Sleuth vs. ChatGPT, Claude, and Copilot

Give Claude a biopharma library

Keep ChatGPT and Claude, and connect Sleuth. They’ll work from a biopharma knowledge base of programs, trials, companies, and deals, with a source on every value.

Keep Claude, and connect Sleuth for the landscape

On their own, ChatGPT and Claude are strong analysts without a library of strategy data. Connect Sleuth, and they work from a biopharma knowledge base instead of rebuilding each landscape from the open web.

On their own

A fast first pass from whatever pages a web search finds that day. Some of it’s wrong, and it’s difficult to tell where.

With Sleuth connected

The same chat, working from every program, trial, and deal in a space, with a source on every value. In our tests, AI tools found 63% more of the landscape.

Claude for Life Sciences

Built for the lab, not the strategy deck

Anthropic, OpenAI, and Google each sell AI for life sciences, but each one is built for research and discovery: papers, protein structures, notebooks, and bench data. The pipelines, deals, sell-side research, and Chinese registries that strategy work runs on aren’t the focus of any of them.

Claude

Claude for Life Sciences connects it to papers, compound databases, and lab notebooks, for scientists designing experiments.

ChatGPT

OpenAI’s life sciences model is built for biology and chemistry: reading papers, planning experiments, and analyzing lab data.

Gemini

Google’s science tools connect Gemini to protein structure and genomics databases, for generating hypotheses.

The strategy library, in the same chat

A strategic decision can shift based on data you’d otherwise miss: a competitor’s program, a deal that sets the market price, a Chinese program nobody flagged. Sleuth adds those records to the same chat, matched to one record per drug, and PubMed can stay on for the papers.

150K+

drugs, by target, modality, and phase

1M+

trials, from 20 registries

400K+

companies, public and private

200K+

deals, with terms where they’re disclosed

About 22,000

Chinese programs, from Chinese-language sources

35+

sell-side research providers

How ChatGPT and Claude compare, alone and with Sleuth

With Sleuth connected

ChatGPT and Claude on their own

Built for

With Sleuth connected

Strategy, BD, and CI work: landscapes, partner and asset searches, deal comps, and indication choices.

ChatGPT and Claude on their own

Early, lab-side science: biology, chemistry, and papers.

Where a landscape comes from

With Sleuth connected

One knowledge base of 150K+ drugs, 400K+ companies, and 200K+ deals, matched to one record per drug.

ChatGPT and Claude on their own

Rebuilt from whatever pages a web search finds, every time you ask.

Trials

With Sleuth connected

1M+ trials from 20 registries, China’s included, linked to the same drug records as programs and deals.

ChatGPT and Claude on their own

Claude connects to ClinicalTrials.gov. Registries outside the US, China’s above all, are hit or miss.

Deals

With Sleuth connected

200K+ licensing, M&A, collaboration, and financing deals, with upfronts, milestones, and royalties where they’re disclosed.

ChatGPT and Claude on their own

Press releases and news a web search finds, where terms are often missing.

Sell-side research

With Sleuth connected

Research from 35+ providers, included in every plan. You don’t need an account with each bank.

ChatGPT and Claude on their own

The AI labs don’t license it, and it’s paywalled.

China

With Sleuth connected

Sleuth reads China’s registries in Chinese and puts each of about 22,000 programs into your landscape in English.

ChatGPT and Claude on their own

ChatGPT and Claude rarely reach China’s registries, which are in Chinese.

Checking a value

With Sleuth connected

Open the project in Studio and click any value to see the filing, registry entry, or broker note behind it, with a confidence score. Sleuth flags the few cells worth checking.

ChatGPT and Claude on their own

Claude cites its sources, but some of what it cites is wrong, and it’s difficult to tell where.

Next month

With Sleuth connected

Turn on monitoring, and the same dataset versions up as trials, filings, and deals come in, with a change log.

ChatGPT and Claude on their own

Ask again and it starts over, so two runs can differ.

AI tools found 63% more with Sleuth connected

We asked Claude, ChatGPT, and Perplexity five landscape and deal questions, once on their own and once with Sleuth connected. All 15 tests found more with Sleuth, 63% more on average.

63%

more of the landscape found with Sleuth connected, on average across 15 tests

+55%

Claude

Alone

With Sleuth

+27%

ChatGPT

Alone

With Sleuth

+108%

Perplexity

Alone

With Sleuth

What changed in the answers

A whole class it missed

Claude’s IL-23 landscape left out the TL1A × IL-23p19 bispecifics and the oral IL-23R drugs. With Sleuth, both were in, and its list grew 97%.

