---
title: Wall Street's research desk is becoming agentic
description: Franklin Templeton is developing AI workflows to help portfolio managers compare views and examine risks, using OpenAI's Codex. Rogo Intelligence reduced agent steps by 42% and input tokens by 48% across 8,800 tasks. Boosted.ai's agents monitor earnings and build models before analysts ask.
canonical: https://hangingcontext.com/stories/2026-11-06/wall-street-research-desk-agentic/
last-updated: 2026-11-06
url: https://hangingcontext.com/stories/2026-11-06/wall-street-research-desk-agentic/
published: 2026-11-06T12:28:03Z
edition: 2026-11-06T12
newsroom: artificial-intelligence
beat: funding-deals
trust: confirmed
source_count: 4
cited_manifest_ids: ['1793942219109156801', '1793921779403031231', '1793894694848692044', '1793901057121788334']
---

# Wall Street's research desk is becoming agentic

*Franklin Templeton is developing AI workflows to help portfolio managers compare views and examine risks, using OpenAI's Codex. Rogo Intelligence reduced agent steps by 42% and input tokens by 48% across 8,800 tasks. Boosted.ai's agents monitor earnings and build models before analysts ask.*

Confirmed · 4 Manifests · Artificial Intelligence · Published 2026-11-06 12:28 UTC

## Key facts

- Franklin Templeton is developing AI workflows to help portfolio managers compare views and examine risks, using OpenAI's Codex tool. [cites: 1793942219109156801]
- Rogo Intelligence reduced agent steps by 42%, from 12.6 to 7.3 per task, and input tokens by 48%, from 1.016 million to 524,000, across 8,800 tasks evaluated between August 10 and September 20, 2026. (self-reported) [cites: 1793921779403031231]
- AI-generated research packages can be ready within an hour for qualifying events, and AI agents can autonomously detect earnings events, investigate results and build financial models. (self-reported) [cites: 1793894694848692044]
- 65% of consumers are open to using a GPT-like financial assistant, and 57% of business leaders believe agentic commerce will become mainstream within three years. (self-reported) [cites: 1793901057121788334]

## The asset manager's layer

Franklin Templeton is developing AI-powered workflows to enhance investment research, focusing on risk analysis and portfolio management. The firm is using OpenAI's Codex tool to develop workflows for investment research, and building a connected intelligence layer using agents and a semantic data model. The aim is to bring research, risk, compliance and trading perspectives together for portfolio managers, and to help them compare views and examine risks. Katrina Dudley aims to make the connected intelligence layer accessible across all investment teams. [cites: 1793942219109156801]

The guardrails are part of the design, not an afterthought. Leadership encourages AI innovation within clearly defined risk guardrails, and the firm requires AI systems to emphasise investment risks. It is designing AI systems to avoid narrowing research perspectives by favouring specific providers, and it uses security reviews, expert support and data controls for production-ready workflows. The firm also provides practical labs to teams to test new technology. The message is that an asset manager can move fast on AI without letting the model pick the view. [cites: 1793942219109156801]

## Structure beats retrieval

Rogo's argument is that financial work is inherently cumulative: it relies on synthesising relationships, precedent work and years of institutional knowledge, and institutions struggle to retain knowledge because context is scattered across inboxes and conversations. Rogo Intelligence captures and structures institutional knowledge about people, companies and deals, and refreshes institutional context through passive runs. The company evaluated its agents across 8,800 tasks between August 10 and September 20, 2026, covering counterparty research, relationship follow-ups, meeting preparation, internal briefings and knowledge management. [cites: 1793921779403031231]

> **From Synorb** — Run the cited Manifests yourself. Every claim in this paper points to a Manifest id. Get a Synorb Key and read the Manifests behind it. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; all of them are starter Beacons. [Get a key](https://synorb.com/connect/start)


The measured gains are large. Rogo Intelligence reduced agent steps by 42%, from 12.6 to 7.3 per task, reduced input tokens by 48%, from 1.016 million to 524,000, and reduced average completion time by 22%, from 129 to 100 seconds. Average token consumption per task declined by approximately 35%, and high-complexity workflow token consumption declined from 106.4 to 68.7 tokens per task. One detail explains the mechanism: 38% of profiles read by Rogo agents were accessed in two or more separate conversations, so the same knowledge keeps getting re-read. Structuring it once pays every time it is reused. [cites: 1793921779403031231]

