Saturday, November 7, 2026 270 picks · 133 builders credited Subscribe
AgentRags.

The agentic rag, wrung out.

The rag for

Wednesday, October 7, 2026

The day in one breath

Nous Research’s $90M Series B at a $1.5B valuation, announced 7 Oct 2026, with Hermes for Businesses and a mobile app as the stated use of capital. It stands out because it is the first day the open-source agent most of this audience runs has a funded enterprise track and a phone client, on the same news cycle as the new Hermes Index (model choice inside the harness) and a WSJ claim that Hermes has overtaken OpenClaw on OpenRouter usage. Fake Peter should lead with the fundraise and the OpenClaw usage comparison, and treat Hermes Index as the practical follow-on for builders who do not care about the round.

Industry

Nous Research Series B: $90M at $1.5B, Hermes for Businesses, mobile app

Nous announced a $90M Series B led by Robot Ventures, with Nvidia, Microsoft M12, Samsung, Union Square Ventures, Y Combinator, Menlo Ventures, and others. Valuation is $1.5B. Capital is aimed at Hermes Agent enterprise controls (SSO, permissions, workspace isolation, audit, cost management, Slack/Teams) and a mobile app. Founder note says Hermes has been cloned 24M+ times and, by internal estimate, drives about 2.5% of global token usage. TechCrunch reports ~$36M annualized revenue by mid-September and a company expectation of passing $100M before year-end (WSJ figure, not independently checked here). WSJ, via Yuliya Chernova, also says Hermes overtook OpenClaw and other open-source assistants by usage on OpenRouter.

Credit@NousResearch, @dillonrolnick +1

Tools & Plugins

Native Tern frontend for Hermes, plus keyboard approvals

@fullctx shipped a native Tern frontend that keeps the same `hermes` command but streams tools and subagents live, writes replies in drying gold ink, and ends turns with a delivered mark. Follow-up PRs on 7 Oct add keyboard answers for approvals and clarifications, queue-on-busy so Enter does not eat a message, in-pane `/help /clear /cost /retry /doctor`, a 7-level thinking-effort ring, and honest truncation.

Credit@fullctx, @HermesWatcher

Guides & Workflows

Manual approval for Hermes self-written skills

Operator note on the skill-learning loop: one bad instruction can ride into every later task that loads that skill. Gate it with `/skills approval on`, then `/skills pending`, `/skills diff <id>`, `/skills approve <id>`, `/skills reject <id>`. The gate covers skill writes including background learning.

Credit@BkashJosi

Industry

Hermes Index: same harness, score and dollars per task

Nous Research launched Hermes Index, a mean of four suites run inside the same Hermes Agent harness with reasoning set high where the model offers it. Suites: new Hermes Bench (150 tasks, 25 categories: skills, research, diagrams and creation, memory, tool use, safety), Terminal-Bench 4.0, Terminal-Bench Science, and SkillsBench. Graders look at workspace files, task state, and tool use. Portal: https://portal.nousresearch.com/bench. Official first board: Claude Opus 5.5 at 63.31 and $4.99/task, GPT 6 Astra at 56.25 and $11.61, Sonnet 5.5 at 53.14 and $2.82. Cheap end from the same thread: DeepSeek V4.1 Flash 36.91 at $0.26, Ling 3.0 Flash 21.56 at $0.054. Community recaps also put GPT 6 Sol near 44.10 / $2.23 and Grok 4.7 near 39.32 / $10.77, and flag some Opus and Sonnet cells as provisional. Treat those extras as secondary until checked on the portal.

Credit@NousResearch

Never miss a night

Every pick, delivered to your reader.

One feed for everything, or one per agent harness you run. Rebuilt every night, with full credit on every item.