Cycle on Sep 28, 2026

I'll start by reviewing the open threads and existing notes for this topic, then gather fresh evidence.

I have a rich set of prior findings (last updated ~Sept 21) and many open watches. This cycle should cover the past week (Sept 21–28, 2026) for new developments. Let me run several searches in parallel to find fresh news.

Some leads already: a Money Marketing piece connecting Claude for Financial Advisors to the Mills Review, Paywhere's programmable AI banking platform, and Tetrix's Series A. Let me dig deeper on the watches and fresh funding news.

Strong leads. The Schwab/Anthropic unveiling at Future Proof and Fidelity's Savvy Wealth backstop are directly relevant to my open custodian watch. Let me fetch the key pages.

Rich haul. Now let me fetch the Money Marketing piece (Mills × Claude for Financial Advisors), the Paywhere piece, and check the Catena watch status.

The Catena watch hasn't fired (still conditional approval, no final authorization or launch — though Sean Neville exiting the board is a new detail). The Money Marketing piece gives a valuable regulatory-lens analysis. Let me verify two more sources: the Saris Series A and the Mariner AI-bots story.

I have strong evidence now. One note: the Saris article turned out to be dated May 29, 2026 — not fresh, so I'll treat it as background only. Let me quickly verify the Paywhere piece (an incumbent-response infrastructure play) before writing notes.

I'll now persist the findings — six notes (three updates to existing slugs, three new subjects), then register watches, then finish. First, the two big wealth-channel stories:

First note saved. The second call was malformed — let me write the RIA custody note properly now.

Both saved. Now the four remaining notes — compliance capital wave, the new agent-trust infrastructure theme, Paywhere's headless banking, and the AI-workforce category update:

All six notes saved. Now registering the forward-looking watches this cycle surfaced:

All watches registered. This cycle surfaced substantial new evidence — here's the summary: **This cycle's findings (6 notes written):** 1. **Updated `frontier-labs-services-land-grab-finance-ai-2026`** — Schwab's exclusive Anthropic deal, unveiled at Future Proof, is **read-only** ("more of a research function" per Schwab's Jon Beatty). Analysts (Lex Sokolin) give Altruist a **2–5 year lead** in agentic capability; RIAs are pushing back on data sharing ("let client data fly into the abyss of Anthropic"); and UK commentary (Ben Goss) maps the product onto the FCA Mills automation spectrum. 2. **New `ria-custody-agentic-ai-2026`** — The custody AI war now has three distinct incumbent strategies: Vanguard buying the AI-native challenger (Altruist, $4.6B), Fidelity renting out an AI front end (backstopping Savvy Wealth's mid-2027 custody launch), and Mariner building an in-house AI workforce ($175M for 700 Humanity Labs bots) — while Goldman, Pershing, and TradePMR stay silent. 3. **Updated `vertical-ai-compliance-regulatory-intelligence-2026`** — Compliance AI funding widened: MCO $100M+ (Accel-KKR **private credit**), Footprint $25M (QED + **MUFG**), Numeral $100M (Insight). 4. **New `agentic-trust-security-infrastructure-2026`** — A new "agent trust" layer got $650M+ in one week: Baselayer $35M (verify agents), Outerlimit $16M (zero-trust agent authorization), Island $400M at $6.4B (**J.P. Morgan** Growth Equity), Cyera $400M (**Goldman Sachs**). 5. **New `ai-assistant-banking-connectivity-2026`** — Paywhere's programmable banking platform (Sept 23) enables "headless banking": banks living inside ChatGPT/Claude with entitlements intact, against a backdrop of $3T deposit migration and only 7% bank AI adoption. 6. **Updated `kastle-ai-workforce-banking-2026`** — The AI-workforce-for-banks category crowded: FintechOS $28M (Santander CIB debt) and Go.AI $85M (on-prem bank AI) joined Kastle. **Watches registered (4):** Vanguard–Altruist close (one_shot), Schwab write-access tripwire (ongoing), Savvy Custodial Platform launch (one_shot, mid-2027), Mariner bot-deployment outcomes (ongoing). **Watches checked, not fired:** Catena OCC final approval (still conditional), rival custodian–frontier-lab deal (Fidelity/Goldman/Pershing all silent), OpenAI vertical expansion (nothing new).

The research cycle is complete. Six findings were persisted (three updates to existing threads, three new subjects), four forward-looking watches were registered, and the cycle closed cleanly. **The strategic headline from this cycle:** the vertical AI landscape for financial services just split into identifiable camps. In wealth, incumbents are choosing between *buying* the AI-native challenger (Vanguard–Altruist), *renting* an AI front end (Fidelity–Savvy), or *building* an internal AI workforce (Mariner) — while Schwab's exclusive Anthropic deal is exposed as read-only, ceding the agentic capability lead for years by analyst estimate. Meanwhile, a new "agent trust" infrastructure layer (identity, authorization, control, data governance for AI agents) drew $650M+ in a single week with J.P. Morgan and Goldman Sachs as strategic buyers — evidence that banks now treat agent oversight as a control requirement. Compliance AI and bank-operations AI both continued their capital acceleration, with private credit and bank-strategic money joining the VC wave.