Loading
Prism is an AI research pipeline that collects financial data, analyses fundamentals, and drafts publication-ready equity reports — so your desk can cover more names, publish faster, and maintain quality your clients can verify. Analysts govern every gate. The agents do the rest.
A research desk with five analysts can maintain deep coverage on maybe 30–40 names. Every mid-cap, every sector newcomer, every client request outside the universe gets the same answer: "we don't cover that." Coverage breadth is capped by headcount.
Quarterly results come out, the stock moves 8% in two sessions, and your research update is still in draft — waiting on data collection, XBRL reconciliation, and peer comparison. By the time it lands, the client has already acted or missed the window.
One analyst verifies every XBRL field and builds a proper DCF. Another eyeballs the numbers and writes from recall. Both reports land in the same format, same branding. The client — and the compliance desk — can't distinguish rigour from intuition.
The peer group rationale, the exceptional item adjustments, the sector context built over four quarters — it all walks out the door. The next analyst starts from zero. Clients lose continuity on names they hold.
It's a governed research pipeline that solves the three problems every research desk faces: not enough coverage for the names clients hold, reports that arrive after the market has moved, and quality that depends on which analyst wrote it. AI agents collect, analyse, and draft. Your analysts verify inputs and approve outputs — at structured gates, not after the fact.
AI research without governance is guesswork in a suit. Governance without AI is a desk that can't scale past 40 names. Prism delivers both.
AI agents do the work. Analysts hold the keys. Every report passes through structured approval gates before a single word goes out.
Enter a ticker and quarter. Maxwell, the data agent, collects price history, financials, corporate actions, shareholding patterns, exceptional items, and parses the annual report for RAG. The analyst's first task is reviewing data — not hunting for it.
Three approval gates: verify XBRL field mappings, confirm the peer group, and review collected data. Each gate is explicit — the pipeline pauses until the analyst approves. No AI output is built on unverified inputs.
Ramanujan analyses the financials. Faraday drafts seven research sections. Leavitt reviews for consistency. The analyst reviews every section, requests regeneration where needed, and approves the final report.
Every data point is sourced. Every approval is logged. The audit trail is the pipeline itself.
Maxwell collects everything an analyst needs to write a research report — automatically, from public sources, with full provenance. The analyst reviews, not reconstructs.
Current market price, 52-week range, and full historical price data — collected and structured automatically for every ticker.
Income statement, balance sheet, and cash flow — both consolidated and standalone, with XBRL field mapping verified by the analyst.
Splits, dividends, buybacks, and bonus issues — anything that affects share price continuity or valuation adjustments.
Promoter, institutional, and retail holdings over time. Changes in ownership structure are early signals for conviction or concern.
One-time gains, impairments, and restructuring charges — identified via XBRL tags or AI extraction. Critical for normalised earnings.
Each report is structured to the standard an equity research desk expects. AI drafts every section. The analyst approves, edits, or regenerates — one section at a time.
Company profile, industry positioning, and competitive landscape drawn from annual report and public filings.
Revenue trends, margin analysis, and balance sheet review — built from XBRL-verified quarterly and annual data.
Near-term and medium-term positive drivers — product launches, regulatory tailwinds, capacity expansion.
Regulatory, operational, and market risks — grounded in filings and sector context, not generic boilerplate.
DCF assumptions, peer multiples, and fair value estimates — with every input traceable to collected data.
The core argument for the stock — bull case, bear case, and the evidence supporting each.
BUY, HOLD, or SELL — with target price, current market price, and upside/downside quantified.
Four client problems. Four people who feel the pain. One pipeline.
A research desk with 5 analysts and Prism covers like a team of 15. Not because individuals type faster — because the pipeline handles data collection, financial analysis, and section drafting. Analysts spend their time on judgement, not grunt work. Clients get coverage on the names they actually hold.
Today, report quality depends on which analyst wrote it. Prism enforces the same pipeline on every report — XBRL verification, peer validation, structured section generation, review gates. The compliance desk can trust the process, not just the person.
Results drop and the pipeline starts immediately — data collected, financials reconciled, sections drafted. The analyst reviews and approves, not reconstructs. Reports reach clients in hours, not the week after the market has already repriced.
Every XBRL mapping, peer group rationale, exceptional item adjustment, and section correction lives in the pipeline — not in someone's head. When an analyst leaves, the next one picks up a company with four quarters of institutional memory intact.
Prism is the only product that combines automated data collection, AI-powered financial analysis, structured section generation, and analyst governance — in a single pipeline from ticker to report.
ChatGPT, Claude, Perplexity Finance
Generate text from training data or web search. No structured data pipeline, no XBRL verification, no analyst approval gates, no audit trail.
Bloomberg, Refinitiv, CapitalIQ
Provide the data. You still need to collect it, structure it, analyze it, and write seven sections. The terminal is a tool, not a pipeline.
Kensho, Alphasense, Visible Alpha
Automate parts of the workflow — data extraction or NLP search. None deliver a full pipeline from ticker to reviewed report with analyst governance at every stage.
They need more coverage, delivered on time, at a quality bar they can verify. Prism is the pipeline that makes that possible.
Get early accessFull annual report PDF parsed, chunked, and indexed for retrieval-augmented generation. Agents cite specific pages, not hallucinated context.
Named after the prism — it takes a single beam of data and refracts it into a full spectrum of insight.
Built by Sigfyn