Power users already build their own systems.
- MEMORY files, system prompts and local RAGs;
- Notion databases, project summaries and context folders;
- specialized agents, conventions and validation procedures.
Investor Deck
Arka Labs builds the work super-agent.
Cortex learns from projects, maintains its memory and mobilizes agent teams.
Professionals already use multiple AIs, multiple tools and multiple workspaces. But every conversation remains fragile: context is lost, decisions scatter, agents start from scratch and memory must be rebuilt by hand.
The problem
AI accelerates production, but fragments work memory.
A team might use ChatGPT, Claude, Claude Code, Codex, Notion, Slack, Linear, Drive, Gmail or Jira in the same week. Each tool holds part of the context. Each agent understands part of the project. Each new session requires reloading a summary, a MEMORY file, a RAG, documentation or a decision log.
The problem n'est plus seulement de produire avec l'IA. The problem est de maintenir la continuité du travail entre les outils, les modèles, les projets et les personnes.
Memory maintenance becomes a job in itself.
- feed, clean and verify sources;
- resolve contradictions;
- synchronize tools;
- verify the agent actually consults the context.
Our thesis
Agents will only become truly useful with governed, visible, provider-independent memory.
The future of AI-assisted work won't be a single model that does everything. It will be an environment that captures work where it happens, maintains structured memory, connects sources, switches models per mission, delegates to specialized agents and makes every action verifiable.
“Which model should I use?”
“How do I get multiple models and agents to work on my real projects without losing the thread?”
The solution
Cortex gives Cortex a persistent work environment.
Cortex is the super-agent. Cortex is the multi-provider desktop app that provides its memory, sources, tools, permissions, connectors, specialized agents, governance and local or cloud workspace.
Cortex
- Governed memory
- Preserved sources
- Connectors
- Permissions
- Specialized agents
- Multi-agent sprints
- Local-first
- Cortex Cloud for teams
Work is captured where it happens.
ChatGPT.com, Claude.ai, Claude Code, Codex, Antigravity et les sources connectées peuvent être suivis lorsque l'utilisateur l'autorise.
Conversations become actionable memory.
Decisions, constraints, hypotheses, risks, open questions, contradictions, sources, actions and validations are organized.
Memory is maintained over time.
Duplicates are reduced, outdated information flagged and contradictions linked to their sources.
Learnings flow across projects.
A weak signal discovered in one project can inform another topic weeks later.
Complex missions become sprints.
Cortex mobilizes specialized agents while maintaining a single point of contact for the user.
Work remains visible and verifiable.
The user sees sources, validations, refusals and consolidated results.
Positioning
A continuity layer for AI-augmented work.
Cortex is neither a simple model wrapper, an enterprise chatbot, a classic document RAG, nor an isolated agent orchestrator. It connects memory, sources, agents, tools, providers, governance and actions.
Free, Pro, Team.
Free validates value on an initial scope. Pro monetizes advanced individual use. Team monetizes shared memory, roles, audit and Cortex Cloud.
AI-native professionals.
Product Managers, consultants, agencies, advanced freelancers, product teams, AI builders, researchers and innovation managers.
Agencies and multi-client teams.
Their pain is acute: multiple clients, multiple projects, multiple tools, multiple agents and a lot of context to preserve.
Differentiation
Memory becomes a governed asset.
Governed memory
Every important piece of information stays linked to its sources and can be inspected, validated or corrected.
Multi-provider by design
Memory stays in the product, even when the user switches providers.
Local-first
Le cœur du produit fonctionne comme une application desktop. Le cloud intervient pour la synchronisation et l'équipe.
Insights interprojets
Cortex connects learnings across topics and projects to surface new insights.
Governance des agents
Cortex mobilizes specialized agents without hiding sources, validations, refusals or progress.
Atomic Intelligence Governance
Each AI capability is designed as an isolated, testable, injectable and governable atom.
Mission Guardian
Missions are bounded by modes, permissions, tool gates, controlled transitions and execution traces.
Cortex comme actif client
The more a user works with Cortex, the more they build a proprietary asset: memory, topics, decisions, sources, methods, skills and cross-project relationships.
Go-to-market
Progressing from design partners to teams and ecosystem.
Phase 1
Design partners
Private beta with demanding users already working with multiple AIs. Goal: validate value on real, long, contextualized projects.
Phase 2
Advanced solo and consultants
Free and Pro activation. Goal: establish Cortex as a personal continuity tool for AI-native professionals.
Phase 3
Teams and agencies
Team plans with shared memory, roles, permissions, audit and dedicated Cortex Cloud.
Phase 4
Ecosystem
Skills.Factory integration, MCP connectors, skill libraries and method marketplace.
Current traction
Product, pricing and beta are in consolidation.
Arka Labs has a Cortex desktop app under construction, a memory/topics/sources/providers/governance foundation, a defined Free/Pro/Team pricing, a delivered pricing page, a prepared “super-agent” campaign and a private beta opening gradually.
ProductDesktopMemory, providers, topics, sources and governance foundation in consolidation.
Design partnersPM · AI · researchEarly conversations around multi-project memory, weak signals and cross-project insights.
To be filledBeta KPIsSignups, interviews, active users, retention, topics, sources, captures, insights and willingness to pay.
Investor need
Accelerate Cortex toward a robust, measurable and marketable beta.
Arka Labs is preparing an acceleration phase to stabilize memory and connectors, finalize Cortex Cloud, strengthen security and governance, structure beta onboarding, measure retention and prepare the Team offering for agencies and teams.
- product and engineering;
- cloud infrastructure;
- security and compliance;
- design partners and beta support;
- targeted go-to-market;
- content and product demos.
Amount and vehicle
Amount sought to be determined. Envisaged format: pre-seed, angel round, BSA Air or another suitable vehicle.
What we want to prove now
Before scaling, Arka Labs wants to validate three points.
Retention
Do users return weekly because Cortex genuinely saves them from rebuilding context?
Value of memory
Is structured, sourced, consolidated memory perceived as superior to a MEMORY file, a RAG or manual documentation?
Transition to teams
Do agencies and teams see Cortex as a collective asset, not just an individual tool?
Investor summary
Arka Labs builds the continuity layer for AI-augmented teams.
Cortex addresses an emerging pain: professionals increasingly use AI but lose work continuity across conversations, models, tools and projects. The product turns this fragmentation into governed, multi-provider, actionable memory.