AI for Financial Research
Ask Archivist a question, get answers grounded in exchange filings all over the world.
- NYSE…
- NYSE American…
- NASDAQ…
- TSX…
- TSX-V…
- CSE…
- NEO…
- Borsa İstanbul…
CoverageLast Updated: …
… Symbols… Filings
What sets us apart?
In plain words
Groundwork
We do the hard data work for you. Filings are collected, organized and indexed before you ask, so your research starts on solid ground, not on faith.
We check answers against both the filings and the web and keep only what holds up. When the evidence is not there, we tell you instead of guessing.
Filings
Filings are where companies go on the record, footnotes and all. We think they are worth reading, so we make reading them easy.
An AI tool on its own skims summaries of summaries. Ours reads the original filings, so every answer comes from the record itself.
Experience
Every statement links to its source. One click opens the filing beside your answer, scrolled to the passage and highlighted, so checking takes seconds.
And we size the work to the question: quick lookups stay quick, deep reads go deep. [6] Hover any citation to see the key details before you open the filing.
Where we fit
We are not a terminal, and we do not try to be one. We are the filings layer beside the tools you already use, at a small fraction of their price.
We shine when the answer is specific and buried: a covenant in a footnote, a metric few vendors extract, or one question asked across many companies.
In full detail, whenever you want it
As told by the team
Built by people who read the footnotes.
We are a small team with experience across a wide array of fields, from deal making and corporate finance to mining, energy, law and cybersecurity. All of it taught us the same thing: the market is loud with noise and opinion, while the facts sit quietly elsewhere.
Accuracy comes down to how many eyeballs get to check the work, so we put yours on every answer. Each one cites its source and takes you to the exact passage.
The companies we followed, with different motivations and questions, across different geographies and languages, required a better search tool that nothing quite answered. So we built the thing we wanted.
The Mosaic Finance teamFounders of ArchivistMore about usHow It Works
Ask
Type a question about any public company in natural language. Add filters to your desire or not, Archivist will figure it out.
Get Cited Answers
Every answer comes with citation badges linked to the exact source. Hover any badge to preview the passage, click to open the filing.
Verify
Click any citation to open the original filing. The PDF viewer highlights the exact section so you can verify every claim against the source document.
There is more to it for you to discover. Get started with your first 10 queries on us.
Behind the Architecture
We talk about what we're building in our blog. Transparency is the product.
A comprehensive overview of common AI misconceptions and the factual realities that should guide engineering decisions.
The math on autonomous agents is brutal. Chain 10 steps at 95% accuracy each and your total success rate is 60%. The hype ignores exponential decay.
If your system breaks because you changed an adjective, the problem isn't your prompt; it's your architecture. The mature approach is System Engineering.
Fine-tuning is effective for teaching form but terrible for injecting facts. The 'domain-specific fine-tuned model' pitch is mostly marketing.
Million-token context windows sound impressive. The reality: models struggle to retrieve information buried in the middle of long prompts.
The 'blackbox' narrative is convenient but wrong. AI models are sequences of math operations. Control the arithmetic, control the output.
LLMs predict the next token. That's it. They've mastered linguistic form without possessing communicative intent.
If code can do it, code should do it. There's a concerning number of AI automations that shouldn't be AI automations.
The industry conflates 'bigger' with 'better.' This ignores basic math. Running a cheaper model multiple times beats expensive single passes.
A comprehensive overview of common AI misconceptions and the factual realities that should guide engineering decisions.
The math on autonomous agents is brutal. Chain 10 steps at 95% accuracy each and your total success rate is 60%. The hype ignores exponential decay.
If your system breaks because you changed an adjective, the problem isn't your prompt; it's your architecture. The mature approach is System Engineering.
Fine-tuning is effective for teaching form but terrible for injecting facts. The 'domain-specific fine-tuned model' pitch is mostly marketing.
Million-token context windows sound impressive. The reality: models struggle to retrieve information buried in the middle of long prompts.
The 'blackbox' narrative is convenient but wrong. AI models are sequences of math operations. Control the arithmetic, control the output.
LLMs predict the next token. That's it. They've mastered linguistic form without possessing communicative intent.
If code can do it, code should do it. There's a concerning number of AI automations that shouldn't be AI automations.
The industry conflates 'bigger' with 'better.' This ignores basic math. Running a cheaper model multiple times beats expensive single passes.