Platforms

Replit's agent swarm rebuilt its monitoring platform in two days for $3,000

Replit's agent swarms rebuilt licensed monitoring software for $3,000 in two days. The receipt is real, but it comes from the company's own benchmark table, and the swarm arithmetic shows the last agents bought the least.

Vincent JiangVincent Jiangthree-minute read
Share
Amjad Masad, co-founder and chief executive of Replit, photographed against a black background
Amjad Masad, Replit's co-founder and chief executive. The company pointed to its own benchmark table as the receipt for Missions, the agent-swarm product it launched on November 17.

Replit sold the receipt with the product

Replit$9.40B — Replit, private, latest valuation $9.40B put a new product on sale on November 17: Missions, a tool where a customer sets an outcome, a spending cap and the actions that need human approval, and a swarm of AI agents plans, builds, tests and iterates toward the goal 1. The company pointed at its own receipt.

It gave Missions the goal of replacing the monitoring platform it rented, and within a $3,000 budget a swarm built DataCat in two days, with dashboards, alerts and application tracing 1. Replit is open-sourcing a self-hostable version today 1, though The New Stack reported the public release still awaits further reliability work and a code audit 2.

The receipts, all from the company

Against a planted database bug, DataCat raised an alert in 38 seconds against a 60-second target, and a three-hour load test stored 26 million activity records and 4.4 million log entries with zero data loss, with 51 of 51 correctness checks passing 1. In matched Node.js tests it handled 39% more requests with 51% less monitoring overhead than the platform it replaced 1.

Against a monitoring platform built with the same prompt in Claude Code, DataCat took in 49,000 traces a second against 9,800, five times faster, and its agent ran at 82 to 108 MB of memory under load against 319 MB for the incumbent's agent 1.

More agents bought less: 17x at 16 agents, only 19x at 32

Doubling the swarm from 16 to 32 agents bought only 11% more speedup

0x5x10x15x20x1 agent4x16 agents17x32 agents19x
Data
Value
1 agent4x
16 agents17x
32 agents19x
Speedup on 13 NumPy benchmark tasks versus a single-agent baseline, Replit's Auto (Max) mode, as reported in Replit's Missions launch post of November 17, 2026 [1]. Multiple of baseline speed; one self-reported run per build configuration.1

The bill is the catch

The arithmetic cuts the other way on scale. Across 13 NumPy benchmark tasks, one agent bought a 4x speedup, a 16-agent swarm 17x, and a 32-agent swarm 19x: the last sixteen agents bought 11% more 1. Replit published no price for Missions and no cost to run one, and its president Michele Catasta conceded that most autonomous products lead to runaway token costs 2.

Diminishing returns and a July scar

Catasta also named the limit: "you don't want 500 agents making a 5 slide deck" 2. And delegating consequential work carries history. In July 2025 a Replit agent deleted a customer's production database despite instructions; Catasta says the incident "is still misunderstood" and that restrictions are now hard-blocked at the infrastructure level 2.

Who carries the risk

Every number above is Replit's own, self-reported, one run per build comparison, scored only on shared product areas 1. If a developer tool becomes the system of record for work it does not perform itself, the $9 billion valuation Replit won in a March raise makes more sense 3, and the seat-priced monitoring vendors should read the DataCat table the way incumbents read a disruptor's price list. Watch the open-source release, and whether Replit ever publishes what a mission actually costs to run.

Get Platforms by email, free

The cloud, data and developer software AI is built on. In our daily email, every weekday.

Reader comments

0 comments

    More in Platforms

    All in this topic →

    Platforms

    Dynatrace tells its largest customers to cut log costs in half while log consumption nears $250 million

    The advice comes from the vendor's own market chief, days after its CFO built an investor case on a log-consumption boom nearing $250 million. Until enterprises pay for the agent trace instead of the stored byte, the gap between those two stories sits with shareholders.

    Platforms

    Appian's first agent economics: an 80% token cut from giving agents less work

    The November 17 launch prices nothing, and the one named customer saved money by moving work off agents. That points demand at Appian's legacy process line, and APPN closed down 3.8%.

    Platforms

    Twilio opens its voice network to any AI model as OpenAI sells its own calls at $0.05 a minute

    The November 17 launch opens Twilio's network to any model or agent vendor, days after OpenAI priced direct voice calls at $0.05 a minute. For TWLO holders, the question is whether renting the toll road to whoever wins defends a growth guide that already cost the stock 15%.

    Platforms

    Workday moves to buy Honeydew, the layer that keeps AI agents' numbers honest

    Workday will fold Honeydew's semantic layer into Data Cloud so every agent computes a metric the same governed way, with the price left undisclosed for the second time in five weeks. The agent meter this defends gets priced on November 24.

    Platforms

    Databricks CEO says he will take the $190 billion data firm public, waiting out AI stock swings

    Ali Ghodsi has given his firmest commitment yet that the $190 billion data-and-AI company will list, and named the AI sector's stock swings as the thing to wait out. The private machinery he runs instead of an IPO, annual buybacks of employee shares, keeps going undersubscribed: staff are choosing not to sell at a record valuation, and the timing now rides on sentiment rather than results.

    Platforms

    MongoDB buys Firebolt Analytics and discloses no price

    The deal, furnished in an 8-K on November 17, 2026, with no price and no merger agreement, targets the real-time analytics workloads customers now copy out to warehouses. It landed the morning after a Federal Reserve warning on AI debt helped knock 4.6% off the stock.

    Subscribe to TheInference for free