inAiinAi
Partners

Academia / Research

Research collaboration with inAi for universities, labs, researchers, professors, doctoral candidates, independent researchers, and research groups.

Email researchAGI as a System

Research collaboration with inAi

inAi studies AI as it moves from models and demos into systems, tools, work, products, agents, and public life.

This page is for universities, research labs, professors, doctoral candidates, independent researchers, research groups, and technically serious collaborators who want to explore research with inAi.

We are especially interested in work that connects intelligence, agents, tools, memory, knowledge, decision-making, business operations, Open Source, and real-world AI products. We are not trying to create research for appearance. We are interested in research that helps clarify how AI systems work, where they fail, how they can be evaluated, and how they can become useful in the world.

For the broader company mission, read AI in the Real World. For the AGI framing behind our research layer, read AGI and AGI as a System. For how we think about maturity, boundaries, openness, and public claims, read How We Build.

AI in the Real WorldAGIAGI as a SystemHow We Build

Who this page is for

This path is for people and organizations that approach AI research seriously.

This page is not only for formal institutions. Independent researchers can be relevant if the work is serious, specific, and connected to inAi's research and product architecture.

  1. A university lab studying AI, agents, intelligence, human-computer interaction, software systems, knowledge work, business operations, or AI evaluation.
  2. A professor or researcher interested in collaboration, supervision, review, co-authored work, seminars, or applied research questions.
  3. A doctoral candidate or graduate researcher working on agents, tool use, memory, evaluation, knowledge creation, AI-native software, decision systems, or organizational AI.
  4. An independent researcher with serious work and a clear proposal.
  5. A research group interested in AGI as a systems problem rather than only a model-capability problem.
  6. An applied AI researcher interested in product-facing consequences of AI systems.
  7. A lab, institute, or academic program interested in Open Source tools, AI education, public understanding, or real-world AI adoption.
  8. A research partner who can help design, evaluate, critique, or explain AI-native systems.

How inAi thinks about research

Research at inAi is connected to the company thesis: AI is becoming a new operating layer for software and work.

That shift cannot be understood only through product marketing. It needs research into how intelligence behaves, how agents make decisions, how knowledge is created, how tools change reasoning, how software changes when AI becomes an operator, and how organizations adapt when AI systems enter real workflows.

inAi's public AGI stance is that general intelligence may not emerge from one isolated model alone. It may emerge from systems of models, agents, tools, memory, perception, execution, coordination, feedback, and environment.

That stance does not mean inAi claims to have AGI. It means the company studies intelligence as a systems problem and builds products in a way that respects that view.

Research collaboration should strengthen that understanding.

Research directions

inAi's public research layer is organized around four directions.

Limits of Intelligence

This direction studies where intelligence succeeds, where it fails, and what limits appear when models, agents, tools, memory, context, and environments interact. Possible collaboration themes include limits of model reasoning, limits of agentic planning, evaluation of long-running AI work, failure modes in AI-assisted decision-making, reliability across multi-step workflows, hallucination, uncertainty, verification, context limits, memory limits, and intelligence as a system rather than a single capability score.

Limits of Intelligence

Agentic Decision Systems

This direction studies how agents make decisions, use tools, follow goals, recover from errors, maintain state, and operate within constraints. Possible collaboration themes include agent decision loops, tool use and tool selection, stateful workflows, agent memory, permissioning and review, planning and replanning, task decomposition, action reliability, interfaces and documentation designed for agents, and human-selected tools now and agent-discovered tools later.

Agentic Decision Systems

AI for Knowledge Creation

This direction studies how AI changes research, writing, synthesis, explanation, learning, analysis, and the production of new knowledge. Possible collaboration themes include AI-assisted literature review, AI-supported research workflows, knowledge synthesis, public explanation of technical ideas, reasoning across documents and sources, quality control in AI-generated knowledge work, human-AI collaboration in writing and research, and accessible explanations of AI for broad audiences.

AI for Knowledge Creation

AI and Business Operations

This direction studies how AI enters real organizational work: product data, catalog operations, supplier information, decision support, workflows, compliance support, documentation, and business processes. Possible collaboration themes include AI in business workflows, product-data and catalog systems, supplier-information processing, evidence-linked outputs, decision support in operations, human review and AI support, measuring AI value in real workflows, AI-native product design for organizations, and practical limits of automation in business settings.

AI and Business Operations

Areas where research collaboration may fit

AGI as a systems question

inAi is interested in research that treats general intelligence as a system-level problem. That can include models, agents, tools, memory, perception, execution, coordination, feedback, environment, human supervision, software interfaces, and the systems around AI models. A good research proposal in this area does not need to claim that AGI is solved. It should help clarify what kinds of systems may be necessary, how they can be evaluated, what failure modes matter, and what product consequences follow from that view.

