Jayeeta Putatunda, forward deployed AI engineering lead at Turing, demonstrated a multi-agent workflow on the Zero to Agent in 30 Minutes episode that transforms the daily influx of financial headlines into structured analyst briefings. The workflow aims to keep internal research, coverage assignments, and market views current while highlighting which incoming items require an analyst’s attention.
The system, step by step
-
Define the scope
The process starts by setting the analyst’s focus area, research questions, and the time window to consider — for example, the last one day, seven days, or 30 days. Specifying preferred sources and narrowing scope up front directs searches so they produce relevant results rather than generic hits.
-
Plan the coverage
A coverage planner agent breaks the research topic into discrete sections (such as market backdrop or company catalysts), enabling subsequent searches to run in parallel instead of as a single long sequential query.
-
Search and gather
A news researcher agent runs multiple queries against the preferred sources first, falling back to general web searches only if those sources don’t return enough results.
-
Validate sources
A validation agent checks each link for a working, clickable URL and correct publication date, removes duplicate stories covering the same event, and filters out paywalled pages that won’t yield usable content.
-
Generate the briefing
A briefing writer agent assembles the validated developments into a set structure: an executive summary, key bullet points, and a section-by-section breakdown explaining what each development means for the analyst and what to watch next.
-
Capture feedback into memory
A feedback agent logs corrections such as formatting or terminology preferences and saves them to a memory database so the next briefing run applies those preferences automatically without the analyst having to re-prompt.
Implementation details and broader takeaways
Putatunda implemented the entire stack on open-source models so newcomers can run it without an API key; she recommends starting with a smaller model before scaling up. Building several single-purpose agents instead of one monolithic agent means a failure at one stage doesn’t force restarting the whole pipeline, and each agent’s output remains easier to trace and debug.
The full code base, including sample data and the demo UI, is available in Jayeeta’s GitHub repository, allowing readers to clone the project and run the briefing agent on their own systems.
What’s coming next
This week, Maxim Salnikov will join Zero to Agent in 30 Minutes to build a supply chain for agent context, demonstrating how to source approved packages from a trusted registry, pin and hash-verify them on any harness, and enforce organizational policy with a CI gate that cannot be bypassed.
