What Anthropic's Head of Growth Does With AI All Day (and the Agency Version of Every Move)

Anthropic went from $1 billion to over $19 billion in ARR in 14 months, the fastest revenue run in software history, and its head of growth, Amol Avasare, just gave his first public interview on Lenny's Podcast explaining how he actually works. Strip away the scale and three moves stand out, because they're not really AI tips. They're an operating pattern, and it happens to be the same pattern we bet Space on. Here are the three moves, and what each one looks like when an agency runs it for clients.

Move one: prototype the idea, skip the document

Avasare says 60 to 80 percent of what his team ships has no PRD at all. Small things run on Slack; when he wants to articulate an idea, he prototypes it and shows it rather than writing the memo about it. Documentation survives only for the projects genuinely large enough to deserve it.

The principle underneath: a document describes a decision, a prototype is one. Most documentation in most companies is a coordination tax, written to align people who could have aligned faster around a working thing. We run the same rule at Space, published as policy: if the honest version of a deliverable is one page, the client gets one page, and the sixty slides never get made. Our sprints jump to the working thing (positioning tested against simulated audiences in week one, sites in build while identity develops), and even our paperwork follows the pattern: the SOW generator drafts a complete statement of work from the conversations themselves, in minutes, so the document catches up to the decision instead of gating it.

Move two: machines watch the dashboards, humans get the verdict

Avasare has a scheduled AI task that scans 20 to 25 charts every morning and tells him only what deserves attention: what's concerning, what's interesting, what to ignore. He looks at a couple of charts for pleasure and delegates the long tail entirely, trusting the system more as its false positives fall.

The principle: senior attention is the scarcest resource in any company, and spending it on surveillance is malpractice when machines monitor tirelessly and humans decide better. Space runs this pattern outward, for clients, and has for over two years: our competitive intelligence systems monitor every client's press, their competitors' moves, and every page of competitor websites continuously, then deliver the analyzed verdict (what changed, what it means, what to do) into the client's Slack, usually before the client has seen the news. Internally our PM Co-Pilot does the same for project noise and our Strategic Radar for the AI market itself. Nobody at Space opens dashboards for a living. The machines watch; the seniors move.

Move three: simulate the perspective you can't constantly ask

The move Avasare saved for last is the one worth stealing first: every week, he asks AI to review his work through a simulation of his manager, built from her public writing and their working context, and tell him where he's dropping the ball. Feedback that used to arrive quarterly, if ever, now arrives weekly, from a perspective he can't reasonably request that often.

We'd go further: this is the single most underused AI capability in business, and it's the one we've productized hardest. Space's audience simulation runs the same play at brand scale, putting rebrands, announcements, and campaigns in front of hundreds of modeled psychographic profiles before anything irreversible ships, because the market is the manager you really can't ask twice. And the Taste Decoder does it for the client relationship, extracting the unspoken principles behind a decision-maker's approvals so the creative team can consult their judgment without booking their calendar. The pattern in all three: borrow the perspective before the stakes arrive.

The two lines that could have come from our site

Two moments in the episode stopped us cold. Asked how Anthropic won as the smallest, least-funded player, Avasare credits "freedom through constraints": no cash, no distribution, no first-mover advantage, therefore total focus. And describing growth philosophy, he argues the biggest mistake practitioners make is squeezing every last dollar, that great companies stay comfortable leaving money on the table to protect trust, brand, and quality. We've published both convictions as our own operating theses, learned across six challenger brands: constraints create greatness, and customers love brands that visibly bet against themselves. Watching the fastest-growing company in history run on the same two beliefs is the most encouraging thing we've read all year. The pattern isn't Anthropic's or ours; it's what the environment now rewards.

What to steal this week

Three moves, each installable in a day. Schedule a morning AI scan of every dashboard you're supposed to watch, and let it tell you what matters. Prototype your next initiative instead of writing the memo, and see what the meeting is like when there's a thing on the screen. And ask an AI, loaded with your goals and your last two weeks of output, to critique you as your toughest board member or biggest customer. If you want the versions of these that run at brand scale, with senior operators attached, that's the business we're in.