What is Spiderbrain?
Your AI’s second brain. Your company’s brain. Spiderbrain, sometimes searched as Spider Brain, is a deterministic context and memory layer that understands how your work is connected, what decisions have been made, and what could break when something changes.
Your AI can generate an answer. Spiderbrain gives it memory, structure, context and a map of how everything connects. It turns your code, documents, decisions, dependencies, rules and institutional knowledge into a living brain that stays connected to the underlying work, so an agent is reasoning over the real shape of your project instead of re-reading text every time.
AI is getting smarter, and it still forgets
Every AI session starts with the same problem: context disappears. Your assistant may be extremely capable, but it does not naturally know:
- why a system was designed the way it was
- which files depend on the one you are changing
- which decisions are still valid, and which have gone stale
- which parts of the system are load-bearing
- what your team learned three months ago
- what another engineer decided last week
- what will break if you make a small-looking change
That knowledge exists. It is just scattered across code, documents, tickets, commits, conversations, people and systems. Humans gradually lose the map. AI agents never had the map in the first place. Spiderbrain builds it and keeps it current.
From second brain to company brain
A personal second brain remembers what matters to you. A company brain remembers what matters to the organization. Spiderbrain is the same engine at three scales, connecting the things that normally stay disconnected: code, systems, dependencies, documents, decisions, rules, people and organizational memory.
A second brain for your project
For one developer, Spiderbrain is a second brain that holds the shape of the work: what connects to what, why it was built that way, and what a change will reach. It persists between sessions, so your AI never starts from zero.
A shared brain for your team
For a team, the same brain becomes one shared understanding. Two engineers asking the same question get the same answer, because they are reading the same map rather than each rebuilding it from memory.
A company brain for your organization
For an organization, it becomes a persistent layer of institutional memory that people and AI agents can query, reason over and build on. Code, systems, documents, decisions, rules and the relationships between them, in one place.
The top rung is a product. Spiderbrain for business is the platform that brings your business data in, resolves identity across it, and runs the Spiderbrain engine as the brain underneath, on an instance provisioned for your organization alone.
Instead of asking an AI to rediscover this context every session, Spiderbrain gives it a persistent, queryable representation of the organization. The result is an AI that does not simply know what something says. It can work out how things are connected, what depends on what, what has changed, and what the consequences are. That is the foundation for persistent enterprise AI.
The problem Spiderbrain solves
A large project or organization is mostly invisible relationships. Which component is load-bearing? What breaks if this function changes? Which systems depend on this policy? Why was this decision made, and has it gone stale? Which piece of institutional knowledge is about to disappear when someone leaves?
Traditional AI retrieves pieces of information. Spiderbrain maintains the relationships between them, as a living context graph that makes those relationships explicit and continuously measurable.
Three sentences explain Spiderbrain
- It tells you what a change reaches. Point it at your code, your documents, your records or your rules. It answers what depends on this, what breaks if it changes, and what is holding the system together.
- It maps every connection and walks every path. Spiderbrain follows the real structure of the system. Every relevant relationship is mapped and every path evaluated by the same deterministic process, rather than by whatever an AI happened to retrieve.
- It produces an answer that is complete, consistent and provable. Complete because the relevant paths are followed. Consistent because the same method is applied every time. Provable because the brain can be rebuilt and independently checked.
This is context engineering for AI systems: turning the structure of your work into something an AI can reliably reason over.
How Spiderbrain works
- It builds a context graph. Every file, dependency, decision and relevant artifact becomes part of one deterministic graph, parsed natively inside your own environment.
- It scores what matters. The graph is scored on our EU servers so the important and potentially dangerous parts become visible, each node carrying measures such as severity, importance and blast radius. Your source stays on your device; only the derived, source-free graph is sent for scoring.
- It finds the keystones. Some parts of a system matter disproportionately. Spiderbrain surfaces those hubs as masters, the natural firebreaks in a cascade and the natural points for review and protection.
- It gives agents persistent context. The brain is served over MCP to Claude, Cursor, Continue, Zed, Windsurf and any Model Context Protocol client, so your AI does not start from zero every time.
Spiderbrain as a decision engine
A company brain is not just a database of information. It has to understand decisions. Spiderbrain makes them durable, scored, traceable, and capable of going stale when the underlying reality changes. People and AI agents make the decisions; Spiderbrain remembers, scores, ages and audits them, and never silently overwrites history.
The same question gets the same answer
Whoever asks, wherever they are, whenever they ask. Two people working from the same brain are working from the same underlying facts.
