SciSpace
Last reviewed
Jun 4, 2026
Sources
26 citations
Review status
Source-backed
Revision
v1 · 1,929 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 4, 2026
Sources
26 citations
Review status
Source-backed
Revision
v1 · 1,929 words
Add missing citations, update stale details, or suggest a clearer explanation.
SciSpace is an AI-powered research platform and literature-review assistant aimed at academics, students, and researchers. It lets users search a corpus of more than 200 million scientific papers, chat with PDFs through an AI "Copilot" that explains text, math, and tables, build literature-review tables that extract structured data from many papers at once, and draft and format manuscripts. The product is made by PubGenius Inc., an India-rooted company founded in 2015 by Saikiran Chandha and Shanu Kumar, and was known as Typeset (typeset.io) until a May 2022 rebrand to SciSpace. The platform pivoted hard toward generative AI in 2023 with the launch of SciSpace Copilot and by 2025 had repositioned itself around agentic, multi-step research tools.
The company behind SciSpace, PubGenius Inc., was started in 2015 by Saikiran Chandha and Shanu Kumar. Chandha, an alumnus of VIT, was selected for Stanford University's Ignite program, an entrepreneurship course, in 2015, where he connected with Kumar, a 2013 graduate of IIT Kharagpur who had also gone through Ignite. The first prototype was built at an AngelHack hackathon in Bengaluru, and the team went on to win AngelHack's Global Demo Day in San Francisco.
The original product, Typeset (typeset.io), was a research-writing and formatting tool, often described in early coverage as a "Google Docs for research communication." Its signature feature let authors write a paper once and then auto-format it to the template of any of thousands of academic journals in a single click, addressing a common pain point in scientific publishing. The product was publicly launched around mid-2017. By that point Typeset had drawn early traction: roughly three months after launch the company reported more than 11,000 researchers across 103 countries, about 5,000 new users a month, and 120-plus papers completed weekly, with the company saying around 95% of new users came through word of mouth.
Typeset's funding history is comparatively lean for a company of its later reach. In September 2017 it raised an $850,000 pre-Series A round. The round was anchored by Haresh Chawla, a partner at the private-equity firm True North (formerly India Value Fund), investing in his personal capacity; Chawla had first backed the company in 2015. Startup databases later list additional investors associated with the company, including Inventus Capital Partners and Silicon Valley Quad, with the most recent recorded round dated to December 2021. Reported totals vary by source: PitchBook lists about $3.85 million raised. The company has generally kept its capitalization low-profile and, as of late 2024, said it was operating profitably rather than relying on fresh venture rounds.
| Date | Round | Amount | Notes |
|---|---|---|---|
| 2015 | Initial investment | Undisclosed | Haresh Chawla (True North partner, personal capacity) |
| September 2017 | Pre-Series A | $850,000 | Anchored by Haresh Chawla |
| (per databases) December 2021 | Later round | Undisclosed | Investors include Inventus Capital Partners, Silicon Valley Quad |
In May 2022 the company rebranded Typeset to SciSpace, signaling a shift from a writing-and-formatting tool toward a broader, end-to-end research platform. Alongside the new name and visual identity, SciSpace opened free access to a large repository of research papers (then described as roughly 270 million articles with metadata, plus tens of millions of open-access PDFs) and pitched a single workspace to discover, read, write, edit, and publish research. The legacy typeset.io domain continued to redirect to SciSpace.
The decisive turn came with the broad adoption of large language models. In April 2023 SciSpace launched Copilot, an AI assistant that answers questions about research papers, summarizes sections, and explains difficult passages, equations, and tables in plain language. The tool built on large language model capabilities and, according to the company, combined third-party foundation models from OpenAI and Anthropic with proprietary models fine-tuned for scientific text. By mid-2023 the company reported around 400,000 active researchers on the platform and about 100,000 users of its newly launched browser extension, and was positioning Copilot as a "ChatGPT for researchers." In September 2023 Copilot added inline citation interlinking so that answers traced back to original sources, an approach aligned with retrieval-augmented generation.
Through 2024 and 2025 SciSpace layered on increasingly agentic features. In February 2025 it launched Deep Review, a multi-agent assistant that performs systematic literature reviews by iteratively searching, screening, and synthesizing papers into a first-draft review. The company also introduced a specialized Biomedical Agent for biology and medicine, and consolidated its tools into a broader "SciSpace Agent" that the company describes as connecting dozens of academic databases and around 150 research tools to automate multi-step scholarly tasks from a single prompt. A SciSpace mobile app for iOS was announced in February 2025 and introduced in early March 2025, with an Android app following.
SciSpace bundles several tools around the central idea of reading, searching, and writing scientific literature with AI assistance.
SciSpace's marketing consistently cites a corpus of more than 200 million research papers, and its product pages describe metadata coverage on the order of 280 million papers along with tens of millions (roughly 50 million) of open-access full-text PDFs. The exact figure has grown over time, from the ~270 million quoted at the 2022 rebrand to 280 million-plus in 2025 materials. The platform aggregates from open scholarly sources and data partnerships rather than hosting paywalled full text directly.
SciSpace is built on top of machine learning and large language models. The company has said it uses foundation models from OpenAI and Anthropic together with its own fine-tuned models for tasks specific to scientific writing and comprehension. Its agentic features (Deep Review, the Biomedical Agent, and the broader SciSpace Agent) reflect the wider industry move toward AI agents that plan and execute multi-step tasks. Earlier, as Typeset, the company's engineering team published work on applying natural language processing to scholarly documents for tasks like parsing and structuring papers.
SciSpace offers a free tier with usage limits alongside paid subscriptions. Its Premium plan has been listed at roughly US $20 per month when billed monthly, or about US $12 per month when billed annually, unlocking higher-quality models and effectively unlimited use of literature search, chat, the paraphraser, the AI writer, citations, and exports. The company has also offered a higher Advanced tier and a short trial window before billing. Prices and tier structure have changed over time and vary by region.
In September 2022 SciSpace partnered with Turnitin to integrate text-similarity and plagiarism detection into its research-writing workflow, letting authors check originality without leaving the platform.
By 2024 to 2025 the company described its user base as more than one million students, PhDs, and researchers across over 100 countries, drawing on users at institutions such as MIT, Stanford, Yale, and CERN, and counting partnerships with 150-plus publishing companies. SciSpace is frequently covered in roundups of AI research assistants and is commonly compared with tools such as Perplexity, Elicit, Consensus, Research Rabbit, and ChatPDF. As with other AI reading tools, reviewers caution that AI-generated summaries and answers can contain errors or hallucinations and should be verified against the underlying papers, a general limitation of large language model systems.
The founders and the company have received Indian startup-press recognition over the years, and Saikiran Chandha is regularly cited as the company's co-founder and CEO.
SciSpace (the AI research assistant from PubGenius / formerly Typeset) should not be confused with unrelated entities that use similar names, such as academic venues, journals, or other software products that have at various times used "SciSpace" or similar branding. This article refers specifically to the scispace.com research platform.