The market for AI agent orchestration platforms is growing crowded, but London-based startup Mindstone has introduced a product that stands out for its architectural philosophy rather than just its feature list. The company this week officially launched Rebel, a local-first, agentic AI operating system that uses a novel approach to one of the field’s hardest problems: giving AI agents a reliable, shared memory.
Rebel is distributed under a Fair Source license, meaning teams of up to 100 concurrent users can adopt and customize it for free, while larger organizations require a commercial license. It is available now for macOS on both Intel and Apple Silicon, as well as Windows, with Linux support in development. Mindstone has raised $5 million from investors including Pearson Ventures, Moonfire Ventures, and Zanichelli Venture.
Why Local Markdown Files Are the Core of Rebel’s Architecture
What makes Rebel distinctive is its local-first architecture, which is a deliberate departure from the approach taken by developer-heavy agent frameworks such as LangGraph, CrewAI, and AutoGPT. Those platforms typically require teams to wire together databases, cloud infrastructure, and state-management logic. Rebel, by contrast, stores its core agent memory and instructions across local markdown (.mdcodecodecodecode) text files.
This choice is partly about cost. Common office formats like Word documents and PDFs carry formatting and metadata overhead that consumes model token context and raises API costs. Markdown keeps information closer to raw text, allowing more of a model’s context window to be spent on the actual task. It is also a hedge against vendor lock-in. If a company’s agent instructions, automations, and memory are stored locally as text files, they are not trapped inside one SaaS provider’s interface or database.
A primary configuration file, agents.mdcodecodecodecode, acts as the agent’s core instruction layer and runtime boundary. Users can create repeatable workflows called “Skills,” adjust agent behavior for specific tasks using “Operators,” and run scheduled background tasks via “Automations.”
Automatic Model Routing for Cost, Privacy, and Task Fit
Rebel’s multi-model orchestration capability is another important feature. The system can break a task into parts and route different steps to different models, dynamically switching between local and cloud-based ones. A more powerful model can handle planning or complex reasoning, a cheaper model can handle routine work, and a local model can handle sensitive steps or approval checks that must not leave the device.
This model-agnostic setup gives companies granular control over cost and security. Data-heavy work can run on lower-cost models such as Llama or DeepSeek. Higher-level reasoning can be reserved for more expensive models. Sensitive work can be routed through a local model running on the user’s machine, keeping that information from leaving the device.
Mindstone CTO Greg Detre explained the goal in a recent interview: “I want to be able to say, ‘Help me with this,’ and it knows what’s personal, what’s sensitive, and what can be shared with the whole company.”
Shared Memory That Doesn’t Dump Everything Into One Database
Rebel’s memory system is designed to avoid a common problem in enterprise AI: dumping large amounts of company information into a database and hoping search will retrieve the right context later. Instead, Rebel uses a tiered memory structure. When an interaction happens, the system estimates how likely that information is to be useful again.
Information with a high expected value is written into a local readme.md file tied to a specific project space. Information with a moderate expected value becomes a reference link back to deeper historical records. Lower-priority material is stored in an indexed memory directory, where it remains available but dormant until a relevant task calls it back.
“Shared memory is the most empowering thing you could possibly do with a knowledge-worker AI,” Detre said. “You get this feeling of being a super-organism as a company that just gets smarter and smarter.”
Practical Security and Permissions in a Local-First Agent
Rebel’s launch on Product Hunt this week prompted technical questions about how a local-first agent should handle permissions, safety checks, and shared memory. Detre responded by explaining Rebel’s separation between planning, execution, and background safety logic, adding that companies can configure Rebel to rely entirely on a local model for gating decisions.
That distinction matters for corporate security teams. Autonomous agents often need broad permissions to read files, draft emails, or interact with internal systems. If the final approval layer depends on an external cloud model, some companies may see that as a compliance risk. Rebel can keep those approval boundaries local.
On shared memory, Rebel uses the user’s local “Chief-of-staff README” and defined spaces to separate private, team, and company-wide information. When the agent encounters ambiguous context, the system pauses and asks the user for approval before proceeding.
As CEO Joshua Wöhle put it in a LinkedIn post: “If an agent is going to sit inside your workspace, remember your context, and ask permission before changing the world, you should be able to see how it works. Not because everyone will read the code, but because someone can.”
Enterprise Case Study: Epignosis and the ‘Potatoes Effect’
Mindstone says Rebel has already been deployed across the 250-person workforce of customer Epignosis, covering sales, engineering, product, finance, and customer success teams. Over a 12-week deployment, the company says Epignosis recaptured the equivalent capacity of eight full-time roles. Adoption spread organically after employees saw colleagues automate time-consuming work, a pattern employees reportedly called the “potatoes effect.”
“The border between learning and doing is fading out — and that changes everything about how you scale,” Epignosis CEO Dimitris Tsingos said in a statement.
What This Means for Enterprise AI Buyers
Rebel arrives as companies are trying to move from AI experimentation to AI operations. The first wave of enterprise adoption centered on access: giving employees chatbots, copilots, and model subscriptions. Mindstone is betting the next wave will center on coordination. That means shared memory, reusable workflows, local control, flexible model routing, and measurable business impact.
For larger organizations, Mindstone Pro adds an Impact Dashboard designed to show where Rebel is saving time and money across business units. The dashboard uses a separate, closed LLM to evaluate telemetry and calculate business impact, calibrated conservatively to avoid inflated productivity claims. It is isolated from individual workspaces, allowing IT and business leaders to evaluate adoption and ROI without reading employees’ private agent activity.
Rebel is a clear editorial statement from Mindstone: buying AI seats is not the same as building AI infrastructure. The system is designed to turn scattered employee experiments into an operating layer for work, using the simplest possible file format as its foundation. For teams that value inspectability, data sovereignty, and the ability to move their workflows elsewhere, Rebel offers a compelling alternative to the growing number of cloud-only agent platforms.