Every answer can identify its source
Not a paraphrase. Spiderbrain can name the specific node and version behind a judgment, so the reasoning can be reviewed and defended.
Knowledge expires when reality changes
A decision should not go stale because a timer ran out. Spiderbrain follows structural change: rewire what something depends on and the relevant judgment reopens. Knowledge ages because reality changed, not because the calendar did.
People stay accountable
The engine proposes. A named person approves. That approval is what turns a candidate into an organizational rule, and the approval itself becomes part of the record.
What Spiderbrain does
Stochastic AI in, deterministic knowledge out
AI is probabilistic, which is useful for generating possibilities and risky when a guess silently becomes organizational knowledge. Spiderbrain holds AI proposals as candidates, admits measured structure to the record, and requires a person to turn a candidate into an approved fact.
Contradictions are surfaced, not silently resolved
When two recorded decisions conflict, Spiderbrain does not quietly pick a winner. Both are shown and a person chooses, so the conflict becomes a visible decision with an owner and a date.
Only derived understanding travels
Parsing is native and happens on-device. Your source code, file contents and commit messages do not leave your machine. The scoring layer works from a source-free representation of paths, edges, hashes and privacy-safe signals.
Determinism is tested, not asserted
Repeated runs are tested for byte-identical output, and every validation figure is published against the engine version that produced it. The goal is simple: same input, same brain.
Your AI’s second brain
Spiderbrain includes an agentic chat grounded in the brain of your project. Instead of asking a generic coding assistant to rediscover your architecture, the agent gets the context it needs to understand the system before it acts.
It knows your codebase
The chat retrieves the files that matter, ranked by importance (webscore), so the AI answers with the right context instead of filling a window with more text.
It knows what your change could break
Before any edit, it shows the blast radius, the files and components affected by what you are about to change, and waits for your approval.
Your models, your keys
Multi-provider: Anthropic, OpenAI, Google, or local models through Ollama and LM Studio. Bring your own key, or keep chat inference private with local models.
Ask, or Agent
Read-only questions when you want understanding. Agent mode when you want scoped changes proposed and applied, each behind a diff with one-step undo.
Blueprint: give your AI a picture
Every Spiderbrain project carries a Blueprint, an architectural canvas where you represent systems as blocks and connections instead of paragraphs of description. Draw the architecture you want to build, or let Spiderbrain derive one from the project it has already mapped.
- From drawing to instruction. Share the Blueprint to your AI over MCP (a Pro feature), and the agent reads the same boxes and wires you drew, so "build this" finally has a picture attached.
- From code to drawing. A built brain can auto-draw its own architecture, so you start from the architecture that actually exists rather than a blank canvas.
- You stay the editor. Updates flow when you pull them. Nothing redraws or overwrites your plan behind your back.
Memory tree: the memory behind the brain
A second brain needs memory. The Memory tree holds the decisions, constraints, patterns and preferences that shape a project, anchored to the parts of the graph they belong to and carried across sessions. The reasoning that shaped the work stops evaporating when a chat window closes.
- Readable, not a black box. Open the Memory tree on any project and see exactly what is remembered, grouped by type and status.
- Steerable. Comment on any memory to ask your AI to change it. Requests queue until the agent applies them, and nothing changes silently.
- Persistent and private. Hosted memory is encrypted on our EU infrastructure (a Pro feature) and served to agents over MCP, so every session can begin with the project’s existing memory rather than an empty context window.
Spiderbrain sees where your system is going
A brain should not only understand the present. It should understand direction. Every Spiderbrain brain is a sealed, fingerprinted release, and comparing releases makes it possible to see how the structure of a project is changing over time. No additional instrumentation is required: the signal emerges from comparing the same map across versions.
What becomes load-bearing next
A component whose reach keeps growing may be turning into a future bottleneck even though it looks ordinary today.
Where the next breaking change starts
Repeated change combined with wide reach identifies the places where future changes are more likely to cascade.
What a piece of work will actually touch
The measurable reach of a proposed change gives a better indication of effort and risk than intuition alone, and it can be checked before the work starts.
What Spiderbrain gives your team
Agents that remember
Persistent project memory across every session, instead of context that disappears with each conversation.
Know what breaks
Severity and blast-radius scores reveal the real impact of a change before you ship it.
Find what matters
Importance scoring surfaces the handful of components carrying disproportionate weight in your system.
Keep the brain current
Rebuild in one click, with drift detection flagging what changed. No manual re-indexing, no stale embeddings. Pioneer founding members get live watch: the map updates as files change.
Keep your source private
Parsing is native and on-device; only a source-free graph is scored. Your source code stays where it belongs.
