How JamLabs built embedded analytics for a global interdealer broker 4x faster with Claude Code

JamLabs built the embedded analytics dashboards, including a chatbot, inside a post-trade analytics platform for the data and analytics business of a leading global interdealer broker, using Claude Code and Claude Cowork. The team delivered about six months of planned roadmap in roughly six weeks, which is about 4x faster than planned.

The client brings clarity to complex OTC markets through multi-asset data and analytics. Its post-trade analytics platform helps trading desks, compliance teams and portfolio managers measure and improve execution quality.

“JamLabs delivered on time and on budget, with a functional, client-led outcome.” Chief Information Officer, the client’s data and analytics business

Results at a glance

• 4x faster delivery: about six months of planned roadmap shipped in about six weeks

• 14 client dashboards built from scratch and live in production

• Full planned scope delivered, with nothing cut to hit the date

• Plain-English questions, answered by Claude, live from the first release

The challenge: six months of roadmap on a fixed date

The client wanted a new analytics experience inside its platform: embedded in the product, running on its own cloud data, and ready for AI from the first release. The plan called for about six months of roadmap work, against a much shorter fixed date.

How did Claude Code and Cowork deliver 4x faster?

JamLabs has engineered and operated the platform with the client for five years, so the team knew the data, the users and the edge cases. Claude turned that knowledge into shipped work at a new pace.

• Claude Code was the team's primary development interface. Engineers directed it to write and test the dashboards, the data model behind them and the access rules, then reviewed and shipped the results.

• Claude Cowork carried the work around the code: specifications, test plans, documentation and status reporting.

• Jitto, JamLabs' agentic data engineering and analytics product, is how the platform is delivered and supported.

What's delivered

• A Claude-powered chatbot, embedded in the platform, that answers client questions in plain English

• 14 client dashboards, embedded directly in the platform

• Scheduled reports, generated from the same data

• A governed semantic layer: each user sees only their own data, and every query is audited. The chatbot follows the same rules, so every answer respects the same access controls.

• No data movement: the platform runs directly on the client's existing cloud data infrastructure.

What's next: the platform inside the AI tools clients already use

JamLabs is building an MCP server that brings the platform into Claude and the other AI tools clients already work in. Clients will be able to pull governed execution-quality data into their own analysis and workflows, under the same access rules as the dashboards.

The chatbot is getting smarter too:

• Answers that explain, not just report: follow-up analysis on why execution quality changed, not only what changed

• From question to deliverable: turning a chat answer into a saved view or a scheduled report

• Proactive insights: flagging notable shifts in a client's execution quality before they ask

FAQ

What did JamLabs build for the client?

The embedded analytics inside the client's post-trade analytics platform: 14 client dashboards, scheduled reports and a Claude-powered chatbot, built with Claude Code and Claude Cowork and now live in production.

Which Claude tools did JamLabs use?

Claude Code for building and testing, and Claude Cowork for specs, test plans, documentation and reporting. Claude also powers the chatbot that answers client questions in plain English.

About Jitto and JamLabs

Most teams don't lack analytics. They lack access to data they can trust. Jitto is JamLabs' AI data team: agents that do the data engineering and answer the questions, with independent checks, human sign-off and a visible trail behind every result.

• Jitto Build turns a request into a working data warehouse. Describe what you need, and Jitto plans, builds and tests the models, then separately checks row counts, totals and keys. A person on your team reviews the evidence and approves every change before it merges.

• Jitto Analyze lets your team ask questions in plain English and see the work behind each answer: the definition, the calculation and the source tables it came from.

JamLabs is a Toronto-based AI and data engineering firm and the maker of Jitto. The team builds with Claude Code and Claude Cowork, and staff members hold active Claude and hyperscaler certifications.

Building analytics into your product? See how Jitto can help at jitto.ai.