that you see the world in a specific topic or a small number of topics for a prolonged period of time.
Voice is a minority component; most writing is commodity language
Source
The Startup Ideas Podcast — The Right Way To Write With AI (September 21, 2026) · at 56:34 ▸
When this applies
when someone avoids producing output by claiming an intangible 'style/fit' problem rather than engaging with the work
The takeaway
A vague complaint about 'voice' is often a proxy for fear of putting oneself out there, discoverable only by actually writing and reflecting, not by waiting for clarity in advance.
Why it works
Across hundreds of ghostwriting engagements, voice differences were mostly a small set of vocabulary tweaks layered on an otherwise standard structural style; people who complain vaguely about 'voice' usually can't articulate what they like or dislike, revealing the complaint is really fear of writing, not an actual style mismatch.
One level up
Vague dissatisfaction that resists articulation is often avoidance in disguise; the fix is action (doing the work) rather than more deliberation.
Where it stops holding
This stops working when someone genuinely CAN articulate specific stylistic preferences — in that case it's a real, addressable style mismatch, not avoidance.
Detail
rule: If someone can't articulate what they do or don't like about their own or someone else's voice, treat the complaint as likely fear-based avoidance rather than a real style issue.
threshold: Used the same authoritative, concise voice for the vast majority of 300+ ghostwriting clients and nearly all said 'this sounds like me.'
rationale: You can't discover your real stylistic preferences without writing a lot first — the preferences are an 'unmade data set' until you produce output.
In their words
I truly believe that 99% of the time when someone says that, they can neither articulate what makes their voice their voice, what it is they don't like, and they can't articulate what makes your voice your voice.