Thesis

Cover Block

PUBLIC

Attribute	Value
Company Name	Thesis
Tagline	Autonomous AI research lab platform accelerating AI discovery 10x faster/cheaper
Headquarters	San Francisco, United States
Founded	2025
Stage	Seed
Business Model	SaaS
Industry	Deeptech
Technology	AI / Machine Learning
Geography	North America
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding Label	Seed
Total Disclosed	~$500,000

Links

PUBLIC

• Website: https://thesislabs.ai
• LinkedIn: https://www.linkedin.com/company/thesis-labs-ai
• Y Combinator Profile: https://www.ycombinator.com/companies/thesis

Executive Summary

PUBLIC Thesis is an early-stage platform that aims to automate the scientific discovery process in artificial intelligence, positioning itself as an autonomous research lab that could significantly compress the timeline for developing foundational models [Y Combinator, Fall 2025]. Founded in 2025 and backed by Y Combinator, the company's central proposition is to treat AI research as an intelligent search problem, a methodology intended to make state-of-the-art discovery ten times faster and cheaper [Y Combinator, Fall 2025]. Its product vision, articulated as an applied AI lab, targets a critical bottleneck for academic institutions, startups, and independent researchers by seeking to systematize the experimental workflow that leads to breakthroughs like AlphaFold [Y Combinator Launch, 2026].

The founding team, Sergio Charles and Luigi Charles, leads a core of two people, and the company's initial capitalization consists of the standard Y Combinator seed investment of $500,000 [Y Combinator, Fall 2025]. Operating as a SaaS business model, Thesis is in a pre-product, concept-validation phase with no publicly disclosed customers, deployments, or external press coverage beyond its accelerator profile. For investors, the next 12-18 months will be defined by the translation of this ambitious technical vision into a demonstrable product, the recruitment of initial research partners, and evidence that its automated systems can generate novel, verifiable results in machine learning.

Data Accuracy: YELLOW -- Core claims sourced from Y Combinator profile and company website; team size and funding round are consistent across primary sources. No independent third-party validation of product or traction.

Taxonomy Snapshot

Axis	Classification
Stage	Seed
Business Model	SaaS
Industry / Vertical	Deeptech
Technology Type	AI / Machine Learning
Geography	North America
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding	Seed (total disclosed ~$500,000)

Company Overview

PUBLIC

Thesis emerged from Y Combinator's Fall 2025 batch, positioning itself as an autonomous AI research lab platform. The company's stated mission is to accelerate AI discovery by treating the research process as an intelligent search problem, aiming to make state-of-the-art breakthroughs 10x faster and 10x cheaper [Y Combinator, Fall 2025]. Co-founded by brothers Sergio Charles and Luigi Charles, the company is headquartered in San Francisco and was founded in 2025 [Y Combinator, Fall 2025] [Crunchbase, 2026].

Key milestones to date are limited to its accelerator participation and initial funding. The company's primary public milestone is its acceptance into Y Combinator, which provided a standard $500,000 seed investment in late 2025 [Y Combinator, Fall 2025]. No subsequent product launches, customer announcements, or partnership disclosures have been made public beyond the initial YC launch profile.

The team currently consists of the two co-founders [Y Combinator, Fall 2025]. There are no open job postings listed on the company's careers page, which remains active but shows no positions [Thesis Labs, 2026]. The absence of public hiring activity or team expansion news since the YC batch suggests the company is in a pre-product, foundational building phase.

Data Accuracy: YELLOW -- Company details confirmed via Y Combinator profile; team size and founding details are single-sourced.

Product and Technology

MIXED Thesis frames its core product as an autonomous AI research lab, a platform designed to treat the process of AI discovery as an intelligent search problem [Y Combinator, Fall 2025]. The company's stated mission is to make state-of-the-art AI discovery ten times faster and ten times cheaper, with the explicit goal of enabling researchers to discover foundational models like the next Transformer or AlphaFold [Y Combinator Launch, 2026]. This positions the platform as a tool for applied AI labs aiming to accelerate the frontier of AI discovery.

The specific product mechanics are described at a high level. The company is building AI systems that automate parts of the machine-learning research workflow, which includes planning experiments and training and evaluating models [FYI Combinator, 2026]. This suggests a system that orchestrates the iterative, computationally intensive cycles of hypothesis testing and model iteration that characterize modern ML research. The wedge appears to be scaling breakthroughs by providing AI-native data science tools, though the exact user interface and degree of autonomy remain unspecified.

No product launch, demo, or detailed technical architecture has been publicly announced. The technology stack is not described on the company's website or in available sources. The platform's current capabilities and the nature of its "autonomous" functionality are defined solely by the company's own claims.

Data Accuracy: YELLOW -- Product claims sourced from company and YC materials; no independent technical review or user validation.

Market Research

PUBLIC The ambition to automate scientific discovery represents a frontier in applied AI, moving beyond task-specific models toward systems that can formulate and test hypotheses, a shift that could unlock new vectors of productivity in R&D.