Connect to the tools you already use
MCP-native: Claude, Claude Code, Cursor, Continue, Zed, Windsurf and more, with no glue code.
Your company already has an ontology
Most enterprises spend months designing ontologies for their data. Your organization already contains one, hidden inside the work itself. Your code contains dependencies. Your documents contain relationships. Your policies contain rules. Your decisions contain constraints. Your systems contain architecture. Your history contains patterns.
Spiderbrain derives that structure instead of asking the organization to describe itself first. Typed nodes, dependencies, containment, constraints and decisions, extracted deterministically from the underlying material. No ontology workshop, no OWL, no modeling committee, no months spent documenting the business before AI can understand it. The company brain starts with the company itself. More on code ontologies in the FAQ, or see how a context layer differs from a semantic layer.
Capabilities
Structure, not guesswork. Everything Spiderbrain learns about your project, served to your agents as queryable, deterministic intelligence, over the graph API and MCP.
Context engineering, as a graph
Every file, dependency and decision becomes a node and an edge in one deterministic graph. Spiderbrain parses real relationships rather than relying on text similarity, scores every node, and serves the result as a living map your agent can query over MCP.
How Spiderbrain prevents context rot
Language models get less reliable as irrelevant context accumulates. More context does not mean better context. Spiderbrain identifies the important parts of the graph so your agent receives the load-bearing few rather than the noisy thousands.
Blast radius, answered
“What breaks if I change this?” gets a real answer. spikescore traces conditional co-failure across the graph, so the measurable reach of a change is visible before you ship instead of estimated afterwards.
Severity that ranks itself
Not every component matters equally. webscore rates every node by how much its failure would hurt, so your AI understands not only what exists, but what matters.
Hosted MCP endpoint
One hosted MCP server at mcp.spiderbrain.ai. Connects Claude, Cursor, Continue, Cody, Zed and any MCP client. No glue code, no local server.
Project decision log
Every decision is append-only, signed and hash-chained, anchored to the file or area it is about. A record you can audit: judgments, reasons, and when they went stale.
Private and deterministic
Your source code never leaves your machine; only a source-free graph is scored. Same input, same brain, every time.
Reads everything your work is made of
Parses real imports across languages, plus documents and chat history, so the graph reflects the whole of your work, not just the code.
A brain your systems can query
The same structural brain your agents read is available over one authenticated REST API and hosted MCP endpoint. Wire it into CI pipelines, engineering dashboards, internal applications, governance and risk workflows, or your own agents. The brain becomes an infrastructure layer rather than another isolated app.
A company brain built from reality
The most useful organizational knowledge is the hardest to maintain. People leave. Systems change. Policies evolve. Architecture drifts. Documents go out of date. Decisions get forgotten. Traditional knowledge management asks people to keep the knowledge base accurate, which is a job nobody has time for and everybody is blamed for.
Spiderbrain takes the other approach: derive the brain from the organization itself, keep it connected to the underlying reality, and let structural change tell you when the knowledge needs to change. That is what separates a company brain from a document repository.
What the scoring stands on
None of the mathematics underneath Spiderbrain is invented. It composes established work from four fields, and the composition is what is ours: conformance checking, argumentation and drift, applied to a decision graph sitting on a dependency graph, under a determinism rule.
Network science
Guimera and Amaral, Nature. Their method for telling a local hub from a bridge between communities is how Spiderbrain separates near spread from far spread.
Search ranking
Brin and Page, with the local computation method of Andersen, Chung and Lang. The algorithm that ranked the web, pointed at dependency graphs, at a cost that does not grow with the graph.
Reliability engineering
Fault tree analysis, and United States Nuclear Regulatory Commission guidance on modelling failures with a common cause. Nuclear safety analysis has asked this question for fifty years.
Neuroscience
Hebb on how connections strengthen, and Bi and Poo on the spike timing that decides whether they strengthen or weaken.
The scores that come out of it are measured rather than asserted, on real corpora, with every figure tagged with the engine version that produced it. See the validation report.
Where Spiderbrain can be used
The pattern appears anywhere written rules and dependencies outnumber what any individual can remember. Spiderbrain started with software, and the underlying graph does not care whether its nodes are files, documents, records, policies, schemas or decisions.
Software and cloud
Understand why systems were built the way they were, what a change reaches, and which components are load-bearing. Engineering reasoning survives a team change instead of leaving with the people who created it.
Autonomous vehicles
Safety-critical constants and rules end up distributed across configuration, code, schemas and documentation. Spiderbrain connects them so they can be reviewed, versioned and traced to the number they move.
