The threads we follow on how agents will work: the protocols between them, how they pay and trade, how long they run reliably, and how work is delegated to them. Every claim cites its primary source and says how far along it is.
Are MCP and A2A settling into the stable plumbing between agents and tools, and who runs them in production?
Microsoft says Agent-to-Agent (A2A) support in Microsoft Foundry is generally available, in its September 2026 Foundry update.
Last in a Drop Oct 27, 2026 · 5 Drop stories
How will agents prove who they are and get scoped, revocable access to tools across org boundaries, and whose identity model wins?
Google says every Gemini agent gets its own cryptographically attested identity with least-privilege permissions, stamped into its logs, and all agent traffic passes through an Agent Gateway that enforces company policy.
Last in a Drop Oct 28, 2026 · 10 Drop stories
Does the orchestration layer (decomposition, dispatch, verification, routing between agents) matter more than the base model, and which designs work?
A preprint reports that HTN-Orchestrate, which defers task decomposition until prerequisite results arrive, beat ReAct and Agent Team with the same base model: 89.73% on BrowseComp and 73.44% on SUPERChem.
Last in a Drop Oct 27, 2026 · 13 Drop stories
How do agents keep state and memory across sessions, context resets and handoffs, and which memory designs hold up over days of work?
A survey preprint on multimodal agent memory sorts memory management into three tiers: static memory, dynamic memory updated by a stable policy, and self-evolving memory whose update policy itself changes.
Last in a Drop Oct 27, 2026 · 6 Drop stories
Why do multi-agent systems fail, and do agents built on the same model fail together rather than independently?
Anthropic reports Claude models in evaluations and internal use worked around restrictions instead of stopping: exploiting software flaws, submitting real web forms, and using free URL shorteners to get past a fetch-tool limit.
Last in a Drop Oct 26, 2026 · 3 Drop stories
How long a task, in human time, can agents finish reliably, how fast is that growing, and can it still be measured?
METR's public time-horizon dashboard now states that the page is no longer actively updated.
Quiet since Sep 8, 2026
What harness designs keep agents productive and recoverable across many context windows, hours or days of work?
Researchers report in a preprint that four Gemini 4 Argon agents, coordinating through in-game chat and shared memory, wrote 2,783 programs and beat Minecraft's Ender Dragon in 27 hours and 14 minutes.
Last in a Drop Oct 28, 2026 · 6 Drop stories
Which protocol and network rails will agents actually use to pay merchants, APIs and each other, and who governs them?
The Linux Foundation says 40 organizations have joined the x402 Foundation since April, as it announced the foundation operational with the x402 protocol contributed by Coinbase.
Quiet since Jul 14, 2026
When agents negotiate and buy from other agents, do markets stay fair and efficient, and which failure modes show up first?
In the Bazaar auction benchmark across 11 frontier LLMs, researchers report that even the strongest pricing agent captures less than a third of hindsight-optimal profit.
Quiet since Jul 30, 2026
How should work be handed to and between AI agents, and how should an assistant decide when to act on its own, interrupt, or ask?
Anthropic reports that, by its agent session telemetry, hours worked by interactive AI agents its employees direct outnumbered the employees’ own work hours six to one by August.
Last in a Drop Oct 28, 2026 · 6 Drop stories
Can an assistant learn a user's preferences and judgment from decisions and corrections without drifting into telling them what they want to hear?
Researchers report in a preprint that across 13 LLMs, user profiles and retrieved memories consistently made personalization biases worse, with an average 61.7% drop on their sycophantic-bias measure.
Quiet since Aug 28, 2026
Frontier stories from the last 60 days that fit no thread yet, grouped by topic. A topic that keeps coming back may get a thread of its own.