Quantifying the total addressable market for an autonomous research platform is challenging, as the category is nascent. Thesis frames its mission around accelerating AI discovery itself, which places its initial serviceable market within the global AI research and development sector. According to a 2025 report from Grand View Research, the global artificial intelligence market size was valued at approximately $196.6 billion, with a compound annual growth rate (CAGR) of 36.6% from 2024 to 2030 [Grand View Research, 2025]. While this figure encompasses hardware, software, and services broadly, it provides an analogous market context for the scale of investment flowing into AI infrastructure and tools. A more direct, though still adjacent, proxy is the market for machine learning operations (MLOps) platforms, which IDC estimated at $4 billion in 2023 and forecast to grow to $8 billion by 2026 [IDC, 2024]. Thesis's proposition sits upstream of operationalization, targeting the discovery phase of the ML lifecycle.

Demand drivers are anchored in the escalating cost and complexity of frontier AI research. Training large language models now routinely requires capital commitments in the hundreds of millions of dollars, concentrating advanced research within a handful of well-funded corporate labs [Stanford AI Index, 2025]. This creates a structural tailwind for tools that promise to increase researcher productivity and lower the marginal cost of experimentation, potentially democratizing access to state-of-the-art discovery. Concurrently, the proliferation of open-source models and scientific datasets has increased the raw material available for automated systems to explore, a necessary precondition for an intelligent search approach.

Key adjacent markets include both traditional scientific computing software and emerging AI-for-science (AI4Science) platforms. The former, dominated by established players like MathWorks and simulation specialists, addresses deterministic modeling. The latter, exemplified by initiatives from large tech companies and startups, applies AI to specific scientific domains like drug discovery or material science. Thesis's positioning appears more horizontal, focusing on the meta-problem of the research workflow itself rather than a vertical application. Regulatory and macro forces are presently neutral but bear watching; increased scrutiny on AI development, particularly around compute-intensive training runs, could influence the operational environment for automated research platforms.

Data Accuracy: YELLOW -- Market sizing relies on analogous, third-party reports for broader AI and MLOps categories; direct TAM for autonomous AI research is not publicly defined.

Competitive Landscape

MIXED Thesis enters a nascent but rapidly formalizing category of AI-native research tools, positioning itself as a fully autonomous platform rather than an incremental productivity aid.

Given the early stage of the company and the category, a direct, feature-for-feature competitor comparison is not yet possible. However, the competitive map can be segmented into three layers: direct platform competitors, established research infrastructure, and adjacent workflow tools.

• Direct platform competitors. This is the most speculative layer, populated by other startups aiming to automate or accelerate AI discovery. Sakana AI, cited as a competitor, represents a different model: an AI research lab that builds its own foundation models, not a platform sold to external researchers [Y Combinator, Fall 2025]. Its positioning is as a producer of AI, not an enabler. Other potential entrants in this autonomous research platform space are not yet publicly named, but the ambition suggests eventual competition with any startup seeking to productize "AI that builds AI."

• Established research infrastructure. This includes the entrenched toolchains used in academic and industrial AI labs today. Frameworks like PyTorch and TensorFlow, coupled with experiment trackers such as Weights & Biases or MLflow, represent the incumbent workflow. Thesis's proposed differentiation is not to replace these components but to orchestrate them autonomously, framing the entire research process as an intelligent search problem [Y Combinator, Fall 2025]. The competitive edge here would be a step-function improvement in researcher productivity, moving from assisted tracking to autonomous hypothesis generation and testing.

• Adjacent workflow and automation tools. This segment includes code-generation assistants (GitHub Copilot, Codium), cloud-based AI development platforms (Google Vertex AI, Amazon SageMaker), and AI-powered data science tools. These are substitutes that address parts of the researcher's workflow,writing code, managing compute, analyzing data,but do not claim to automate the high-level research strategy. Thesis's bet is that integrating and surpassing these point solutions creates a new category.

Thesis's defensible edge today is almost entirely conceptual and talent-based. The platform's core premise,treating AI discovery as a search problem to be solved by AI,is a specific technical thesis that, if executable, could create a structural advantage [Y Combinator, Fall 2025]. This is coupled with the early validation and network access provided by Y Combinator, whose partner Tom Blomfield is listed as the primary contact [Y Combinator, Fall 2025]. At this pre-product stage, however, this edge is perishable; it depends entirely on the founding team's ability to translate the concept into a working system that demonstrates the promised 10x acceleration.

The company's most significant exposure is its lack of a moat against well-capitalized incumbents or fast-following startups. Any advantage in algorithm design or system architecture could be replicated if the underlying concept proves valuable. Furthermore, Thesis does not own the distribution channel to researchers. Adoption would require displacing entrenched workflows and convincing time-constrained academics and industry labs to trust an autonomous system with their research direction, a high barrier to change. A competitor like Weights & Biases, with deep integration into existing workflows and strong researcher loyalty, could decide to layer autonomous capabilities on top of its tracking suite, leveraging its existing distribution.

