Kray.AI search and document assistant
2021–2024
An AI search and document-assistant platform sold as Kray.AI, with a WooCommerce plugin and a demo store.
Cloud and Platform
Elasticsearch
I built Elasticsearch search indices for Kray AI and related products to power full-text.
See its threads in the worldUsed on
Kray AI product search
2024–2026
An AI product search for e-commerce stores that replaces the store's keyword search with intent and specification matching.
E-commerce search for a mechanical-seal manufacturer
2023
Graph-backed e-commerce search over a mechanical-seal catalogue, for customers and staff finding parts by nomenclature, type or compatibility.
Site search and chatbot for an industrial-caster manufacturer
2023–2025
AI site search and chatbot for a caster manufacturer's e-commerce site, delivered under the Kray.ai product name.
Kray store rollouts
2023–2026
Work that puts Kray, an AI e-commerce search product, into real stores: a store cloner, demo store and grocery rollouts.
Keyword Country research product
2019–2020
A keyword research product that expands one seed phrase into a large, tagged keyword set with search-volume data.
Industrial fastener marketplace
2026
Online marketplace for industrial fasteners with a single canonical product catalogue and competing seller offers, so trade buyers can compare prices.
Product search for a fastener distributor
2026
A product search for a stainless-steel fastener distributor that understands buyers' abbreviations and thread standards.
Market-analysis tool for a market-intelligence firm
2020–2023
A tool that shows who competes for a market in chosen locations and how visible each competitor is, for a market-intelligence firm.
SEO analysis platform
2016–2021
SEO analysis product that crawls a customer site, analyses its pages and turns findings into assignable tasks for marketing teams.
Stories where it mattered
2024 · Kray.AI search and document assistant
One model for everyone, or one per customer
- Situation
- A document assistant for a fleet operator needed an architecture, and one model trained on every vessel's documents was the assumed default.
- Task
- Lay out the real trade-offs of cost, latency and access control instead of just picking one.
- Action
- I wrote both options out for client and team: one shared model versus one model per vessel, with pros, cons and the replica arithmetic.
- Result
- The partitioned design went forward, with the loss of cross-fleet questions accepted openly.
Goes with
Full-Text Search
Vector Search
Embeddings
Python
API DesignThrow me a paper plane
Write a line; I’ll find it in my inbox.