AI Lab

A place to test AI ideas before they become products

Not everything we try becomes a product. The Lab is where we run controlled experiments on assistants, automations and applied AI — with each entry honestly labelled as Experiment, Internal, Prototype, Private Beta, Public or Archived.

Workflow copilot

Internal

Problem: Internal admin tasks eat time that should go into product work.
Hypothesis: A narrow copilot with strong context can remove specific repeated steps.
Tools: LLM API · TypeScript · Supabase
Current state: Used internally on selected workflows.
Next step: Measure time saved on a defined task before considering it a product.

Match-summary generator

Experiment

Problem: Post-match write-ups are time-consuming for community moderators.
Hypothesis: Structured match data plus a constrained prompt can produce useful drafts.
Tools: LLM API · NoLagZA data model
Current state: Prototype run against sample match data.
Next step: Evaluate quality with real moderators before exposing publicly.

Support triage assistant

Prototype

Problem: First-line support questions repeat and pull attention from complex ones.
Hypothesis: A retrieval-based assistant can answer the repeatable ones with sources.
Tools: LLM API · Embeddings · Docs corpus
Current state: Prototype limited to internal documentation.
Next step: Trial on a scoped support surface with clear escalation paths.

Applied AI for your business

Turn a Lab idea into a workflow

If a specific problem in your workflow keeps burning time, that's usually where applied AI earns its keep.

Talk to DigiAI