CLIRUN and the tools you already use.
ChatGPT, Claude and the rest are built to help with almost anything. CLIRUN does one job. It traces how an event reaches an Indian market and puts a public source behind every step. Plenty of people will use both, so here is where they differ. We describe each product from its own documentation and link to it.
Checked on 21 September 2026
The short version. Choose by the job in front of you.
- Draft a memo, summarise a filing, think a problem throughChatGPT or Claude
- Have an agent work through your files and produce a deckClaude Cowork
- Build your own research pipelineCodex
- Check a live price or read an earnings call transcriptPerplexity Finance
- Find out whether one thing really reaches another, with the proof for each stepCLIRUN
- Decide what to buy or sellYou. None of these should make that call for you.
In detail. What each is good at, and where CLIRUN differs.
ChatGPT will research almost anything. CLIRUN researches one kind of question, under rules it won't bend.
What ChatGPT is good at
It drafts, edits and explains across any subject, and it's very good at a first pass. Its deep research mode browses the web for 5 to 30 minutes and writes a report with citations.1 Since February 2026 you can limit that search to sites you choose and bring in other data through apps and MCP servers.1
| ChatGPT | CLIRUN | |
|---|---|---|
| What you ask | Anything at all. | How one thing reaches another in an Indian market. |
| What you get back | A written report with citations. | A chain of typed links, each with its own quoted source, plus a reviewed answer and a report. |
| What supports a claim | The pages the model read and chose to cite. You decide how much weight each one deserves. | A sentence quoted word for word from a published document, stored with the time it was read. Sources are tiered by who issued them. |
| When the evidence runs out | At launch, OpenAI said deep research can make factual errors or incorrect inferences, and often fails to convey uncertainty.2 | The chain stops at the last evidenced link, and the answer lists what was searched. |
| What it costs | A subscription, with monthly deep research allowances that depend on the plan.1 | One monthly plan. Each analysis is quoted in credits before it runs. |
| Best for | Breadth, drafting and first passes. | A claim you'll have to defend. |
Using them together
Run the question in CLIRUN, export the report as Markdown, and ask ChatGPT to help turn it into your memo. The citations travel with it.
- 1Wikipedia, ChatGPT Deep Research. Duration, February 2026 changes and plan allowances.
- 2OpenAI, Introducing deep research, February 2025. Limitations section.
Claude is a careful generalist. CLIRUN is a specialist with a fixed standard of proof.
What Claude is good at
Claude reads long documents closely and writes clearly. Anthropic's financial services tools connect licensed data from providers including FactSet, S&P Global, Morningstar and PitchBook, and Anthropic says every claim links directly to its original source.12
| Claude | CLIRUN | |
|---|---|---|
| What you ask | Anything, including questions about documents and data you bring. | How one thing reaches another in an Indian market. |
| What you get back | Prose, analysis, and files such as documents and spreadsheets. | A chain of typed links with quoted evidence, a reviewed answer and a report. |
| What supports a claim | Citations to the pages or data it used, including licensed data if you've connected it.1 | Public documents only, quoted word for word and tiered by who issued them. |
| Data it can reach | Your uploads, the web, and the services and data providers you connect.2 | The public web. CLIRUN doesn't use licensed data feeds. |
| Best for | Reading, writing and reasoning over your own material. | Tracing a mechanism through the public record. |
Using them together
Ask Claude to write the analysis around a CLIRUN chain. The exported report gives it an evidence section that's already cited.
- 1Anthropic, Claude for Financial Services. Connected providers and source links.
- 2Claude Help Center, Install financial services plugins. Data providers and plans.
Cowork is an agent you direct across your files and tools. CLIRUN does one bounded job, and asks before it spends.
