PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchUnni T A’s project portfolio traces a shift from making an LLM generate an answer to building applications that can retrieve information, use tools, manage memory, and handle workflow failures. In a DEV Community article published September 29, 2026, the author describes seven projects across research, finance, appointment automation, compliance, and enterprise retrieval. The descriptions are the author’s account; they do not independently verify each project’s implementation or production behavior.
What ties these AI projects together?
The projects address different jobs, but the recurring design question is what an AI system needs around its model to do useful work: access to information, a way to act on it, checks on its output, and rules for handling errors and sensitive data. The author frames the work as a progression beyond a simple prompt-and-response interface, toward applications that combine tools, retrieval, memory, evaluation, security, and conventional software logic.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL160_.jpg)
As an Amazon Associate I earn from qualifying purchases.
“My goal is not just to make an LLM generate an answer,” the author writes. That is a useful lens for the projects below: they are not all the same kind of agent or RAG system. Some are described as tool-using workflows, some combine structured records with document retrieval, and one is explicitly an architecture demonstration rather than a finished system.
Recommended Free Tools
Seven projects, organized by the work they do
Deep Research Agent: gather, check, and synthesize information
The Deep Research Agent is described as a multi-stage research workflow. It takes a topic, generates search queries, gathers information, extracts facts, identifies gaps, searches again, and produces a structured report. The author names FastAPI, LangGraph, Tavily, and ChromaDB as part of the project.
#1 Best Overall
The described features extend beyond the initial report: export, follow-up questions, summaries, counterarguments, translation, and job history are included in the account. The author also says the project has Docker deployment and an Ollama-based local mode. These details illustrate a research pipeline that can revisit missing information rather than treating the first model response as the finished product.
Autonomous Financial Research Agent: combine sources and memory
This project is described as a ReAct-style agent for financial research. Its source material includes SEC EDGAR filings, earnings transcripts, financial data, and news; its reported tasks include sentiment analysis, peer comparisons, fact checking, and calculations.
The author describes working memory, semantic memory implemented with FAISS, and episodic memory for earlier runs. The account also lists conflict resolution, PII redaction, prompt-injection protection, rate limiting, and evaluation. These features point to the operational questions that arise when an agent works across potentially conflicting financial sources—not just how it generates a summary.
Autonomous Dental Appointment Bot: connect conversation to actions
The appointment bot is described as supporting booking, rescheduling, and cancellation across web, SMS, WhatsApp, and voice. Its named components include PostgreSQL, Redis, Celery, Stripe, and Google Calendar.
The author highlights slot locking, payment webhooks, duplicate-event handling, logging, health checks, and error handling. Those are application concerns with direct consequences for a scheduling workflow: a conversational interface must not merely propose a time; it must coordinate availability and external services without creating duplicate or inconsistent bookings.
NexusBase: an enterprise RAG architecture
NexusBase is described as an enterprise retrieval-augmented generation architecture using Next.js, FastAPI, LangGraph, PostgreSQL, and pgvector. The author mentions query routing and retrieval evaluation, emphasizing the steps between a question and the context presented to a model.
Rank #3
The article’s publication-time status is specific: the backend is labeled functional, while the frontend was being redeployed. That is a time-bound description from the author’s article, not confirmation of the project’s current state.
MedComply: compliance-oriented application structure
MedComply is presented as a medical-compliance SaaS monorepo with a Next.js frontend, FastAPI backend, and Supabase migrations. The author lists organizations, users, documents, authentication, role-based access control, document processing, and AI-assisted analysis among its components.
This description puts the AI feature inside a larger product structure: identity, permissions, and document handling are part of the application, rather than details that can be delegated to a prompt.
Aequitas FI: pair structured analysis with document retrieval
Aequitas FI combines structured financial data with RAG. The author describes a separation between SQL analysis and document retrieval, using LangGraph, PostgreSQL, and pgvector. Reported capabilities include temporal comparison, PII redaction, audit logging, human feedback, and automated testing.
This is a distinct pattern from relying on document search alone. Structured questions can be handled through SQL, while unstructured source material can be retrieved separately; the two paths can contribute to the same financial analysis without being treated as interchangeable.
Context Synthesizer: a demonstration with important gaps
Context Synthesizer is presented as an enterprise RAG architecture demonstration for a possible retrieval workflow across Slack, Jira, Google Drive, and Notion. It should not be mistaken for a complete deployed connector-based system. The article explicitly says live connectors, the vector database, embedding pipeline, backend retrieval engine, authentication, and production LLM inference are not implemented.
Best Value
That distinction matters: an architecture demonstration can communicate how a system might be organized, but it does not establish that the integrations or end-to-end retrieval behavior are working in production.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the project mix says about building AI applications
Across the portfolio, the work shifts with the task. Research agents need search and fact extraction; a finance agent needs multiple source types and ways to handle disagreement; appointment automation must coordinate state and external services; compliance software needs application-level access controls; and enterprise RAG needs retrieval and evaluation. Aequitas FI makes the boundary between structured SQL analysis and document retrieval especially explicit.
The author’s reflections emphasize a set of practical design decisions: what information a system can access, which tools it may use, what it should remember, how retrieval works, how results are checked, how failures are handled, when the agent stops, how sensitive information is treated, and how the system is deployed. These are the author’s lessons from the projects, not independently measured findings or a guarantee that each listed safeguard has been validated.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Source and status
The project descriptions and reflections in this article are drawn from Unni T A’s DEV Community post, “My Journey Building AI Agents, RAG Systems, and AI-Powered Applications,” published September 29, 2026. Implementation details and status labels above are attributed to that post; they are not an independent audit of repositories, demos, or production systems.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
