Multiple agents or one smart prompt?

For a complex AI use-case, there are currently two schools of thought...

School 1: You need to spend a lot of time and thinking (and investment) in creating a chain of specialist agents that each do a different job.

School 2: Just hit the smartest, best chain-of-reasoning model with an amazing one-shot prompt and let it do the job.

I think, wherever possible, you want to attempt to go with path 2.

Why?

Because the first approach gives you fine-grained control and observability.

However, the second approach allows me to bet that...

a) These billion-dollar frontier model companies will create smarter and smarter models that just get the job done without babysitting.

b) A model with full context of what it's done and what it's planning to do next is going to do a better job than different agents trying to infer intent and picking up the ball from each other

What do you think?

The future of UI powered by AI

Increasingly, apps will need to be "dual modality."

They will need a structured UX/UI with buttons and dials, as well as a chat interface for conversational change.

The real trick is how those two interface models co-exist and - most interestingly - how they cross over.

For example,

1. How little, strategic, and elegant affordances appear inside the structured UI to suggest "AI Intervention",

2. How little moments of structured interaction will surface inside chat to help users fast-forward to the right answer without long-form typing.

I've been working hard on this in multiple apps with my startup portfolio.

If you represent a funded startup or enterprise that wants to learn more, drop a comment below, and we can chat.

A product is not what you think it is

A product is not what you think it is.

A product is not a piece of software you make.

A product is a packaged, repeatable, scalable, predictable, marketable, monetizable unit of value delivered to market.

It may or may not require custom software.

It always involves the orchestration of multiple skills to deliver scalable solutions to people.

Costs don't matter - ignore them.

While you always want to manage your burn, your primary goal as a startup or scaleup is to geometrically accelerate your growth so that costs become rounding errors compared to usage, revenue, and, eventually, profit.

No, I won't sign your NDA

Every now and then, I am asked to sign an NDA before reviewing some of the initial details of a startup that has reached out to me.

Like an angel investor or VC, I do not sign NDAs to discuss your startup.

Here's a quick summary of why I stick to this:

1. Ideas aren't unique; execution is. My focus will be on the 10,000 decisions required for execution, not just the initial idea.

2. Avoids conflicts. I meet with many founders, and I may have already heard similar ideas (or will hear them in the future). Signing an NDA creates unnecessary legal risk that I might be accused of sharing information, even if I haven't.

3. Reduces friction. It eliminates paperwork and administrative friction that slows down initial discovery.

4. Filters for quality. It helps filter for founders who understand that the primary value is in execution and who are coachable.

That said, I do, of course, always act in a professional manner, treat the information I review as strictly confidential, and use it only in the best interests of the company and founders, even without a formal NDA in place. 

Of course, my formal engagement contract includes a confidentiality clause, since we're dealing with very important lower-level details.

Leader code switching

A big part of senior leadership is knowing how to "Code Switch".

Code switching is temporarily changing the way you speak (tone, intonation, word choice - even accent) to adjust to different people, personas, and personalities in your life.

For example, you may have heard people do this when talking to others of their culture - suddenly their accent deepens, or they use different words!

You probably do it when talking to your family vs talking to colleagues at work.

It's natural and typically happens entirely subconsciously.

Being a leader in a work environment requires the same skill. However, it typically takes more intentionality.

As a leader, you need to know what to say, how to say it, and when to say it to various audiences and forums in your business. The scenery doesn't change very much. The people are very similar, but the circumstances often require a very different 'code' or style for maximum success.

It's a key part of sending the right message and gently and effectively guiding action and morale.

AI has replaced me!

One of the product leaders I work closely with as a Fractional CPO has trained an AI Agent on all my public posts on product, leadership, start-ups, etc.

Now he checks all his work with it before running it by me.

Has he made me redundant, or has he made himself indispensable?

I still have a much better sense of humor (and incisive taste and judgment) that it can't match, though.

I still think I win this round.

How about you? Have you been replaced, or made yourself indispensable?

Did you stumble into a great new business?

If one or more of these things meaningfully change, it's a different business.

• Product

• Business model

• Target customer

• Geography

Don't do it casually or unintentionally. Each business requires a meaningful new investment in sales, marketing, customer support, etc.

This, of course, needs significant new capital/cash flow.

Non-verbal signals that are working (for or) against you

Rightly or wrongly, you can quickly (and detrimentally) communicate your seniority with your personality, word choice, and general confidence in any given room.

People can try to look past it to hear the substance of what you're saying, but the fact is, it's very, very hard and takes enormous intention and cognitive load.

Don't leave it to the other person/people to make up their own mind. Own the signal.

What signal are you sending?

What signals do you look for?

Effective M&A Strategy

What if you had unlimited money to buy companies to accelerate your business? How should you think about M&A?

That’s a "kid in a candy store" style question, right?

First, you would want to map out the entire business architecture and strategy for the next 12 to 24 months.

Then you want to ask the question "which companies might help us accelerate the execution of this strategy?"

But of course, you must factor in the enormous cost and complexity of integrating different tech stacks, cultures, and processes. Successful M&A is hard, costly, and disruptive.

In order to qualify as a candidate to meaningfully accelerate part of the business strategy, the target company would need to have a massive unfair advantage in the particular category of the business you're trying to accelerate.

