Perplexity for Financial Research: Spaces and Deep Research

A Bristol-based investment research firm with twelve analysts came to us with a specific bottleneck. Each analyst covered four to six companies. Monthly, they needed to update their coverage with recent news, earnings calls, regulatory filings, and analyst note summaries. The preparation phase before they could write a view took three hours per company. The writing itself took 45 minutes. The maths was wrong.

They were using ChatGPT for the writing, but doing the research manually: opening tabs, reading articles, copy-pasting into documents, cross-referencing. The Head of Research had heard Perplexity described as a research tool with citations. He was sceptical. Two weeks after we set up Perplexity Spaces for their coverage universe, research preparation time was under 40 minutes per company.

This guide explains why Perplexity is different from general-purpose AI for research tasks, and how to configure Spaces and Deep Research for a finance context.

Why Perplexity for Finance (Not Just ChatGPT)

The fundamental difference between Perplexity and a general-purpose AI like ChatGPT is how they handle information. ChatGPT responds from its training data, which has a knowledge cutoff and cannot access current web information unless specifically enabled. Perplexity queries the live web for every response and cites its sources inline.

For financial research, this distinction is significant. A ChatGPT response about a company's recent performance might be accurate for a period six months ago. A Perplexity response pulls from recent earnings announcements, regulatory filings, news coverage, and analyst commentary published this week.

The Bristol firm ended up using Perplexity and ChatGPT in sequence: Perplexity to gather and structure current information, ChatGPT to analyse and draft the investment view based on the Perplexity output. Each tool doing what it does best.

For a broader comparison of AI tools for finance, see our guide on ChatGPT vs Copilot vs Claude for finance teams.

Setting Up Perplexity Spaces for Research

Perplexity Spaces is a shared workspace that combines document upload, source configuration, and collaborative querying. Unlike a basic Perplexity search, a Space can be configured with specific sources to prioritise, uploaded documents for context, and instructions that shape how the AI synthesises its responses.

Finance Space configurations that work

Configuring Space instructions

Each Space can include an instruction set that tells Perplexity how to synthesise results. For a company coverage Space, useful instructions include: present information chronologically (most recent first), highlight any earnings beats or misses versus consensus, flag any material changes to guidance, and always cite the date of each source.

These instructions do not replace professional judgement. They ensure the output is structured consistently enough that an analyst can scan it quickly and identify what needs deeper investigation.

Deep Research for Company Diligence

Perplexity Deep Research runs an extended, multi-step web query across many sources before synthesising a structured report. Unlike a standard Perplexity query (which runs one search pass), Deep Research iterates across sources, identifies gaps in its findings, and runs additional queries to fill them before producing output. The result is a longer, more comprehensive report with 20-40 citations.

For finance teams, Deep Research is most valuable for: initial company diligence before a meeting, sector deep-dives before starting new coverage, and M&A target background research from publicly available sources.

Deep Research query structure for finance

The quality of Deep Research output depends on prompt specificity. A vague prompt produces a general summary. A structured prompt produces a report that directly answers the questions you need answered.

Produce a diligence brief on [Company Name] covering:

1. BUSINESS MODEL: Core revenue streams, customer segments,
   geographic mix. Recent changes to the model.

2. RECENT PERFORMANCE: Last two reported periods. Revenue
   trend, margin trajectory. Any guidance changes.

3. COMPETITIVE POSITION: Main competitors. Market share
   narrative. Any recent shifts in competitive dynamics.

4. KEY RISKS: Regulatory, competitive, operational. Any
   recent developments that have changed the risk profile.

5. MANAGEMENT AND GOVERNANCE: Key leadership changes in
   the last 12 months. Any governance concerns.

Prioritise information from the last 12 months.
Cite the source and date for every material claim.
Flag where information is limited or conflicting.

This prompt takes five minutes to run. The output provides a structured first-pass view of a company that would previously have taken two to three hours of manual research to assemble. The analyst's role shifts from gathering to evaluating: reading the Deep Research output critically, identifying what to probe further, and forming a view rather than summarising facts.

