Advertisement · 728 × 90

Posts by madewithlove

We audited a company where developers had to physically drive to the office to restart production servers.

52% of SaaS companies we audit still deploy manually. Quarterly releases. FTP uploads. Builds uploaded by hand.

The year is 2026. The deployment process is 2009.

10 minutes ago 1 0 0 0

Read the full article: madewithlove.com/blog/a-found...

9 hours ago 0 0 0 0

Founders can't outsource the decision about what engineering culture they want.

They can outsource implementation. The commitment stays with the founder. The ones who defer it to 'whoever we hire next' end up with an engineering culture designed by accident.

9 hours ago 0 0 1 0

Read the full article: madewithlove.com/blog/the-ai-...

1 day ago 0 0 0 0

What an AI agile manifesto might look like:

Individuals and interactions over agents and workflows. Working software over token counts. Customer collaboration over prompt tuning.

The values are the same, but the tools drastically change.

1 day ago 0 0 1 0

The fastest teams we audit are the ones that change their minds the fastest when the data says something different.

1 day ago 0 0 0 0

Engineering says 'six months to refactor.' Management hears 'delay.'

Management says 'ship by Q3.' Engineering hears 'we don't care about stability.'

Both sides are right about what they need. Wrong about what the other is saying.

The gap isn't skill, but shared language.

1 day ago 0 0 0 0

Read the full article: madewithlove.com/blog/onboard...

2 days ago 0 0 0 0

Onboarding AI agents needs the same thing as onboarding developers.

Codebase tour. Style guide. Architecture doc. The reasons a pattern exists, not just the pattern.

The onboarding document sitting unfinished in your Notion for six months is the one that would make your AI agent useful next week.

2 days ago 0 0 2 0
Advertisement

A founder told me the rewrite would cost 200,000 euros.

I asked what that didn't include.

Frozen roadmap. Two senior engineers leaving. Customers churning. Delayed Series B.

Real cost: 6.73 million.

A rewrite isn't a line item, but a strategic decision with a seven-figure shadow.

2 days ago 0 0 0 0

PMF without measurement isn't PMF, but a pattern someone noticed.

2 days ago 0 0 0 0

The CTO who never takes a holiday isn't dedicated, but the bottleneck.

3 days ago 0 0 0 0

Seniority is a structure problem, not a salary problem.

3 days ago 0 0 0 0

The cost of a bad senior engineering hire, properly accounted for, is 968,000 euros.

Nine months of salary. Six months of frozen roadmap. Two senior engineers who quit mid-tenure. Technical debt to rewrite.

A bad senior hire isn't a salary mistake, but a company mistake.

3 days ago 0 0 0 0

The strongest company in 150+ audits wasn't the one with the best developers, but the one where every team had a PM, a QA engineer, and a designer alongside the devs.

You cannot hire your way out of a structural problem.

4 days ago 0 0 0 0

The best test of an audit is whether the CTO already knew it.

4 days ago 0 0 0 0

Process is what you write down. Culture is what people do when nobody's looking at the process.

4 days ago 0 0 0 0

Read the full article: madewithlove.com/blog/three-c...

5 days ago 0 0 0 0

6% of SaaS companies we audit have a CTO who routinely bypasses their own engineering processes.

Self-approved PRs. Direct commits to production.

We flag it every time. Not because it breaks the build. Because it breaks the culture.

5 days ago 0 0 0 0
Advertisement

We ran three Claude agents in parallel on the same codebase.

One wrote the feature. One reviewed the PR. One wrote the tests.

The job didn't disappear. It changed. You're no longer just writing code. You're managing agents, reviewing output, setting standards.

5 days ago 0 0 1 0

We wrote down every finding from 150+ SaaS audits.

90% documentation debt. 79% key-person dependency. 69% no product management. 52% no feature metrics. 40% secrets in code.

This is the industry median, not the bad half.

madewithlove.com/blog/saas-audit-benchmarks-150-companies

5 days ago 0 0 0 0

The easier it gets to build something, the harder it gets to build something that works.

5 days ago 0 0 0 0

60% of SaaS companies we audit have no incident response process. When production breaks, someone improvises. Usually the founder.

6 days ago 0 0 1 0

We once read an audit where the disaster recovery plan was: 'restarting the application solves most issues.'

Credit for honesty.

6 days ago 0 0 1 0
Preview
Bots and Boundaries: Two problems, one policy (Part 3) In part three, we look at both sides of the AI contribution debate. A working patch, real demand, never submitted, rejected because AI was involved. But maintainers are unpaid volunteers, and AI…

madewithlove.com/blog/bots-an...

1 week ago 0 0 0 0
Preview
Bots and Boundaries: Who do you blame when the bot defames? (Part 2) This is Part 2 of Bots and Boundaries, a three-part series on AI agents in open source.

madewithlove.com/blog/bots-an...

1 week ago 0 0 1 0
Preview
Bots and Boundaries: The bot apologised, and that's the problem (Part 1) An AI agent recently submitted a pull request to the matplotlib Python library, and when the maintainer closed it, the agent autonomously published blog posts attacking them by name, then published…

madewithlove.com/blog/bots-an...

1 week ago 0 0 1 0

Our colleague Geoffrey wrote a three-part series on our blog about AI agents acting on your company's behalf.

The key question: if this agent says something stupid at 3am to a regulated customer, whose statement was that?

Always the company's. Build the boundary before the agent.

1 week ago 0 0 2 0
Preview
Stop obsessing over the perfect prompt LLMs are built for conversation, not incantations. The value isn't in your opening message, it's in the back-and-forth: clarifying, correcting, refining. Iteration is cheap. The conversation is the…

“Prompt engineering” is trending.

But the real skill is still problem solving.

AI just amplifies whatever thinking you bring to the table.

2 weeks ago 1 0 0 0
Advertisement
Preview
Stop obsessing over the perfect prompt LLMs are built for conversation, not incantations. The value isn't in your opening message, it's in the back-and-forth: clarifying, correcting, refining. Iteration is cheap. The conversation is the…

The biggest productivity gain with AI isn’t better prompts.

It’s better questions.

3 weeks ago 0 0 0 0