How to measure AI adoption in your company
Plus: The principles to build successful ventures
· 578 words
Christian Woese & Melanie Gabriel
April 07, 2026 • Read time: 3 minutes
Dear hustlers, founders, operators and visionaries,
Having a bad day? Can’t be as bad as Anthropic’s, who accidentally leaked 512,000 lines of proprietary code through a single debug file. Or Oracle’s, who told somewhere between 20,000 and 30,000 employees via a 6AM email that their jobs no longer exist. AI and data centers need the budget now. Follow the Gradient and stay tuned.
PS: Has this e-mail been forwarded to you? Sign up here.
Your talk track for this week’s cocktail party
Finnish Oura interviews banks for IPO as soon as this year
The health ring maker is exploring a public listing at an $11bn valuation after raising $875m in its Series E last year. Revenue doubled from $500m in 2024 to over $1bn in 2025, with $1.5bn expected this year. They have sold over 5.5m rings sold worldwide.
Revolut fined €11.5m in Italy over misleading fee claims
Italy's antitrust authority hit Revolut with an €11.5m fine for misleading "0% commission" investment ads that obscured the risks and limitations of fractional shares. The regulator also flagged aggressive practices in managing account suspensions, preventing users from accessing their funds. Revolut disputes the findings and plans to appeal.
What do you meme?
Food for thought
AI startups are growing faster than ever. At lower cost?
AI-native applications are hitting record growth with fewer and fewer employees. But this isn't lean efficiency. Operational costs are simply shifting from payroll to compute. If everyone's buying the same tokens, differentiation comes from distribution, data loops, or workflow lock-in, not from having a better team.
We talked to Max Buckley of Exa (Youtube / Spotify / Apple) for an exclusive look inside an AI-first company seeking to beat Google, and to Andreas Goeldi of b2venture (Youtube / Spotify / Apple) about his take on how the service-as-software shift connects directly to this payroll-to-token swap.
Eight principles for building a company?
a16z distilled Elon Musk's operating playbook: question every requirement, delete before you optimize, shorten timelines, keep design close to production, hire for attitude. Critics argue this only works with near-infinite capital, while defenders call it a decision-making framework for the AI age.
Smart hack
From our partners
One brand built 30+ landing pages through Viktor without a single developer.
Each page mapped to a specific ad group. All deployed within hours. Viktor wrote the code and shipped every one from a Slack message.
That same team has Viktor monitoring ad accounts across the portfolio and posting performance briefs before the day starts. One colleague. Always on. Across every account.
5,700+ teams. 3,000+ integrations.
From us
Should companies measure how many AI tokens employees burn? Meta, OpenAI, and Shopify already do, complete with internal leaderboards ranking individual consumption. We appreciated the Inside Look of Victoria from ElevenLabs on her take-away’s when doing this:
It's a crude proxy. One engineer may have burned billions of tokens. That tells you nothing about what they built.
Zero-usage outliers matter more. Check for people who are not using AI at all. They might be stuck or missing tools entirely.
Track at team level, not individual. Leaderboards invite gaming. Think lines-of-code metrics from the 90s.
Ask top users what they built. Make sure you understand outcomes over consumption.