Husband • Dad • Developer Building with @cursor_ai, @Grok and @Bot
Joined January 2012
- Tweets2.5K
- Following1.8K
- Followers952K
- Likes1.5K
If you're launching a new feature, don't measure success by how many people clicked the announcement.
Measure whether the intended users discovered it, tried it and received the expected benefit.
A feature can generate plenty of curiosity without changing anyone's behaviour.
The important question isn't whether people noticed the launch.
It's whether the feature made their work better.
A useful framework for deciding whether to build or buy a feature:
Build it when it differentiates your product or solves a problem central to your customers.
Buy or integrate when the capability is common, well served by existing tools and not a reason people choose you.
Your time is limited.
Spend it where your product can become meaningfully better than the alternatives.
If customers keep asking whether your product integrates with another tool, don't assume the integration itself is the whole opportunity.
Gary O’Neill retweeted
A product demo should answer five questions in this order:
1. What problem does this solve?
2. Who experiences that problem?
3. How does the product solve it?
4. What changes after using it?
5. What does the customer need to do next?
If your demo starts with 15 minutes of navigation, you may be making people work too hard to understand the value.
Gary O’Neill retweeted
A simple way to make a SaaS landing page more convincing:
Replace vague claims with evidence.
Instead of "Save time with automation", show the task being automated.
Instead of "Easy to use", show the actual workflow.
Instead of "Built for teams", explain who uses it and what they accomplish.
Specificity makes it easier for a buyer to judge whether the product is relevant.
Gary O’Neill retweeted
Your changelog can be a marketing channel.
Instead of writing:
"Improved reporting."
Write:
"Reports now show which jobs are overdue, so managers can spot delays without checking each record individually."
Explain the change in terms of the customer's working day.
The feature is the mechanism. The improvement is the story.
A useful way to test whether an automation deserves to exist:
Calculate the annual cost of the manual process.
Include the frequency of the task, the time spent and the cost of correcting mistakes.
Then estimate the build cost, ongoing maintenance and remaining human oversight.
Sometimes the numbers justify automation immediately.
Sometimes a better checklist is the smarter solution.
Both are good outcomes.
Before rebuilding a messy codebase, identify which problems are causing real damage.
Are deployments unreliable?
Are bugs recurring?
Is every new feature taking longer?
Are developers afraid to change certain areas?
Refactor around demonstrated problems rather than aesthetic discomfort.
Clean code has value, but the best refactoring priorities are connected to outcomes you can actually improve.
So Anthropic's biggest bill is basically going to Elon. What a timeline!
Anthropic has officially revealed in its IPO filing today that the company expects to spend up to $120.5 billion on @SpaceX SI compute through 2029.
• Monthly payments up to $3.1B, or $37.2B annually.
• SpaceX is expected to deliver all contracted computing capacity by Dec 2026.
• Anthropic (and SpaceX) can terminate the agreements with 90 days' notice after the initial contract periods end in early 2027.
Here's the exact wording directly from the filing: "Between May and October 2026, we entered into and amended several agreements with SpaceXAI to access compute powered by Nvidia GPUs. Under the SpaceXAI arrangements, we agreed to spend an aggregate of up to $120.5 billion over terms extending into 2029, consisting of approximately $3.1 billion in the aggregate per month once all capacity has been delivered, which we expect in December 2026, with different fees during capacity ramping periods. Following initial non-cancelable periods ending in January 2027 and March 2027 under the respective agreements, each agreement may be cancelled by either party upon 90 days’ advance written notice."
An experiment for your next landing page:
Remove every sentence that could appear on a competitor's website without changing a single word.
"Powerful platform."
"Streamline your workflow."
"Save time and money."
Replace those phrases with details about the actual problem, the actual user and the actual outcome.
If your copy could describe any product, it probably isn't explaining yours well enough.
A useful rule for indie builders:
Never let a successful launch convince you that you've found a reliable acquisition channel.
A launch creates a spike of attention.
A repeatable channel brings qualified people to your product consistently.
Track where paying customers actually come from, not just where sign-ups came from.
The distinction matters when the initial excitement disappears.
If you're unsure which feature to build next, make a list of the last 10 customer problems you've heard about.
For each one, record:
• How many customers mentioned it
• How frequently it occurs
• How severe the consequences are
• Whether the customer has a workaround
This gives you a more defensible priority list than choosing whichever idea sounds most exciting.
One of the most useful questions to ask a customer who cancels:
"What would need to be different for you to consider using us again?"
Don't argue with the answer.
Don't immediately offer a discount.
Look for patterns across multiple cancellations.
A single response is an anecdote. Repeated reasons can reveal a problem worth fixing.
A good way to find your first B2B customers without running adverts:
Choose one narrow type of business.
Find 20 companies that fit.
Research how they currently handle the problem you solve.
Contact them individually with a relevant observation and a simple question.
Don't send the same generic pitch to everyone.
The aim is to start a useful conversation, not win a bulk-email competition.
A product can have excellent retention for the wrong reason.
Sometimes customers keep paying because moving their data elsewhere is difficult.
That isn't the same as being genuinely loved.
Ask existing customers what they would miss if they left.
Their answers can tell you whether you've built real value or merely created switching costs.