What happens when you fall into a rabbit hole on Wikipedia? Often it looks something like starting on a page about Melinda Gates and ending two hours later reading about the Renaissance (not such a big stretch really).
The process of research isn’t too dissimilar. Sometimes a root folder will hold other relevant information nested somewhere deep inside it – but that journey down itself holds value because of the context you gather on the way to that target piece of information.
At SpikeJam we don’t fall, we leap into rabbit holes. So we followed our curiosity here and built a tool that would amplify our natural patterns of research.
We spent time speaking to different types of people that had a recurring workflow based around research. We settled on the small world of policy consultants that advise businesses on regulatory and political affairs. After some cold outreach (see our thoughts on this here), we found a design partner (who we’ll call Jane) with a pain in this very area.
Working side by side with our design partner:
Working side by side with users as they walk through their work has been the single highest leverage activity we’ve done. Breaking down the barriers and showing that we’re there to understand and help has resulted in much higher quality conversations, with feedback that users might not otherwise be comfortable sharing over a Zoom call.
She had a weekly recurring deliverable that involved researching, summarising, and formatting recent developments, on a topic relevant to each client, into a concise newsletter. This was an entirely manual task and took nearly 4 hours a week to complete. We knew this was a good problem to solve when she showed us the custom GPT she duct-taped together that helped with some of the last mile.
There were other research tools out there but we found the summaries couldn’t just be of the recent news article or latest legislation, they needed to summarise the development given the wider context of the market and policy landscape.
And so, Supafetch was born (naming credits, as always, go to Andy):
Supafetch is a tool that combines web scraping from chosen sources on the web, with the semantic analysis of AI, allowing users to ingest sources they trust and generate content based on specific sources of knowledge.
With Supafetch, the most valuable element of research becomes the actual selection of the information sources, as opposed to their interpretation. Said otherwise – with the output quality remaining consistent, the quality of the input became far more important.
We worked with her in the weeks leading up to the US election, so as an example, with Supafetch she could ingest articles profiling both candidates, transcripts of speeches they had made, and documents with policy positions from both campaigns. Having this greater context surrounding the topic allows for more robust summaries that draw beyond a single source.
As we were building out Supafetch, we realised that we had created something that held far more value outside of this narrow use case. We decided to broaden the scope of the product to address not just researchers, but people who must refer to text to inform their work, opening it up to a wider audience.
The potential use cases are quite broad. One that stands out is for regulated workplaces or workflows. For example, if your range of motion is constrained to a couple of options outlined somewhere in a Standard Operating Procedure, but you have thousands of those and other reference points and other sources of truth, being able to quickly get to the relevant text is incredibly value.
If quickly getting to insights is valuable in your workflow, play around with the tool here and let us know what you think.
Andy and Anton come from an engineering and commercial background respectively, they intersect in their obsession with product.
Taking their joint experiences in growth and product — the two areas they spike in — Andy and Anton are working on what their shared curiosity leads them to. To hear where their curiosity led them to next, sign up for notifications on the next post.