OR

You should really only be considering companies you can strip for parts without planning to integrate their legacy tech, culture, structure etc.

Which parts?

A "part" might be “the team”. For example, you could buy a small amazing little consumer product startup and turn them into one of your cross-functional product squad to work on a specific product mission/problem.

A “part" might just be something like their customers. Take their database, migrate some percentage over to your core products (assume only a conservative fraction), and kill the rest of the business.

Ruthless? Yes.

Efficient. Also yes.

Everything else quickly becomes a quagmire.

AI supercharges whoever you already are

AI has a wonderful way of supercharging your superpowers and your weaknesses.

If you tend to talk a big game and lack rigor in your thinking, then AI will be a great venue for you to talk/think endlessly and create mountains of AI slop.

That's why AI tools require top 1% people to produce high-quality 100x outcomes.

Measuring KPIs and OKRs when things are fundamentally broken is DUMB.

Measuring KPIs and OKRs when things are fundamentally broken is DUMB.

Why?

Because it's like asking what the score is when the players are in the wrong ballpark.

Get to the right ballpark, start playing the right game, and then you can start to ask what the numbers are.

Alignment is the engine that powers disruption and success.

How do you know if your startup or scale-up is working well?

Everyone is aligned and working on basically the exact same thing for the exact same reason - at the same time.

Alignment is the engine that powers disruption and success.

Agent management and orchestration

Every management job description will soon include “agent management and orchestration” as a core skill

SaaS vs. AI

Running a SaaS company?

Your pitch must be something along the lines of...

"AI lets you ask questions. We give you answers to questions you didn’t even know were worth asking."

You must also have a comparison chart that justifies your existence beyond AI chatbots and AI-generated software.

I've developed extensive thinking on this for my portfolio of startups.

Reach out if you'd like to get access

Stop spinning plates and start landing planes

Value creation may not be where you think it is

You (and your manager) should not be measuring your value creation by how many plates you have spinning in the air, but by how many high-quality, needle-moving deliverables you land FAST.

1 high-quality deliverable that moves the needle for the business is worth 100x 5 plates spinning in the air.

The same is true for your Startup.

If your users/customers don't feel it, chances are high it's worth ZERO.

Deliverable quality: Great teams vs. Mediocre teams

Mediocre teams: Along every link in the chain, the clarity of requirements and the quality of the outcomes degrade.

With teams like this, you can work to define every single little detail - every pixel, every non-functional requirement, every edge case, etc. And even then, the deliverable will come back broken.

Great teams: Along every link in the chain, the clarity of requirements and the quality of the outcomes improve.

With teams like this, you only have to point in the vague direction of where you want to go, and every pixel, every non-functional requirement, every edge case, etc., is handled even better than you could have imagined.

Why?

Partly because of competence.

Partly because of mindset.

In great teams, each person isn't just passing along a message. They don't believe their job is just to "pass the info along."

Instead, each person is a skilled craftsman who believes, "It's my job to polish and refine this using my particular skills and perspective.

This can be the difference between high-quality outcomes, fast. Or low-quality outcomes, slow.

And of course, this is often the difference between success and failure.

AI makes strong product craft more important than ever

AI development is to product development what CGI & CGI Cameras are to directing films.

When you can do "anything," the real question becomes "how do you tastefully make choices that tell a cohesive story?"

In modern cinema, where many shots are entirely CGI, you can place the camera anywhere and move it in any way you like. No limits.

So many mediocre directors misuse this freedom to dramatically move the camera all over the place, failing to give a sense of scale, proportion, motivation, and emotional journey.

Great directors understand that placing and moving the camera takes taste, judgement, and intentionality. They also understand that, often, the best answer is restraint and obeying the laws of physics.

See Pacific Rim 1 vs. Pacific Rim 2.

Great product management is the same.

When engineers can build anything, FAST, the real question (and challenge) becomes WHAT should you build and HOW should it fit into a cohesive product story to drive utility and delight.

In other words: Taste, judgement, and senior product craft become more valuable than ever.

AI can work against shared understanding

The problem with AI-generated analytics that allow anyone to create any insight at any time means that very few people across the business are looking at the same story at the same time.

This is true of AI interfaces and tools across many applications.

Consistency and shared mental models are just as important as AI-powered flexibility in everyone's hands.

Be careful about power undermining shared narratives and mental models.

This is a concern for both operating a business and designing great product experiences.

The Silicon Valley Playbook

This is the Silicon Valley playbook.

1. Sell the dream (Fundraise)​

Raise money on a deck or an early MVP - invest ahead of value, growth, revenue, and profit.

2. Massive Value Creation 

Disrupt the status quo by making something 10-100x cheaper, faster, and/or easier than before. Often, this involves redefining the old rules by using software/internet efficiencies and cutting out incumbent players and processes.

3. Fast-forward to massive scale

Make things standardized and repeatable. Avoid services/high-touch processes. Execute fast and hard to grow really big, really fast. If you have to (and can afford to), subsidize the cost of your product to accelerate adoption.

4. Own the category

Use flywheels (self-reinforcing loops at scale) to build an unbeatable moat with ever-increasing value.

5. Capture a piece of the (massive) newly created value

This is when serious Silicon Valley-style companies seriously start thinking about breakeven (much less profit). Not before.

This is why they win while other players are capital, ambition, and/or execution starved and have small exits or die on the vine.