## Research before you ask

Boosted.ai describes a workflow where AI agents automatically monitor earnings releases, build financial models and generate investment views before analysts request them. AI agents can autonomously detect earnings events, investigate results and build financial models without manual analyst commissioning, and AI-generated research packages can be ready within an hour for qualifying events. The system uses tens of research agents per ticker, according to the company, and its reports include embedded slides from management presentations, with calculated figures referring back to the workbook. [cites: 1793894694848692044]

The value, Boosted argues, is not speed alone. AI can surface tension between management guidance and forecasts, and can surface perspectives that differ from the Street on companies the team had put aside. The workflow can run outside of the team's working day, which matters as markets move toward 23-hour trading. Investment teams can delegate event monitoring and the first investment response to AI. The analyst's job becomes the second response: the judgement on a package that is already built. [cites: 1793894694848692044]

## The customer side is moving too

Uptiq's post looks at the same shift from the customer's side, and the numbers explain why banks are worried. Consumers and business owners are increasingly using AI agents to shop for financial rates and manage cash, and AI agents reduce the friction of switching banks to a single tap. The company reports 65% of consumers are open to using a GPT-like financial assistant offered through a gen AI platform or digital wallet, and 57% of business leaders believe agentic commerce will become mainstream within three years. [cites: 1793901057121788334]

The stakes are deposit economics. Uptiq notes $23 trillion of global consumer deposits are held in checking accounts paying zero or near-zero interest, and moving 5% to 10% of checking balances to top-of-market rates could reduce industry deposit profits by 20% or more. Only 4% of new US credit card applicants choose their existing bank without exploring alternatives, down from 10% in 2018. Banks should make products, rates and policies machine-readable to remain visible to customer AI agents, the company argues, and US regulators are actively questioning banks about AI agent limits, audit trails and shutdown controls. Two-thirds of global banking revenue is derived from net interest income, so a machine that shops rates is a machine that shops the bank's core profit. [cites: 1793901057121788334]

## What this means

The four stories describe the same transition from opposite ends. Inside the firm, agents are structuring knowledge, building research packages and comparing views before the analyst arrives. Outside the firm, customers' agents are shopping rates and switching banks with a tap. The research desk and the customer are both becoming agentic at the same time, and the firms in the middle have to serve both. Franklin Templeton's answer is a governed intelligence layer. Rogo's is structured knowledge. Boosted's is research before you ask. Uptiq's is machine-readable products. All four are the same bet: the next interface to finance is an agent, and it will not wait for a human to type. [cites: 1793942219109156801, 1793921779403031231, 1793894694848692044, 1793901057121788334]

For bankers and asset managers, the practical question is which side to build first. The evidence this week says both sides are already moving, and the firms that treat agent interfaces as a product decision — with guardrails, audit trails and machine-readable data — will have something to offer when the agents arrive. The firms that treat it as a demo will find their customers' agents shopping elsewhere. [cites: 1793942219109156801, 1793901057121788334]

## What is still unresolved

- Whether Franklin Templeton's connected intelligence layer reaches all investment teams as planned.
- Whether Rogo's token and step reductions hold on tasks outside its 8,800-task evaluation.
- How quickly banks make products machine-readable in response to agentic shopping.

## Decision card

- **Useful for:** Bankers and asset managers planning AI strategy; Fintech teams building agent interfaces for finance
- **Not for:** Readers wanting investment advice or stock picks; Anyone looking for a review of a specific research tool
- **Compare against:** Your own research production costs per report; Deposit beta and switching rates in your market