AGI as a System

Agents and software use

AI agents use software differently from humans. Humans use screens, buttons, dashboards, and manual navigation. Agents need callable tools, structured inputs and outputs, readable instructions, state, memory, predictable actions, permissions, and recovery paths. Research in this area can help define what agent-ready software means and how agents should interact with tools, workflows, documentation, memory, and external systems.

Products for Agents

Evaluation and reliability

AI systems must be evaluated in the context where they are used. A model answering one question is not the same as an agent working through a multi-step process. A catalog workflow is not the same as a personal career workflow. A research assistant is not the same as an autonomous tool-using system. We are interested in evaluation work that respects those differences.

Limits of Intelligence

AI-native software

AI-native software is not just old software with an AI feature attached. It may involve different input flows, output structures, interfaces, user expectations, review layers, data handling, tool use, and action models. Research can help answer what changes when intelligence becomes part of product logic, what should stay under user control, what should be automated, what should be structured for agents, what needs evidence, review, or reversibility, and how maturity and limits should be communicated.

Products for Agents

Public AI understanding

AI is too important to be understood only through technical jargon, fear-driven headlines, or promotional claims. inAi's AI for Everybody layer exists to explain AI clearly for broad audiences. Research partners can help make public explanation more accurate, more grounded, and more useful.

AI for Everybody

Open Source and research tooling

inAi releases selected tools and experiments in the open when they can be useful beyond inAi. Research collaboration may connect to Open Source when the work benefits from public code, examples, documentation, evaluation tools, or reproducible demonstrations.

Open Source

Types of collaboration

Research conversations

Some collaborations begin as focused conversations: a research topic, a paper discussion, a method review, a research-direction critique, or a question around evaluation, agents, or AI-native products. These conversations are useful when they are specific. A message that says only “let's collaborate on AI” is usually too broad.

Academia / Research

Research notes and papers

Some work may become public research notes, essays, articles, papers, or other research outputs around AGI as a System, agentic decision systems, limits of intelligence, AI-assisted knowledge creation, AI in business operations, agent-ready software, public AI explanation, evaluation methods, or AI-native product design. Not every research output needs to be an academic paper.

AGI / Research

Thesis and student projects

Some academic collaborations may fit as student projects, internships, thesis topics, supervised research, or applied work. Students should use Internships if the goal is primarily an internship. Use this page when the core proposal is a research collaboration with a lab, professor, supervisor, or research group.

Internships

Evaluation and review

Researchers may help evaluate ideas, products, outputs, research notes, Open Source tools, agent-facing workflows, or public explanations. This can include critique of assumptions, evaluation methodology, benchmark design, failure-mode analysis, literature grounding, public-claim review, research-direction review, and technical review of Open Source tools.

Agentic Decision Systems

Workshops, seminars, and talks

Some collaboration may take the form of workshops, seminar discussions, guest sessions, recorded conversations, or educational formats. These should be scoped. The goal should be to improve understanding or research quality, not to create vague visibility.

Academia / Research

Open Source collaboration

If the collaboration is mainly code, repository work, documentation, examples, or tooling, start with Contributions or Open Source. If the code work is part of a broader research question, this page is also relevant.

Contributions

Grants and funded research

If the collaboration requires external funding, public funding, research grants, or institutional programs, use Grants as the funding route. This page can define the research fit. The Grants page should define the funding fit.

Grants

What inAi can offer research partners

inAi is not a large academic institution and should not be treated as one.

What inAi can offer is different.

A systems-level AI thesis

inAi has a clear public stance: intelligence may emerge from systems, not one isolated model alone. That stance creates research questions around agents, tools, memory, interfaces, feedback, coordination, evaluation, product design, and real-world workflows.

AGI as a System

Product-facing research context

Research connected to inAi does not stay only abstract. The company builds across Business Products, Consumer Products, Products for Agents, and Open Source. That creates practical questions about how AI systems behave in real product contexts.

Products

Public explanation layer

Through AI for Everybody, research can be translated into clearer public understanding. That matters because AI should not be explained only by experts speaking to experts.

AI for Everybody

Open Source connection

Some research may connect to public repositories, examples, tools, developer workflows, documentation, or reference projects. That does not mean all work becomes public code. It means Open Source can be a useful route when public tooling helps the ecosystem.

Open Source

Fast conceptual iteration

inAi is small enough to move quickly across product, research, writing, Open Source, and public architecture. That can make collaboration useful when the scope is specific and the partner understands that inAi is a product company with a research layer, not a university department.

Work with us

What inAi expects from research partners

Good research collaboration starts with clarity.

Specificity

A useful proposal should state the research question, not only the domain. Better: We are studying how tool-using agents maintain state across multi-step workflows and would like to compare this with your Products for Agents thesis. Too vague: We want to collaborate on AI.

Academia / Research

Seriousness

The work should be intellectually serious, even if it is early. Seriousness does not require a famous institution. It does require a real question, relevant background, and respect for claim boundaries.

