insora maps organizational knowledge, detects gaps through adversarial analysis, and resolves them - grounding every decision in complete, verified information. It builds knowledge graphs from your data, identifies gaps by simulating optimal knowledge, and routes questions to the right experts.

How does insora work?

insora builds three knowledge graphs: a factual graph mapping knowledge and relationships, a gap graph that uses adversarial agents to surface missing information, and an organizational graph mapping expertise. Agents work across documents, external systems, created artifacts, web sources, and your team's experts to answer questions and close knowledge gaps.

What makes insora different from other AI tools?

insora uses adversarial structures to boost output quality and pursues a stochastic approach by simulating optimal knowledge for an organization and comparing it to the current knowledge system. This surfaces knowledge gaps before they cause bad decisions. Zero hallucinations through grounded, source-traced retrieval.

insora was founded by Benno Rothe (CEO, KIT & UC Berkeley, Industrial Engineering & Operations Research) and Adam Kuča (CTO, St Andrews & U Michigan, Computer Science, former ML engineer at Blindspot AI).

Better decisions, faster.

Make decisions with your entire organization's verified knowledge -
built for organizations that cannot afford uncertainty.

Book a demo Test insora

EU-sovereign Zero hallucinations No US LLMs GDPR by design

Based on decision theory and research

Your organization wastes money on inefficient work.

Information search

2.5 hours / day

Lost hunting across tools and inboxes.
~19% of the week goes to finding internal information.

~€12,500 / person / year

Duplicate work

~30% of work

Rebuilt decks, re-answered questions, parallel workstreams - nobody could find what already existed.

Scales with team size

Suboptimal decisions

68% of data unused

Every choice under incomplete information is a bet.
Gains in decision certainty unlock millions in value.

Risk & opportunity cost

McKinsey Global Institute · IDC for Seagate

KNOWLEDGE FOUNDATION

From scattered data to decision-ready knowledge

Step 1 - Connect

Connect everything your organization already uses - Microsoft 365, SAP, Slack, Salesforce, and 40+ integrations. Setup takes less than an hour.

Step 2 - Extract

insora detects entities, relationships, and facts across every connected source. Each extraction is quality-scored. Sensitive data is pseudonymized before processing - GDPR-compliant by architecture.

Step 3 - Organize

Knowledge structures itself by domain - legal, engineering, finance, operations, market intelligence - reflecting your organization’s actual knowledge density and relationships.

Step 4 - Simulate optimal knowledge

Using Bayesian Value of Information analysis, insora models what your organization’s optimal knowledge looks like - tailored to your industry, size, decision patterns, and domain structure. The blue overlay shows where you should be.

Step 5 - Detect blind spots

Every gap between your actual and optimal knowledge is surfaced and ranked by decision impact - missing facts, stale documents, unexplored domains. You see exactly which blind spots cost you the most.

Step 6 - Resolve and verify

Specialized agents close gaps automatically: retrieving missing information, routing questions to the right expert, researching open topics, and connecting isolated knowledge. Every fact is adversarially verified and source-traced - a self-improving knowledge base where every future decision draws on complete, validated information.

Gap detected Suggested action

Upload a document

To close this gap, upload a document that contains the missing information.

Research Just now

Want me to draft a brief?

I can research this topic and create a short summary you can review and add to the knowledge base.

Expert High impact

You may be the best source

There is a gap in your organization’s knowledge that you could close. Can you answer these three questions for me?

Connect Graph update

Link isolated facts

Agents propose connections between previously separate entities so nothing stays stranded.

AGENTIC SYSTEM

From query to verified answer

Step 1 - Query

Step 2 - Plan

Step 3 - Execute

Step 4 - Resolve

User query

Orchestrator

Planning & coordination

Evaluator

Assess completeness

Output

Sub-task internals

Specialized agent

Operative

Evaluator

Results sufficient?

Report to orchestrator

✓ sub-task complete

ENTERPRISE-GRADE

Built for your ecosystem

40+ integrations. Setup in less than an hour.

Enterprise readiness

Keycloak SSO · RBAC · On-premises · ISO 27001 prep · SLA-ready

Role and access import. On-premises deployment option. Scales with your organization - maps structure, expertise, and knowledge patterns as you grow.

Security

EU-only infrastructure. No US language models. Data pseudonymized before processing.

Transparency

Full audit trails. Source tracing for every answer. No black-box decisions.

GDPR by architecture

Data minimization. Right to deletion. Privacy is structural, not a policy layer.

Zero hallucinations

Every output is anchored to verified knowledge in your organization.

VALUE CALCULATOR

Calculate your savings

Knowledge workers spend 19.8% of their time searching for information¹. Another 7.8% goes to duplicating work that already exists².

Across the average team, this adds up to 3.5+ hours per person per day lost to search, creation, processing, and sharing of information that should be instantly accessible.

insora reduces this by retrieving only verified, relevant facts from the knowledge base. No noise. No redundancy. No hallucinations.

On top of time savings, this frees capacity for higher-value and more enjoyable work; as decision frequency and capital intensity increase, insora can add several million in value for larger companies through higher certainty in decisions.

¹ McKinsey Global Institute, The Social Economy. ² IDC Knowledge Worker Productivity Study.

Team size

Avg. hourly cost per person (€)

Hours lost/day per person

Annual cost of lost productivity

€443,300

Expected savings with insora: €354,640

EARLY ADOPTERS

Where we are

The application period is completed. We are now working with our early adopter companies. If we have capacity, we will onboard more early adopters.

Book a demo Test insora

Nov – Dec 2025

Application phase Open call for early adopter organizations. 50+ applications received.

December 31, 2025

10 early adopters selected Cohort finalized. Custom onboarding and direct founder access for each organization.

Jan – Mar 2026

Iterative improvement with early adopters Weekly feedback cycles. Knowledge graph optimization. Agent capability expansion based on real organizational data.

April 2026

Security audit & enterprise hardening Penetration testing. ISO-27001 preparation. Keycloak SSO integration finalized. Enterprise deployment playbook.

Apr – May 2026

Enterprise rollout Onboarding of enterprise clients with dedicated integration support, SLAs, and on-premises deployment option.

ABOUT US

The people behind insora

Background

We met at an accelerator in London and are building insora because we believe AI should be accessible, intuitive, safe, and highly capable for organizations of all kinds - helping their teams make better decisions faster and reduce day-to-day pressure.

How we work

We pursue a highly data-driven approach, but care just as much about how insora feels - we believe users should feel LIGHT when using insora.

Benno Rothe

CEO

Hey, great to see you here - I'm Benno! I graduated from Karlsruhe Institute of Technology (KIT) and UC Berkeley with a degree in Industrial Engineering & Operations Research with distinction. At KIT, I founded and led SkyLab, a student club focused on autonomous drone technology. In contrast to my technical background, I've always been passionate about art and design... and if I'm not creating things, you'll find me playing basketball, snowboarding in the Alps (since I was 5!) or taking my old Yamaha XT 500 on an adventure which I restored with my dad.

Adam Kuča

CTO

Hey, I'm Adam! I studied Computer Science at the University of St Andrews and U Michigan. Before co-founding insora, I built production ML systems at Blindspot AI, and a complete system to enhance satellite imagery. Currently, I'm focusing on Applied Machine Learning and building highly capable & useful agentic networks; at university I led the largest student tech & computing society. When I'm not training models, I'm usually in the gym or learning something new before most people wake up.