Alphaholic started from a familiar problem. The evidence behind a view lives in too many places: a filing, an expert-call note, a spreadsheet, a saved article. They all bear on the same question, and pulling them back together takes longer than forming the view did.
The problem
The hard part of research is rarely finding one more piece of information. It’s keeping its context, tying it to the company or question it belongs to, and finding it again when the view changes.
I wanted a workspace for the whole loop: collect material, read it, check it against other evidence, write down a view, and come back to it later.
What it is
Alphaholic connects research sources, company information, notes, files and AI help, so less work gets repeated between reading, organising and analysing.
- Research: material from many sources in one stream, easy to browse and to find again.
- Companies: financials, disclosures and research context, organised around each business.
- Memos and files: working notes that stay useful after the conversation that produced them.
- AI: summaries, retrieval and research tools inside the workflow, plus agent-facing APIs so the same tools work from outside it.
The loop
Collect, connect, investigate, decide, revisit. The useful unit is a question with its evidence attached, not an answer on its own.
An AI summary speeds up the first pass. What it’s worth depends on whether I can go back to the source, check the statement and decide what it means. That’s why source context and the original material run through the whole product.
A few choices
- Keep the way back to the source. A fluent answer is only a step.
- Build around recurring tasks. Company work, notes and reading each need their own context.
- Show missing data and failed processing. An uploaded file is not automatically a searchable one.
- Keep access to material in line with the permissions it came with.
Keeping score
Ideas from my AI and technology coverage go into a market-neutral paper portfolio, with the record published on manage.alphaholic.app. It’s paper, not client money; the point is to hold myself to my own calls.
My part
I work on the product and the code: the interface, data workflows, integrations, and how the AI tools fit into everyday tasks. The direction comes from using it in my own research every day. The next useful change is rarely a new model. More often it’s a familiar workflow made clearer, or easier to come back to.
