Less talking, more building: ๐ ๐๐น๐๐ถ-๐๐ด๐ฒ๐ป๐ ๐ฅ๐ฒ๐๐ฒ๐ฎ๐ฟ๐ฐ๐ต ๐ฆ๐๐๐๐ฒ๐บ ๐ณ๐ฟ๐ผ๐บ...
Aurimas Griciลซnas@Aurimas_Gr
19 views
Nov 02, 2025
~2 min read
Advertisement
1
Less talking, more building: ๐ ๐๐น๐๐ถ-๐๐ด๐ฒ๐ป๐ ๐ฅ๐ฒ๐๐ฒ๐ฎ๐ฟ๐ฐ๐ต ๐ฆ๐๐๐๐ฒ๐บ ๐ณ๐ฟ๐ผ๐บ ๐๐ฐ๐ฟ๐ฎ๐๐ฐ๐ต! ๐
Some weeks ago I released an episode of my Newsletter and an update to the ๐๐ ๐๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ ๐๐ฎ๐ป๐ฑ๐ฏ๐ผ๐ผ๐ธ GitHub repository.
There I implemented a Deep Research Agent from scratch without using any LLM Orchestration frameworks (using DeepSeek-R1 for some planning tasks).
In the project we implement the following Agentic topology:
๐ญ. A user provides a query or topic to be researched.
๐ฎ. A LLM creates an outline of the final report that it will be aiming for. It will be instructed to produce not more than a certain number of paragraphs.
๐ฏ. Each of the paragraph description is fed into a research process separately to produce a comprehensive set of information to be used in report construction. Detailed description of the research process will be outlined in the next section.
๐ฐ. All of the information will be fed into a summarisation step that will construct the final report including conclusion.
๐ฑ. The report will then be delivered to the user in MarkDown form.
Each of the research steps are following given flow:
๐ญ. Once we have the outline of each paragraph, it will be passed to a LLM to construct Web Search queries in an attempt to best enrich the information needed.
๐ฎ. The LLM will output the search query and the reasoning behind it.
๐ฏ. We will execute Web search against the query and retrieve top relevant results.
๐ฐ. The results will be passed to the Reflection step where a LLM will reason about any missed nuances to try and come up with a search query that would enrich the initial results.
๐ฑ. This process will be repeated for n times in an attempt to get the best set of information possible.
Detailed walkthrough Blog Post: newsletter.swirlai.com/p/building-deeโฆ
GitHub with implementation code and Notebooks to follow: github.com/swirl-ai/ai-anโฆ
Happy Building!
Be sure to leave a like or star if you find the content useful!
#AI #LLM #MachineLearning
Some weeks ago I released an episode of my Newsletter and an update to the ๐๐ ๐๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ ๐๐ฎ๐ป๐ฑ๐ฏ๐ผ๐ผ๐ธ GitHub repository.
There I implemented a Deep Research Agent from scratch without using any LLM Orchestration frameworks (using DeepSeek-R1 for some planning tasks).
In the project we implement the following Agentic topology:
๐ญ. A user provides a query or topic to be researched.
๐ฎ. A LLM creates an outline of the final report that it will be aiming for. It will be instructed to produce not more than a certain number of paragraphs.
๐ฏ. Each of the paragraph description is fed into a research process separately to produce a comprehensive set of information to be used in report construction. Detailed description of the research process will be outlined in the next section.
๐ฐ. All of the information will be fed into a summarisation step that will construct the final report including conclusion.
๐ฑ. The report will then be delivered to the user in MarkDown form.
Each of the research steps are following given flow:
๐ญ. Once we have the outline of each paragraph, it will be passed to a LLM to construct Web Search queries in an attempt to best enrich the information needed.
๐ฎ. The LLM will output the search query and the reasoning behind it.
๐ฏ. We will execute Web search against the query and retrieve top relevant results.
๐ฐ. The results will be passed to the Reflection step where a LLM will reason about any missed nuances to try and come up with a search query that would enrich the initial results.
๐ฑ. This process will be repeated for n times in an attempt to get the best set of information possible.
Detailed walkthrough Blog Post: newsletter.swirlai.com/p/building-deeโฆ
GitHub with implementation code and Notebooks to follow: github.com/swirl-ai/ai-anโฆ
Happy Building!
Be sure to leave a like or star if you find the content useful!
#AI #LLM #MachineLearning