Academia / Research

Respect for public and private boundaries

Research collaboration must not require inAi to reveal private architecture, internal prompts, private datasets, model or infrastructure details, security findings, customer data, supplier data, candidate data, credentials, or unreleased internal systems. Public pages can explain inAi's stance and public architecture. They are not an invitation to inspect private infrastructure.

How We Build

Useful outputs

A good collaboration should have a plausible output: a research note, public essay, literature review, workshop, evaluation method, prototype, Open Source contribution, thesis project, public explainer, critique of a research direction, partner-funded research plan, or scoped pilot-research link.

Academia / Research

Honest maturity framing

Research collaboration should not be used to imply product maturity that does not exist. Do not frame a research discussion as a partnership launch, customer proof, institutional endorsement, public pilot, or validation unless that is explicitly true and permission-safe.

How We Build

What not to propose

  1. Generic AI collaboration messages with no specific topic.
  2. Requests for confidential internal architecture.
  3. Requests for private prompts, model routing, provider configuration, credentials, or infrastructure details.
  4. Requests for customer, supplier, candidate, or private product data.
  5. Proposals that require inAi to claim it has AGI.
  6. Proposals that present speculative AGI claims as current product capability.
  7. Requests to turn inAi into a research-only lab.
  8. Requests to use inAi's name for false endorsement.
  9. Mass outreach from conferences, journals, or paid publication venues.
  10. Pay-to-publish offers.
  11. Irrelevant surveys with no clear research purpose.
  12. Requests for unpaid work that primarily benefits another institution without clear mutual value.
  13. Grant proposals that use inAi's name without prior agreement.
  14. Papers or public posts that imply partnership before a partnership exists.
  15. Requests to publish private or sensitive material.
Work with usInternshipsGrantsTestersPilots / Corporate partners

Publication, authorship, and attribution

Research collaboration can lead to public material, but publication must be scoped.

Possible public outputs may include research notes, essays, explainers, technical posts, evaluation frameworks, Open Source documentation, public talks, educational material, co-authored articles, and references from inAi pages to external work.

Authorship and attribution should follow real contribution. Institutional names should not be used as endorsement unless there is explicit permission. Research discussions should not be presented as partnerships unless both sides agree.

If publication involves legal, institutional, privacy, IP, confidentiality, or data questions, those must be handled before publication.

Data and confidentiality boundaries

Do not send confidential datasets, private research data, unpublished institutional data, customer data, student data, candidate data, supplier data, credentials, internal documents, or legally sensitive material through a first contact message.

A first contact message should describe the research question and collaboration idea without exposing sensitive information.

If a serious collaboration later requires private material, it must be handled through an appropriate process, with the right permissions, agreements, and legal/privacy review.

LegalPrivacyHow We Build

How to propose a research collaboration

A useful first message is specific and easy to evaluate.

Please include your name; institution, lab, research group, or independent researcher profile; role or title; research area; relevant publications, projects, repository links, or previous work; the research question or collaboration idea; why inAi is relevant to the question; which inAi layer the proposal connects to; what kind of collaboration you are proposing; whether the desired output is private discussion, public note, article, paper, evaluation method, Open Source contribution, student project, workshop, or grant proposal; expected timeline, funding or grant context, confidentiality constraints, and the best way to continue the conversation.

A concise message with a clear research question is better than a broad institutional introduction.

Example proposal directions

Agent memory and state

A researcher studying how agents store, retrieve, and use state across long workflows may want to compare that work with inAi's Products for Agents thesis.

Agentic Decision Systems

Evaluating multi-step AI work

A lab studying evaluation for tool-using agents may propose a framework for assessing multi-step reliability, recovery, and review.

Limits of Intelligence

AI for knowledge creation

A research group studying AI-assisted writing, literature review, source synthesis, or public explanation may connect to inAi's knowledge-creation research and AI for Everybody layer.

AI for Knowledge Creation

AI in business operations

Researchers studying AI adoption in real organizational workflows may connect to inAi's work around product data, catalog systems, supplier information, evidence, and operational support.

AI and Business Operations

Open Source research tooling

Researchers interested in public tools, reproducible workflows, or developer-facing AI utilities may connect through inAi's Open Source layer.

Open Source

Contact route

For academia and research collaboration, use research@inai.world.

Subject line suggestion: Research collaboration — [Your topic / institution]

You can also start from Contact if the website contact menu is available.

For broader collaboration routes, return to Partners.

Email research@inai.worldContactPartners

Related pages

PartnersAGIAGI as a SystemLimits of IntelligenceAgentic Decision SystemsAI for Knowledge CreationAI and Business OperationsProducts for AgentsSoftware for the agent eraOpen SourceOpen Source ApproachAI for EverybodyBusiness ProductsPageMindHow We BuildGrantsInternshipsContributionsContactLegal