The most plausible 18-month competitive scenario hinges on proof of a unique, hard-to-replicate capability. If Thesis can demonstrate a novel search algorithm or optimization technique that yields a published breakthrough in collaboration with a credible research institution, it could establish early thought leadership and attract a dedicated user base. In this scenario, a "winner" could be an adjacent tool provider that moves quickly to partner with or acquire Thesis to embed its autonomy layer. A "loser" could be a generic AI infrastructure startup that fails to move beyond compute orchestration into the higher-value research intelligence layer, finding itself boxed into a commodity offering.

Data Accuracy: YELLOW -- Competitive mapping is inferred from company positioning and one named competitor; most analysis is conceptual due to pre-product stage.

Opportunity

PUBLIC Thesis's opportunity rests on capturing a portion of the multi-billion dollar market for AI research acceleration, a prize that grows as the cost and complexity of frontier AI development escalate.

The headline opportunity is to become the default operating system for AI research labs, a category-defining platform that systematizes the discovery of novel machine learning architectures. The company's mission to make state-of-the-art AI discovery 10x faster and cheaper directly targets a critical bottleneck [Y Combinator, Fall 2025]. If successful, Thesis would not be just another tool but the foundational layer upon which future breakthroughs like AlphaFold or the Transformer are built, embedding itself into the R&D workflows of both academic institutions and commercial AI labs. The Y Combinator backing and the framing of AI research as an "intelligent search problem" signal an approach that is both ambitious and methodical, moving beyond incremental workflow automation toward a new paradigm for discovery [Y Combinator, Fall 2025].

Two or three growth scenarios, each named, The path from a two-person seed-stage startup to a platform of consequence hinges on specific, plausible inflection points.

Scenario	What happens	Catalyst	Why it's plausible
Academic Wedge	Thesis becomes the standard tool for computational research in top-tier AI/ML university departments and national labs.	A landmark research paper from a partner institution is published using Thesis to discover a novel model, demonstrating 10x efficiency gains.	The platform's stated mission aligns with the resource constraints of academic research [Thesis Labs, 2026]. YC's network provides connections to early-adopter faculty.
Commercial Lab Expansion	The platform is adopted by emerging AI startups and scaled into the internal workflows of large tech companies' research divisions.	A successful deployment at a YC peer AI startup leads to a case study showing reduced time-to-model and cost, triggering inbound from larger players.	The company already lists collaborations with "leading institutions, startups, and independent researchers" as a target, indicating this go-to-market motion [Y Combinator, Fall 2025].

What compounding looks like, The core compounding mechanism is a data and algorithmic flywheel. Each research project run on the platform generates structured data on experiment design, model performance, and compute efficiency. This proprietary dataset, absent from public repositories, can be used to train the platform's own recommendation and search algorithms, making them more effective at guiding future research. Over time, this creates a data moat: the platform that has orchestrated the most successful experiments becomes the most intelligent at planning the next one. Early signs of this flywheel are not yet public, but the product's foundational premise,treating research as a search problem,is inherently iterative and data-driven [Y Combinator, Fall 2025].

The size of the win, A credible comparable is the trajectory of AI infrastructure companies that achieved unicorn valuations by becoming essential to the research and development lifecycle. For instance, Weights & Biases, a platform for machine learning experiment tracking and collaboration, reached a reported $1.25 billion valuation in 2021 [Forbes, 2021]. Thesis aims at a more ambitious, upstream layer of the stack,not just tracking experiments but autonomously planning and executing them. If the "Academic Wedge" scenario plays out and Thesis captures a significant portion of the global AI research budget, a valuation in the hundreds of millions to low billions is plausible (scenario, not a forecast). The total addressable market is the billions spent annually on AI R&D compute and researcher time, a portion of which Thesis would aim to capture as platform revenue.

Data Accuracy: YELLOW -- The opportunity framing is derived from the company's stated mission and market context, but specific TAM figures, customer traction, and flywheel evidence are not yet publicly available.

Sources

PUBLIC

1. [Y Combinator, Fall 2025] Thesis (thesislabs.ai) - YC Company Profile | https://www.ycombinator.com/companies/thesis

2. [Thesis Labs, 2026] Thesis Homepage | https://thesislabs.ai

3. [LinkedIn, 2026] Sergio Charles - Thesis (YC F25) | LinkedIn | https://www.linkedin.com/in/sergiocharles/

4. [LinkedIn, 2026] Luigi Charles - Thesis (YC F25) | LinkedIn | https://www.linkedin.com/in/luigicharles/

5. [LinkedIn, 2026] Thesis (YC F25) | https://www.linkedin.com/company/thesis-labs-ai

6. [Y Combinator Launch, 2026] Launch YC: δ Thesis | Y Combinator | https://www.ycombinator.com/launches/OnF-thesis

7. [FYI Combinator, 2026] Thesis | FYI Combinator | https://fyicombinator.com/company/thesis

8. [Crunchbase, 2026] Sergio Charles - Co-Founder & CEO @ Thesis Labs - Crunchbase Person Profile | https://www.crunchbase.com/person/sergio-charles-c7e3

9. [Thesis Labs, 2026] Mission | Thesis Labs | https://www.thesislabs.ai/mission

10. [Thesis Labs, 2026] Careers | https://thesislabs.ai/careers