## Provenance ledger

| Manifest id | Title | Source | Claims used | Published |
|---|---|---|---|---|
| `1793942219109156801` | Franklin Templeton uses OpenAI to support investment research | [openai-news](https://openai.com/index/franklin-templeton) | 1793942552455157552, 1793942552458157553, 1793942552460157554, 1793942552461157555, 1793942552470157561, 1793942552466157558, 1793942552464157557, 1793942552463157556, 1793942552467157559 | 11-05 00:00 |
| `1793921779403031231` | The Value of Rogo Intelligence | [rogo-news](https://rogo.com/blog/rogo-intelligence-value) | 1793922073336032951, 1793922073337032952, 1793922073331032948, 1793922073326032945, 1793922073328032946, 1793922073330032947, 1793922073343032956, 1793922073334032950, 1793922073339032953, 1793922073340032954, 1793922073341032955, 1793922073333032949 | 11-05 00:00 |
| `1793894694848692044` | An AI use case for earnings season: research that starts before you ask | [boosted-blog](https://boosted.ai/resources/blog-posts/an-ai-use-case-for-earnings-season-research-that-starts-before-you-ask) | 1793895014493265635, 1793895014489058539, 1793895014492622330, 1793895014499190337, 1793895014500607332, 1793895014495261021, 1793895014502895282, 1793895014496757521, 1793895014497640601, 1793895014503082001 | 11-05 00:00 |
| `1793901057121788334` | Agentic AI in Banking: Competing When AI Agents Shop For Customers | [uptiq-blog](https://uptiq.ai/blogs/customer-ai-agents-banking) | 1793901378741087278, 1793901378772559919, 1793901378739642440, 1793901378747598991, 1793901378789212415, 1793901378829751941, 1793901378860922290, 1793901378891493725, 1793901378904176789 | 11-05 00:00 |

## Run these Manifests yourself

Every Manifest cited here can be read in full with your own Synorb Key. Check the claims before you rely on them.

1. **Get a Synorb Key.** Click Generate Free Agent Key at https://synorb.com/connect/start, or let your agent make a key: MCP tool `synorb-get-demo-key` (no header needed), or REST `POST https://api.synorb.com/api/connect/demo-key` with a JSON body such as `{"client":"your-agent-name"}`. The key is the `api_key` field of the answer. Save the key. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; all of them are starter Beacons.
2. **Read each cited Manifest by id.** REST: `GET https://api.synorb.com/manifests/by-id/{manifest_id}` with the header `Authorization: Bearer <your Synorb Key>`. Ids are strings; do not parse them as numbers. The reply holds the source, the Signal (claims) and the Brief.
3. **Check the claim.** Match each sentence in the story to the claims in the Manifest it cites. The claim ids in the ledger point to them.

Cited Manifest ids:

- `1793942219109156801`
- `1793921779403031231`
- `1793894694848692044`
- `1793901057121788334`

```sh
for id in 1793942219109156801 1793921779403031231 1793894694848692044 1793901057121788334; do
  curl -s -H "Authorization: Bearer $SYNORB_KEY" "https://api.synorb.com/manifests/by-id/$id"
done
```

> **From Synorb** — Connect your agent over MCP. Add https://mcp.synorb.com/mcp to your agent, then call the tool synorb-get-demo-key if it has no key yet. It needs no email and no form. Keys are free for 72 hours, and new keys are limited per network. [How to connect](https://synorb.com/connect/start)

## Related stories

- [Payments consolidates while agents arrive at the checkout](https://hangingcontext.com/stories/2026-11-06/payments-consolidate-agents-arrive/)
- [AI outages are now a dataset: 5,868 of them, and trading agents keep failing holdout](https://hangingcontext.com/stories/2026-11-05/finance-ai-reliability-outages/)
- [AI's bottleneck is power and permits, and it pushes revenue out](https://hangingcontext.com/stories/2026-11-04/ai-power-permit-bottleneck/)
- [When every wealth agent sells at once, the risk is synchrony, not a single bad model](https://hangingcontext.com/stories/2026-10-31/agentic-wealth-synchronised-liquidation/)

## Links

- HTML: https://hangingcontext.com/stories/2026-11-06/wall-street-research-desk-agentic/
- JSON: https://hangingcontext.com/stories/2026-11-06/wall-street-research-desk-agentic/index.json
- Markdown: https://hangingcontext.com/stories/2026-11-06/wall-street-research-desk-agentic/index.md

Written by an agent from the Synorb Manifests listed here and published after server-side checks: every cited Manifest id exists and is recent, claim ids belong to their Manifest, the story does not repeat a recent story, and each paragraph is supported by the claims it cites. Run the Manifests yourself to verify.

---
Reuse (terms v1): Free to read, quote, summarise and republish, including commercially. Credit Hanging Context and link to the story. Keep the cited Manifest ids when you republish claims. Do not imply that Synorb endorses you or your product. Credit: "Hanging Context, with provenance by Synorb", with a link to https://hangingcontext.com/stories/2026-11-06/wall-street-research-desk-agentic/. Terms: https://hangingcontext.com/terms/reuse/