Building an AI-first venture studio aiming for $25 billion in value

Executive overview

Most founders treat AI as a feature to bolt onto existing products. The window to get ahead is closing fast — and those still waiting are already behind.

Dan Martell and his operating partner Roddy built Martell Ventures as an AI-first venture studio, launching 15+ companies in 13 months. The thesis: start every product from an AI-first premise, not a software product with AI added. Leadership, brand, and standards compound what AI cannot replace.

The companies that will survive are those that redesign around AI's purpose, not those that add AI to their processes.

Why AI is different this time

• Prior tech shifts (internet, mobile) created new job categories; AI will do the same — but the transition window is shorter.
• Coding is now English. Anyone who can speak can direct software.
• Teams that don't adopt AI will leave for companies that do — inaction is not a safe option.
• If you don't think AI-first before starting any task, you are voluntarily working at a fraction of your capacity.

How large language models actually work

• Every word in a prompt is tokenised into an index number pointing to an embedding table of vectors.
• Those vectors pass through a transformer, calculated against billions of pre-trained vectors, producing a new vector — the next word.
• Half a billion to a trillion mathematical operations produce a single token like "sure".
• Understanding this matters: you are directing a system with PhDs in every discipline, not a search engine.

The AI-first vs AI-added distinction

• Red ocean: take an existing SaaS product, add AI-powered features to the interface.
• Blue ocean (Martell Ventures approach): start from an AI-first premise — what would this solution look like if AI were the foundation?
• Example: Hero (recruiting platform) — no interface, no login. A phone call creates an account, posts the job, runs the ad, and handles agentic automation behind the scenes.
• The venture studio launched 15+ companies in 13 months using this model.

Director, not doer: the leadership shift AI demands

• Most professionals think AI can't disrupt them because they're skilled craftsmen — it already has.
• A CEO's purpose is not to type; it is to lead, vision-cast, and coach. Strip 90% of task work that doesn't serve that purpose.
• The shift: doer (doing everything yourself) → director (telling people what to do, scales to ~7 people) → designer (building systems that don't need you).
• A designer's presence feels additive, not necessary — that is the target state.

Building brand and media as a moat

• In a world where AI can spin up any SaaS product in seconds, the scarce resource is distribution and attention.
• Brand is the asset that precedes you before you walk into the room; AI cannot replicate it.
• Media compounds: invest heavily at the start, build momentum, then double down — results accelerate after the flywheel spins.
• Martel Media grew from 1 person to 15+ full-time over eight years using the doer → director → designer framework.

Sam's five rules for leading a team

1. Hard for the soul, training for the role — ask: can they do the work? Do they care? Are they a good person?
2. Default to trust — give new hires irresponsible amounts of trust, then inspect what you expect.
3. Train, don't tell — show what a high performer looks like before expecting it.
4. Measure what matters — pick a small number of North Star metrics (e.g. revenue, followers, revenue per follower).
5. Lighthouse, not tugboat — model the behaviour you want before demanding it from your team.

Todd's Builders Pyramid

• Standards are the greatest asset — not people. People are the evidence of your standards.
• Talent: hire better than yourself and develop those people.
• Strategy: strategies are a sacrifice, not a to-do list. Clarity on what to do implies clarity on what to stop.
• Dreams: the story and vision get people in the door before money can. Start with dreams, not numbers.

Martell Ventures operating model

• 13-month-old studio; 15+ portfolio companies at revenue.
• ~15 people on the operating side; one finance lead covering 12 companies simultaneously.
• Co-located team — physical proximity is a deliberate design choice for speed.
• Goal: $25 billion in enterprise value by helping SMB founders win with AI-first products.
• Core mission: become the global leader in AI-powered SMB innovation.

Default to action

• Momentum compounds — more results in the last two years than the prior eight.
• If something speaks to you, act immediately. Overanalysis kills compounding.
• The number one thing that changes your life is the environment you spend time in.