Limitations to keep in mind

Deep Research draws from publicly indexed sources. Paywalled content (FT, Bloomberg Premium, broker research) is not accessible. For companies with limited English-language public coverage, results will be thinner. And for any figures you plan to use in client-facing work, verify against primary sources: the company's own filings and announcements, not Perplexity's synthesis of them.

The Bristol firm's standard operating procedure was: use Deep Research for the first pass, verify any specific figure in the company's most recent filing before putting it in an investment note. This took five minutes of verification for what used to require two hours of source gathering.

Competitor Monitoring with Spaces

Competitor monitoring is one of the highest-value Perplexity use cases for finance teams. Most teams either monitor competitors manually (time-consuming and inconsistent) or not at all (they react when something has already happened). Perplexity Spaces makes structured competitor monitoring achievable for teams without a dedicated intelligence function.

Competitor Space structure

Create one Space per competitive category (not per competitor, unless you have a small, well-defined set). A financial services firm might have one Space for retail banking competitors, one for wealth management peers, and one for fintech challengers in their core market.

Upload to each Space: your most recent competitive benchmarking analysis, any strategy documents that define how you assess competitive positioning, and a roster of the specific competitors in scope. Configure source prioritisation to include competitors' investor relations pages, relevant financial press, and sector analyst reports.

Weekly competitor briefing workflow

This workflow replaces an inconsistent manual process that might have run monthly (too infrequent) or not at all. Running it weekly takes 15 minutes instead of an hour, which means it actually happens.

Market and Sector Analysis Workflows

Beyond company-specific research, Perplexity is useful for building quickly updated sector views. Finance teams that produce quarterly market commentary, investment committee briefings, or board papers on market conditions use Perplexity to gather and structure the market evidence that supports their analysis.

Sector briefing query structure

Produce a sector briefing on [Sector] covering the last quarter:

1. MACRO DRIVERS: Key economic factors affecting the sector
   this quarter. Include interest rate, FX, commodity inputs
   as relevant.

2. DEMAND TRENDS: Evidence of demand strength or weakness.
   Volume data, consumer sentiment, order books where available.

3. SUPPLY AND COST: Input cost trends, supply chain
   developments, labour market conditions specific to sector.

4. REGULATORY DEVELOPMENTS: Any new regulations,
   consultations, or enforcement actions in the quarter.

5. PERFORMANCE SNAPSHOT: How sector indices or representative
   companies have performed vs broader market.

6. OUTLOOK: Forward indicators. Consensus analyst view
   if available.

Source from reputable financial and sector media.
Cite and date every material data point.

The output feeds directly into investment committee papers and board market updates. The finance team's role is validation, contextualisation, and judgment rather than evidence gathering. This is the right allocation of skilled analyst time.

For building the analysis layer on top of this research, combining Perplexity's output with Claude Projects for the narrative drafting follows the same pattern described in our Claude Projects for finance workflows guide.

Investment Research Firm: 3 Hours to 40 Minutes

The Bristol investment research firm covered forty companies across four sectors. Twelve analysts, each covering four to six companies, producing monthly coverage updates and quarterly deep-dive reports. The research preparation phase — gathering current news, earnings commentary, and analyst reactions — was eating time that should have been spent forming views.

We configured Perplexity Spaces for their forty coverage companies across three weeks. Each Space was set up with the company's investor relations page as a priority source, three to four relevant financial news outlets per sector, and the analyst's most recent coverage note uploaded as context. Standard monthly briefing queries were written for each sector and saved as templates in the relevant Space.

Monthly update workflow after implementation: run the saved briefing query in the company's Space (two minutes), review the structured output for material developments (ten minutes), verify any figures against primary filings (five minutes where relevant), write the updated view using ChatGPT drafted on the Perplexity output (twenty-five minutes). Total: under 45 minutes per company, down from three hours.

The Head of Research noted that quality improved alongside speed. The Perplexity output consistently surfaced developments that manual reading missed — typically items buried in trade press or secondary analyst commentary that a time-pressured analyst would have skipped. The citation structure meant that anything flagged could be verified in thirty seconds.

Annual analyst time saved: approximately 1,800 hours across the team. This was reinvested into expanding coverage by six companies and improving the depth of quarterly deep-dive reports.

Frequently Asked Questions

Related guides