Jon Osaghae
AI Systems Developer — Agents, RAG & Automation
New to Contra
Jon is building their profile!
AI operations agent with permission-gated actions
Problem: Businesses want AI agents that take real actions — sending emails, updating records — but handing an LLM unsupervised write access is a legitimate trust problem most teams correctly won't accept.
Approach: Built a governed agent for a hospitality studio use case, with a decision layer that checks whether a request is a new action or a question about something that already happened before touching any tool — the failure mode that caused it to draft a new email when asked whether one had already been sent. Sensitive actions hold for human approval, every action is logged in a full trace, and the agent verifies an action actually succeeded before reporting it did.
Outcome: A working agent that does the reasoning while a human keeps the final say on anything that matters — nothing sent or changed without a trace, and no false "sent successfully" confirmations.
Tools: Next.js, Groq, Supabase, pgvector
0
5
AI lead-qualification system with tool-calling analysis
Problem: Inbound leads pile up unsorted — sales teams waste time manually reading every submission to figure out which ones are actually worth a callback.
Approach: Built a tool-calling system that runs parallel analysis on each lead (intent, fit, urgency) and drafts a qualified follow-up, routing results to Slack/email automatically. Forked into a law-firm-specific version with bilingual intake and a hardened no-legal-advice system prompt.
Outcome: Leads arrive pre-qualified with a drafted response ready for human review, instead of landing in an inbox as raw, unsorted submissions.
Tools: Next.js, Groq (parallel tool-calling), Supabase, Nodemailer, Slack Webhooks
0
25
Multi-document AI knowledge base with cross-source citations
Problem: Internal knowledge — policies, contracts, onboarding docs — is scattered across files nobody can search, so answering a simple internal question takes minutes of manual digging or a person to ask.
Approach: Built a RAG system that ingests multiple documents into one searchable knowledge base and answers plain-English questions with the exact source cited — including questions that require pulling and synthesizing facts from two different documents at once.
Outcome: Stress-tested against a dense, realistic document set, answers that used to take minutes of manual search now return in seconds, correctly cited, including answers that span multiple source documents.
Tools: Next.js, Voyage AI embeddings, Supabase + pgvector, Groq Llama 3.3 70B
0
31
Lets.ng (http://lets-ng.vercel.app) — Shortlet Booking SaaS (Founder & Full-Stack Developer)
Built a SaaS platform tailored for Lagos shortlet operators, replacing Instagram DM chaos with a personal storefront link. Operators list properties, customers browse and initiate WhatsApp routed booking inquiries directly. Next.js 14 App Router, Supabase auth and database, RLS, middleware route protection, Tailwind v4. Live at lets-ng.vercel.app (http://lets-ng.vercel.app).
0
69