Selected work
A small set of real engagements, anonymised at the client’s request, including work led by the founding team prior to operating as Axial Foundry.
Below the featured engagements, additional capability themes show the breadth the studio can deliver across.
Featured engagements
Applied AI · Conversational Systems
A multi-turn customer support assistant that handles enquiries on WhatsApp, grounds answers in product documentation, and stays connected to the distributor’s sales workflow.
Challenge
The distributor’s sales and support team was processing repetitive product enquiries, quote follow-ups, and order status questions across WhatsApp and email — with no central record, no automated triage, and inconsistent response times that slowed the sales pipeline.
What was built
Built a conversational assistant on Flask with OpenAI-powered logic, a retrieval layer over the distributor’s product catalogue and FAQ corpus, and Twilio-driven WhatsApp delivery. Conversations were stored against contact records so the sales team could pick up where the assistant left off, and automated follow-up workflows were layered on top to keep slow-moving conversations active without manual prodding.
Outcome
Improved client engagement rates on follow-ups and removed manual triage from the sales team’s daily workflow. The assistant operates as the first responder on the channel, escalating to humans when the question moves outside its scope.
Delivered
Stack
Client name withheld at their request. Engagement led by the studio’s founding team prior to operating as Axial Foundry.
Data Infrastructure · Automation
A scraping and monitoring layer collecting structured data on more than one million restaurants, with operational dashboards built so the client’s team could see pipeline health at a glance.
Challenge
The platform needed continually-refreshed structured data on a very large set of hospitality venues sourced from public listings. Existing collection was brittle, single-threaded, and offered no visibility — silent failures and partial scrapes were corrupting downstream analytics.
What was built
Engineered a scraping infrastructure capable of running across the 1M+ venue set, hardened against rate limits and structure changes. Built Django-based monitoring dashboards backed by SQL so the operations team could watch run health, freshness, and exception rates in real time. Data was normalised on ingestion and surfaced cleanly to the client’s analytics layer.
Outcome
Replaced an unreliable collection setup with infrastructure the client could trust at scale. Pipeline visibility shifted from “nothing came in today, why?” to a dashboard view of every run’s health.
Delivered
Stack
Client name withheld. Engagement led by the studio’s founding team prior to operating as Axial Foundry.
Additional capability themes
Representative of the breadth of work the team can pick up beyond the two featured engagements above.
AI Systems
Representative of RAG-powered question answering work built to make internal or public-facing knowledge easier to access through grounded retrieval.
Speech AI
Speech-oriented system work spanning transcription, speaker diarization, and usable output formatting for operational or research contexts.
Model Adaptation
Representative of model evaluation, instruction tuning, continual pretraining, and benchmarking work used to understand model behavior and quality.
Software Products
Product-oriented engineering work focused on the application layer around user journeys, data handling, dashboards, and business logic.
Ready to talk?
A premium website, an internal tool, a product MVP, a knowledge assistant, or a deeper AI implementation — Axial Foundry helps you shape it, ship it, and keep improving it.
