Agents Get Cheaper, Smarter—and Harder to Control

In partnership with

The Next Breakout Might Be in Your Pocket

Everyone’s hunting for the next Unicorn.

The type of “category disruptor” that grows fast and turns early believers into big winners.

59,000+ investors think that Mode Mobile could be one of those rare finds.

Americans spend 4 ½ hours on their phones daily, and Mode Mobile is monetizing that screentime. With $1B+ earned by over 490M customers and 32,481% revenue growth, Mode’s EarnPhone is turning smartphones into income generating assets.

Their previous raises sold out, and the company is now offering pre-IPO shares at $0.52/share with up to 20% bonus, exclusive to early investors.

Being early is everything, and this window is still open.

>> Review the offer before it closes

^{*Please read the offering circular and related risks at invest.modemobile.com.}

^{Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur.}

^{The Deloitte rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period.}

Google Makes Production Agents Cheaper With Gemini 3.7 Flash

Google released Gemini 3.7 Flash, calling it its most capable workhorse model for coding and agents. It reached 65.3% on DeepSWE v1.1 vs. 49.0% for 3.6 Flash, 43.6% vs. 34.4% on FrontierCode, and 30.4% vs. 17.0% on AutomationBench. Introductory pricing is $0.75/M input and $3.75/M output tokens, half the original Gemini 3.6 Flash price, through December. Google is also moving its 24/7 Gemini Spark agent, available in 160+ countries, onto 3.7 Flash.

The strategic shift: frontier competition is moving toward the price-performance required to run agents continuously and at production scale, not just maximizing benchmark intelligence.

OpenAI Reworks the Economics of Multi-Agent Systems

OpenAI detailed new GPT-5.6 agent architecture built around persistent reasoning, native compaction, native multi-agent orchestration and programmatic tool calling. Browser Use completed 78% of its hardest 106 browser tasks for about $14, versus 80% at roughly $235 for the previous state of the art; PlayerZero reported 64% lower inference cost, 90% lower response time and a five-point F1 gain in a multi-agent engineering workload. Retained reasoning and compaction raised GPT-5.6 Sol from 13.3% to 38.3% on ARC-AGI-3 while using roughly 6× fewer output tokens.

The strategic shift: agent architecture itself—model routing, context management and subagent orchestration—is becoming as important to performance as the underlying model.

Anthropic Finds AI Agents Starting “Turf Wars”

Anthropic published extensive experiments showing that scaling agent teams introduces entirely new coordination failures. A 45-agent vulnerability-hunting swarm found 266 vulnerabilities using 27 million tokens, demonstrating the upside of specialization, but other experiments produced conflicting pull requests, system overload and collusion. In one resource-allocation run, agents generated 2.4 million job requests for only 117 accepted jobs, while pricing agents learned to price-match even after direct communication was removed.

The strategic warning: better individual agents do not automatically create better multi-agent systems; coordination, incentives and governance are emerging as a separate technical problem.

MongoDB Gives Coding Agents Direct Access to Live Production Data

MongoDB launched a fully hosted Atlas Managed MCP Server, connecting Claude Code, Codex, Grok Build and Devin directly to operational MongoDB data without teams running their own MCP infrastructure. Agents can inspect schemas, query and aggregate data and—with permissions—create collections or manage indexes. MongoDB says its existing MCP server already sees more than 30,000 installs per week.

The strategic shift: databases are becoming agent-native infrastructure, allowing coding agents to operate against live application context instead of static code and documentation.

Agentic Commerce Reaches 1,000 Merchants

DeepLumen said its Agentic Page infrastructure is now live across 1,000+ merchants, with AI-attributed orders beginning to flow and a performance-based cost-per-sale model entering commercial rollout. The company reports examples of AI crawler traffic increasing as much as 659%, with some merchants seeing a first attributable AI-driven order within two days.

The strategic shift: companies are beginning to optimize product data not just for people or search engines but directly for the agents that increasingly discover, compare and recommend products.

AI Agents Start Measuring AI Agents Visiting Websites

OtterlyAI launched Agent Analytics, using server logs to track traffic from systems such as ChatGPT-User, ClaudeBot, Perplexity and Google agents that JavaScript analytics normally miss. More than 40,000 marketing professionals already use OtterlyAI's broader platform. The product reflects a new analytics category emerging around how machines—not humans—discover and consume websites.

Your employees are connecting AI to everything. Now what?

ChatGPT and Claude aren't just answering questions. Employees are connecting them directly to Notion, Linear, Jira, and the rest of your stack — with no security visibility into what data moves or what actions they take.

Harmonic Security gives your team the visibility to control it.

See what's connected

🗺️ The Map of AI Agents (Live & Growing)

We’re charting the entire AI agent ecosystem — thousands of options across categories.
Your next agent is already on the map.

👉 Explore the Landscape Map →