What Claude Cowork is good at
Cowork handles longer tasks that span files and apps. You point it at folders and tools, it shows each step as it works, and it hands back finished work for you to review. It has a built-in browser, connectors for Microsoft 365, Google Drive and Slack, and it can run tasks on a schedule.1
| Claude Cowork | CLIRUN | |
|---|---|---|
| How the work is directed | You describe a goal. Cowork plans the steps and works through them, and you can redirect it.1 | You ask one question. CLIRUN runs the same research method every time, and you can't loosen it. |
| Standard of evidence | Whatever you ask for, and whatever you check yourself. | A quoted public source for every link, enforced on every run. |
| Access it needs | The folders and tools you choose to share.1 | None. CLIRUN never reads your files, and search services never see your question. |
| Cost and control | Part of Claude's Pro, Max and Team plans.1 | A quote before each analysis. Nothing runs until you confirm. |
| Best for | Producing the deck, the spreadsheet or the document. | Deciding what the deck is allowed to claim. |
Using them together
Export a CLIRUN report as Markdown and give it to Cowork to build the committee deck. The quotes and their sources are in the file you hand over.
- 1Anthropic, Claude Cowork. Features, connectors, scheduling and plans.
Codex writes software. You could build your own version of CLIRUN with it.
What Codex is good at
Codex is OpenAI's software engineering agent. It reads and edits code, runs tests in a cloud sandbox or on your own machine, and proposes changes for you to review. Since July 2026 it lives inside the ChatGPT desktop app.1
| Codex | CLIRUN | |
|---|---|---|
| What you'd use it for | Building a pipeline that searches, extracts and summarises sources. | Asking a question and getting the finished chain. |
| What you maintain | Everything the pipeline needs: source rules, stored copies of pages, checks on the output, cost limits and upkeep. | Nothing. CLIRUN runs and maintains that pipeline. |
| Standard of evidence | The one you write. | A quoted public source for every link, tiered by who issued it. |
| Best for | Teams that want to own their research tooling. | People who want the answer rather than a new project. |
Using them together
If you do build your own tools, CLIRUN's Markdown exports are easy to feed into them.
- 1Wikipedia, OpenAI Codex (AI agent). Capabilities and the July 2026 merge with the ChatGPT desktop app.
Perplexity tells you what's happening in a market. CLIRUN explains how one thing reaches another.
What Perplexity Finance is good at
Perplexity Finance gives fast answers with sources alongside market data. For Indian markets it shows live NSE and BSE prices, explanations for notable price moves, historical data downloads, and live transcripts of company earnings calls.12
| Perplexity Finance | CLIRUN | |
|---|---|---|
| The question it answers | What's happening, and what are the numbers? | Does A reach B, and through what? |
| Live market data | Yes, including prices and earnings calls.1 | No, on purpose. A price is a reading that changes by the minute. CLIRUN keeps the mechanism and the evidence behind it. |
| What you get back | Answers with sources, and market data views. | A chain of typed links, each with its quoted evidence. |
| Best for | Keeping up with markets. | Understanding an exposure before you rely on it. |
Using them together
Use Perplexity for the latest figures. Use CLIRUN for the mechanism that explains why those figures matter to you.
- 1Analytics India Magazine, Perplexity Finance expands to Indian stock markets, 13 August 2025.
- 2TechCrunch, Perplexity now supports live earnings call transcripts for Indian stocks, 18 August 2025.
Doing it yourself is the method CLIRUN copies. The difference is who does the reading.
What doing it yourself is good at
Nothing beats your own judgment about which sources to trust and what they mean. The cost is time: finding the document, reading it, quoting the right sentence, noting the date, and remembering what you looked for and didn't find.
| Doing it yourself | CLIRUN | |
|---|---|---|
| Who reads | You do. | CLIRUN does, and you review what it found. |
| Record of what was checked | Whatever notes you keep. | Every quote stored with where and when it was found, plus what was searched and not found. |
| Judgment | Yours. | Still yours. CLIRUN shows evidence. It doesn't make the call. |
| Best for | Questions you know inside out. | Questions where you'd like the reading done first. |
Using them together
Let CLIRUN do the first pass, then spend your own time on the weakest link it names.
If we've got something wrong, tell us.
These products change quickly, and we'd rather be corrected than out of date. Every claim about another product on this page links to the page it came from, so you can check it the same way you'd check a CLIRUN chain.