Your database, queryable in any language.
AI agents that read your real content and answer your users with sources, not hallucinations.
How it works
Connect your data sources
CMS, database, API: wherever your content lives.
I build the retrieval layer
RAG, embeddings, vector search, tuned to your content.
Embed the chat
Drop the widget on your site, or expose it via API.
Use cases
Internal AI chats connected to real product data
Two examples of custom assistants I worked on: one for editorial research, one for DeFi analytics and chart generation.
Cryptoast internal tool
Cryptoast: news, guides and rankings, queryable
News articles, long-form guides, live crypto rankings, prices, market caps and volumes are all searchable from one internal chat, with citations back to the source content.
Example question
What's Solana's market cap, and which guide should I read first?
Example answer
Solana sits at $X.XB market cap as of today. For a primer, start with "What is Solana?" then move on to the staking guide.
DefiLlama client work
DefiLlama: LlamaAI turns DeFi data into answers and charts
I designed LlamaAI and participated in its creation. It can fetch the DefiLlama database, answer analytics questions and generate unique charts directly from the data.
Example question
Create a chart with the top 10 protocols ranked by daily revenue.
Example answer
LlamaAI can query DefiLlama data, rank the top protocols by daily revenue, then render the result as an interactive React chart component instead of a static answer.
What's included
Custom retrieval pipeline
Tuned to your content: chunking, embeddings, ranking.
Source citations on every answer
Every claim links back to the article, doc or row it came from.
Embeddable widget
Drop in a single script tag, or query via API.
Monitoring + iteration
I watch real questions and improve the retrieval over time.
Pricing
Setup
$5,000–10,000
One-off. Depends on data sources, volume, and complexity of the retrieval pipeline.
Running costs
$150–600/mo
API + vector DB hosting. Scales with traffic; capped to a budget you set.
Tell me about your project
A few sentences is enough to start. I reply within 24 hours.