The hard columns, filled

Claude marked royalty terms “not disclosed” on 52% of its deal rows. With Sleuth, 89% of rows had real terms.

One drug, one row

ChatGPT listed CPO102 and SYSA1801 as two drugs. They’re the same CSPC ADC, and with Sleuth they became one row.

Past the paywall

Perplexity said comprehensive partnering databases “are generally paywalled” and stopped at 18 deals. With Sleuth, it found 150% more.

Example

Ask in the chat, and keep the dataset

A chat answer is a block of text. With Sleuth connected, the same question comes back as a dataset you can check, keep, and export.

“Which Phase III MASH programs include F4 patients and are positioned to add on to a GLP-1?”

Drug

Company

Mechanism

Fibrosis population

Incretin positioning

Fits the question

efimosfermin alfa

GSK

FGF21

F2–F3 and F4

GLP-1 add-on

Yes

efruxifermin

Novo Nordisk

FGF21

F2–F3 and F4

GLP-1 add-on

Yes

pegozafermin

Roche

FGF21

F2–F3 and F4

GLP-1 add-on

Yes

survodutide

Boehringer Ingelheim

Incretin-based

F2–F3 and F4

Incretin backbone

Near miss: it’s the incretin

Columns Sleuth builds for you

efimosfermin alfa

CompanyGSK

MechanismFGF21

Fibrosis populationF2–F3 and F4

Incretin positioningGLP-1 add-on

Fits the questionYes

efruxifermin

CompanyNovo Nordisk

MechanismFGF21

Fibrosis populationF2–F3 and F4

Incretin positioningGLP-1 add-on

Fits the questionYes

pegozafermin

CompanyRoche

MechanismFGF21

Fibrosis populationF2–F3 and F4

Incretin positioningGLP-1 add-on

Fits the questionYes

One row per program

Every program that fits, not the ten a chat has room for.

A source on every value

Click a cell to see the trial, filing, or broker note behind it.

Yours to keep

Saved to your Studio account, and exported to Excel, a deck, or an interactive analysis.

A first draft, or a dataset you can put your name on

On its own, Claude gives you a first draft from the open web, and it’s difficult to tell which parts are wrong. With Sleuth connected, every value is sourced and scored, so you check only the few cells Sleuth flags.

Claude alone

A first draft from the open web. Some of it’s wrong, and it’s difficult to tell where.

Studio, or Sleuth in Claude and other AI tools

Decision-grade. Sleuth sources and scores every value, so you only look at the few flagged cells.

Concierge and Custom datasets

Production-grade. Every delivery is checked before it reaches you, so there’s nothing left for you to review.

How a value gets checked

Deep dives built from what AI tools can’t reach

Questions about Sleuth, ChatGPT, and Claude

Why not just use Claude?

Claude is a strong analyst without a library. Ask it for a landscape and it rebuilds one from web searches each time. It can’t reach China’s registries in Chinese or sell-side research, and it can’t match thousands of records to the same drugs the same way every run. Some of what it finds is wrong, and it’s difficult to tell where. Sleuth did that work ahead of time, and it works inside Claude, so you can keep using it.

Doesn’t Claude for Life Sciences already do this?

No. Claude for Life Sciences connects Claude to the lab: papers, compound databases, and lab notebooks. The pipelines, deals, sell-side research, and Chinese registries a landscape or partner search needs aren’t in it. Sleuth adds those, and you can use both in the same chat.

What about Gemini or Copilot?

Sleuth works in Copilot through Copilot Studio, the same way it works in Claude and ChatGPT. Gemini wasn’t in our benchmark. Any AI tool that supports MCP can connect to Sleuth with the server URL.

What about deep research modes?

Deep research modes search more of the open web, but they still can’t reach sell-side research or China’s registries in Chinese, and they rebuild the list from scratch each time. Our benchmark ran each app at its strongest standard setting: Claude Opus 5 and GPT-5.6 Sol at high effort, and Perplexity’s Best.

Is Sleuth built on Claude?

Sleuth uses frontier models from several AI labs, Anthropic’s among them, and picks the one that does each step best. The difference is the knowledge base the models work from.

Does connecting Sleuth cost extra?

No. Every plan includes Sleuth in Claude, ChatGPT, and Copilot, and analysis costs the same there as in Sleuth Studio. Your AI tool calls Sleuth only when a question needs it.

Connect Sleuth to the AI tool you already use

Set up in Claude, ChatGPT, or Copilot