Financial Research with Clarifo and Claude: A Practical Workflow

Financial research is still often organised around documents: find the latest filing, download the PDF, search for a number, copy it into a spreadsheet, and then try to work out what actually changed. Claude makes the conversation easier, but the quality of the answer still depends on whether it can reach the right source, fiscal period, metric and context.

With Clarifo connected to Claude, the research can move directly from a question to company filings, reported figures and source-backed analysis without switching between separate tools.

For financial research, the division of labour is clear: Claude frames the question and explains the conclusion; MCP provides the connection; and Clarifo supplies filings, structured financials, calculations, reporting context and provenance. The workflow becomes useful when each layer does its specific job and the evidence remains visible.

Financial AI is moving into specialist workflows

The market is beginning to separate the conversational model from the financial systems behind it. Anthropic and OpenAI are packaging AI around specific finance workflows; the practical question is what evidence, permissions and outputs sit behind the model.

Anthropic starts with the advisor's day

Claude for Financial Advisors is aimed at wealth-management practices. Anthropic describes a set of connectors and workflow skills that bring custodians, portfolio platforms, CRMs, planning tools and meeting context into Claude. The intended work is practical: prepare for a client meeting, review a portfolio, flag a compliance issue, update a CRM or draft a follow-up. Investment recommendations, client communications and compliance decisions still require the advisor's review and approval.

OpenAI starts with the research desk

ChatGPT for Financial Services is a separate ChatGPT plan for eligible financial institutions, initially focused on investment banking and equity research. It combines frontier models with built-in premium financial data and connected sources, lets users trace figures and claims back to supporting tables and passages, and can turn research into editable financial models, research notes and pitchbooks.

Clarifo focuses on the evidence layer

Clarifo gives Claude direct access to company filings, financial history, segment data and source-backed calculations. Instead of finding reports, extracting figures and piecing the data together manually, Claude can retrieve the relevant evidence through Clarifo and spend more of the workflow on analysis. The result is faster research with a clear path back to the underlying filings.

The NVIDIA case: from reported growth to an investment question

Instead of asking Claude to "summarise NVIDIA", ask a question that contains an investment or research decision:

That single question contains several smaller tasks: establish the long-term growth curve, identify the point of structural change, check whether the business mix has changed, read the latest quarter, search for counter-evidence and test the valuation. Clarifo can supply the source material for each task while Claude keeps the narrative coherent.

1. Find the inflection point in the financial history

The first request might be:

The returned series shows why FY2024 matters. Revenue moved from $27.0 billion in FY2023 to $60.9 billion in FY2024, then to $130.5 billion in FY2025 and $215.9 billion in FY2026. Operating income and net income accelerated even faster than revenue.

The analytical point is not merely that revenue grew. It is that the company crossed from a cyclical semiconductor growth pattern into a new demand regime. Clarifo's inflection analysis helps identify the break; Claude's role is to explain what changed and to test whether the filing evidence supports calling it structural.

NVIDIA's FY2026 Form 10-K describes the company as a full-stack, data-center-scale AI infrastructure business. It also describes CUDA, networking, systems and software as parts of the platform. That evidence is more informative than a generic sentence saying that "AI is a growth driver". Read the NVIDIA FY2026 Form 10-K.

2. Check whether the current business is still the same business

The next Clarifo request moves from history to composition:

For Q2 FY2027, NVIDIA reported $96.221 billion of revenue. Data Center contributed $89.023 billion, or approximately 92.5% of the total.

The implication is more specific than saying that "AI is important". NVIDIA's exposure is now highly concentrated in hyperscale and enterprise AI infrastructure. That concentration helps explain the growth, but it also changes the risk profile: a slowdown in data-centre investment, a customer redesign or a supply constraint can affect the group more directly than when Gaming and other businesses represented a larger share.

3. Anchor the story in the latest quarter

A third prompt asks Claude to connect the current quarter to the longer trend:

The important research move is to connect the quarter to the longer story. Q2 FY2027 confirms that the FY2024 inflection did not immediately disappear. It also shows why the risk analysis must become more sophisticated: NVIDIA is no longer only selling components. It is planning capacity, networking, systems and long-duration commitments around the AI buildout. The primary source is NVIDIA's Q2 FY2027 Form 10-Q.

4. Ask for counter-evidence, not only confirmation

A useful research workflow should deliberately look for evidence that could weaken the emerging story:

The latest filing changes the risk discussion materially. NVIDIA says its supply and capacity commitments increased from $119 billion in the prior quarter to $279 billion as of July 26, 2026. It also discloses $36 billion of AI cloud commitments and $25 billion of equity investment commitments. Separately, guarantees related to the SB Energy PORTS Technology Campus can reach $105 billion in aggregate under specified conditions.

The analytical question is therefore broader than whether AI demand is strong. NVIDIA is increasingly securing supply, cloud capacity and infrastructure around that demand. The next step is to monitor whether those commitments continue to translate into revenue and cash generation without creating excess capacity, counterparty exposure or lower returns.

Repeat the workflow with any company

You do not need to know the company before you start. The workflow is designed to begin with a simple question and become more specific only when the evidence gives you a reason to go deeper.

You can also ask a simple question

The full workflow is useful when you want a complete research case, but it should not be a requirement. If you already know what you want to investigate, ask directly:

Clarifo provides the filings and financial data behind the question. Claude can then follow the evidence and decide which parts deserve deeper investigation.

Turn the finished research into a visual

Create the visual only after the analytical point is clear. The chart should show the evidence behind one conclusion, not decorate the page.

For a long-term change, use the financial history. For a changing business mix, use the reported segment or market-platform data. For the latest quarter, use the filing's reported comparison. For risks, visualize only figures that are directly supported by the filing and avoid combining exposures that are not economically comparable.

This article is for information and research-method purposes. It is not investment, legal or financial reporting advice. Periods, figures and filing status should be verified against the linked primary sources before decisions are made.