Hi friend,

Watching this conversation, one thing becomes obvious pretty quickly:

Ramp is not treating AI like a software rollout.

They are treating it like a new operating system for the company.

That sounds dramatic, but the examples in the video are surprisingly practical. Six weeks in, the host says 12% of production code was coming from people who were not engineers. Ramp also shared data suggesting businesses that heavily use AI grew revenue at 27% per year, compared with 3% for companies that did not spend on AI. Whether those numbers hold everywhere is a separate question, but the direction is hard to ignore: the companies getting real leverage from AI are not just buying licenses. They are redesigning how work gets done.

Here is the useful version:

AI adoption is not about getting people to use tools. It is about helping people redesign their own work.

That was the real lesson from Diego Zack, Ramp’s design leader.

Not “which model are they using?”

Not “which tool should everyone install?”

The deeper question was:

What does a company become when everyone can build?

1 idea worth sitting with

Most companies are still asking:

How do we get our team to use AI?

Ramp seems to be asking a better question:

How do we make every person more capable of turning ideas into working systems?

That difference matters.

Using AI means someone opens ChatGPT, asks for help, maybe saves a bit of time, then goes back to the same workflow.

Building with AI means someone looks at a repetitive task and asks:

Why does this job still need to be done this way?

That shift showed up everywhere in the conversation.

Ramp’s design team is not just designing screens anymore. They are building design infrastructure so that more people across the company can create useful, brand-safe, customer-safe experiences without needing a designer in every loop. The goal is not to remove designers. The goal is to let designers spend more time on the work where taste, emotion, judgement, and quality actually matter.

That is a very different view of design.

Design becomes less of a gatekeeping function and more of an enabling system.

1 product/design lesson

The strongest product lesson from the video was this:

The future of design is not just designing interfaces. It is designing the conditions for good work to happen.

Ramp’s internal tools seem to follow this pattern.

One example was Glass, an internal tool that packages access to models, files, company context, Google Workspace, Notion, Slack, Snowflake, Datadog, and other internal systems so employees do not need to learn terminal commands, permissions, or complex setup before getting value. The point was to collapse the setup cost so people could start working with AI immediately.

That is the part many teams miss.

They give people the tool, but not the harness.

They give people access, but not context.

They give people possibility, but not a clear first win.

Ramp’s approach seems closer to:

Access -> Context -> Workflow -> First win -> Peer sharing -> Adoption

That first win matters.

In one story, Diego sat with an account manager who was spending 30-40 minutes after meetings writing follow-up emails. They opened Claude together, worked through the task, and within roughly 40 minutes had automated the summary email in the person’s own style. That became the “whoa” moment: the point where AI stopped being abstract and became personally useful.

That is probably the adoption pattern hiding in plain sight.

People do not adopt AI because leadership says it is important.

They adopt it when it removes a real pain from their actual day.

1 founder tradeoff

The exciting version of this story is:

Everyone becomes a builder.

The uncomfortable version is:

Everyone now has to rethink what their job actually is.

That is a real tradeoff.

Ramp’s AI fluency ladder is useful here. Diego described a progression that roughly looks like this:

Level 1: Use AI to answer questions
Level 2: Use AI to write or prototype things
Level 3: Build your own tools
Level 4: Build agents that build and use tools to do parts of your job

The biggest leap is not from level one to level two.

It is from “AI helps me do my work” to “AI does some of the work, and I design the system around it.”

That is where the role changes.

For designers, it means the craft shifts from producing every screen to shaping the product’s feel, quality, emotional tone, taste, and system-level coherence.

For managers, it means they cannot just manage the work. They need to stay close enough to the craft to show what good looks like.

For researchers, it means a tiny team can create systems that let hundreds of people ask better questions, find the right customers, run sessions, and feed insights back into shared knowledge. Ramp described a three-person research team enabling broad daily research across the company, with around 200-250 research sessions in a month.

The tradeoff:

AI gives teams more leverage.

But leverage without taste creates noise.

Leverage without judgement creates slop.

Leverage without trust creates chaos.

The companies that win probably will not be the ones with the most AI tools.

They will be the ones with the clearest taste, strongest internal systems, and fastest path from idea to useful proof.

The quiet lesson: taste becomes more important, not less

One of the most interesting moments was Diego talking about codifying his taste.

Not in a vague “make it premium” way.

In a practical way: spacing, hierarchy, eyebrows, titles, grouping, pixels, visual rhythm, emotional tone.

That is a big idea.

As AI makes execution cheaper, taste becomes more valuable.

Because when anyone can generate something, the scarce skill becomes knowing:

  • what is worth making
  • what good looks like
  • what should be removed
  • where the emotional detail matters
  • when polish is helping the customer
  • when polish is protecting the maker’s ego

Diego said something especially useful about customer-facing work: early alpha can be wide open, and teams should be willing to embarrass themselves in front of friendly customers quickly. The lesson was that over-polishing a misguided idea is often more about protecting the designer than helping the customer.

That one stings a little.

But it is useful.

For a solo founder, this is the line to remember:

Polish after proof. Not before.

What this means for builders

The practical takeaway is not “go buy more AI tools.”

It is this:

Pick one workflow that hurts, and redesign it with AI until it produces a real first win.

Not a demo.

Not a prompt experiment.

Not a shiny prototype.

A real win.

Something like:

Before:
This task takes 40 minutes and drains energy.

After:
This task takes 5 minutes, keeps the quality bar, and gives the human more space for judgement.

That is the unit of progress.

One painful workflow.

One useful harness.

One repeatable system.

One person who can now do higher-quality work with less drag.

Then that person teaches the next person.

That is how adoption compounds.

A practical experiment for this week

Choose one recurring task and run this tiny audit:

AI Workflow Audit

1. What task keeps repeating?
Example: writing follow-up notes, summarising research, creating first-draft specs, preparing meeting briefs.

2. What part actually needs human judgement?
Example: prioritisation, tone, customer empathy, final decision, taste.

3. What part is mechanical?
Example: formatting, summarising, searching, comparing, drafting, tagging.

4. What context does AI need to do this well?
Example: previous examples, brand voice, customer notes, product docs, decision rules.

5. What would a first win look like?
Example: 30 minutes saved, better consistency, fewer missed follow-ups, faster research synthesis.

6. How could this become a reusable harness?
Example: prompt template, Slack bot, Cursor command, Notion workflow, internal agent, checklist.

The goal is not to automate everything.

The goal is to stop treating AI like a smarter search box and start treating it like a design material.

1 question

Where in your work are you still asking a human to behave like a machine?

That might be the best place to start.

Because the real opportunity is not just faster output.

It is better-designed work.

And maybe that is the bigger lesson from Ramp:

The AI-native company is not the company with the most tools. It is the company where people keep redesigning the work itself.

Big love,
CJ-one