Defence and public administration
Rule-bound decisions depend on orders, amendments, entitlements, postings and procurement rules accumulated over decades. A persistent organizational brain makes those relationships traceable.
Clinical trials
One protocol amendment can reach consent forms, monitoring plans, standard procedures and investigator documentation. Spiderbrain turns that cascade into a connected, inspectable structure.
Pharmaceutical manufacturing
A single standard update can reach master files, procedures, client agreements and batch records. The company brain identifies the connected surface that needs reviewing.
Medicine and oncology
Panel updates and variant reclassifications propagate into reports, recommendations and partner protocols. Spiderbrain surfaces the connected knowledge that may have gone stale.
Compliance, risk and legal
When a framework version changes, affected policies, controls, evidence and engagements have to be revisited. The connected brain makes that surface queryable instead of an annual archaeology exercise.
Who is Spiderbrain for?
Developers and teams using AI agents who want the AI to understand the system behind the code, not simply autocomplete the next piece of it.
Individuals
Who want a second brain for their projects, so the reasoning behind the work is still there next week.
Engineering teams
Who need one shared understanding of a codebase rather than several private mental models.
Organizations
Who need persistent institutional memory that outlives the people who created it.
Enterprises
Who want a company brain sitting beneath their AI agents, so those agents reason over the business rather than around it.
From AI assistant to organizational intelligence
AI is becoming very good at generating. The next problem is helping it remember, understand and reason over the organization it is working inside, and that takes more than a larger context window. It takes a persistent brain: a second brain for your project, a shared brain for your team, a company brain for your organization, and one context layer underneath every agent working across all of them.
Intelligence is everywhere. Context is what makes it useful, memory is what makes it persistent, and a brain is what connects it all.
Built by Perform Digital
Spiderbrain is built by Perform Digital, which has been delivering for enterprise clients since 2019. The people who build the engine are the people who deploy it, so an unusual data shape, an unfamiliar approval flow, or a deployment that has to stay inside your own boundary is treated as an extension of the engine rather than a workaround. More for teams and enterprises.
Second brain and company brain, answered
What does an AI second brain do?
An AI second brain holds what you and your tools already know in a form an AI can act on, so it does not start from zero every session. Spiderbrain builds one by deriving the structure of your systems rather than asking you to file notes: it maps what depends on what, scores which parts matter most, keeps a record of decisions and their reasons, and serves that to any AI agent over MCP. The practical difference from a notes app is that nobody has to maintain it. The graph is derived from the systems themselves, so it goes out of date only when they do.
Is this the same as the Building a Second Brain method?
No, and the distinction is worth stating plainly. Building a Second Brain, by Tiago Forte, is a personal knowledge-management method: you capture, organise, distil and express notes, and the value comes from your own curation. Spiderbrain is software that derives a second brain automatically from systems you already have, for an AI to consult rather than for you to read. If you want a method for managing your own notes, that book is the better answer. If you want your AI to stop forgetting your project, this is.
Is the second brain the same as the gut?
In biology, yes, and that is a different subject. The enteric nervous system in the gut is commonly called the second brain, and medical sources such as Johns Hopkins Medicine and Harvard Medical School are the right places to read about it. This page uses second brain in its computing sense: an external memory that a person or an AI consults. Both meanings are legitimate; they simply share a name.
What is a company brain?
A company brain is one governed understanding of how an organization actually works, shared by every team and every AI agent instead of living in individual heads and scattered documents. It is the same idea as a personal second brain, scaled up: the same derived structure and the same decision record, but organization-wide and governed, so an answer given to one team is the answer given to every team. Spiderbrain describes the rungs of that ladder as a second brain for a project, a shared brain for a team, and a company brain for an organization.
How do you build an AI second brain?
Point Spiderbrain at a project and it builds one. Parsing runs natively on your device, only a source-free graph is uploaded for scoring, and the result is served to your AI over MCP. There is nothing to file or tag. What ships today is software and code, plus database schemas read from a SQL file. The alternative approach, assembling one by hand from notes and documents, is what most second-brain tooling does, and its weakness is that it stays accurate only for as long as somebody keeps maintaining it.
What are the limitations of a second brain?
The usual one is decay. A hand-curated second brain is accurate on the day it is written and drifts from then on, because nothing connects the note to the thing it describes. A derived second brain moves that problem rather than removing it: it stays current with the systems it reads, so it is only as complete as its coverage. Spiderbrain resolves dependencies structurally in JavaScript, TypeScript, Python and Rust, indexes everything else as content, and reads database schemas from a SQL file. Anything outside that is not in the brain, and the honest answer is to say so rather than imply full coverage.