Specor

The Semantic Product Runtime

Your product has
a memory now.

Specor turns every feature, rule, flow, and decision into one living, versioned graph — checked continuously by AI and read live by your engineers' AI tools. Context never walks out the door.

For product, engineering, QA & design Compliance & UX checks on every commit A live product brain your AI tools read

No credit card · Viewers always free · Set up in minutes

Live demo — pick a feature, switch tabs, tap a rule. No signup.

app.specor.ai/graph/nodes/guest-checkout

Guest Checkout

Feature

Lets shoppers complete a purchase without creating an account.

Edit Delete
Capability:Checkout
Domain:Commerce

Needs (2)

Link Need
Buy without creating an accountFaster path to purchase

Decisions (1)

Record Decision
No mandatory signup at checkout

Plugs into the AI tools your engineers already use

CursorClaudeGitHub CopilotVS Code via open MCP standard

What makes Specor different

A graph, not documents

Features, rules, flows, and decisions are connected and queryable — not prose that drifts out of date.

Versioned like Git

Branch, save, review, and revert your product definition — in plain English, no command line.

AI checks every commit

Coverage, conflict, compliance, security, and UX gaps caught automatically — for free.

Read live by your AI tools

Cursor, Claude, and Copilot pull the real spec over the open MCP standard before writing code.

Sound familiar?

Context walks out the door

When a PM leaves, the decisions, rationale, and history go with them. Nobody knows why it was built that way.

Six tools, zero sync

Product lives across Notion, Jira, Confluence, Miro, and Slack. Change one — the other five go stale immediately.

Developers still have to guess

Engineers read a two-sentence Jira ticket, ask a PM in Slack, and ship code that doesn't match what product intended.

In Specor these aren't scattered notes — they're connected. Change one thing, and Specor knows what else is affected.

How it works

From scattered context to a single source of truth

You don't start from a blank page, and you don't need to learn version control. Three steps — the same three you'll watch unfold down this page.

1

It remembers

Bring what you have — a meeting transcript, a PRD, your codebase, or plain English. Specor turns it into a structured spec: features, rules, flows, and decisions, all connected and versioned.

2

It reviews itself

Background agents review every change, flag gaps and conflicts, and propose fixes — instantly for per-node quality issues, and after each commit for compliance, UX, and security gaps.

3

It ships in sync

Everyone works from one source of truth. Developers connect their AI coding tools via MCP, so the code that gets written matches what product actually intended.

Who it's for

One product. Read differently by every team.

Everyone works from the same versioned source of truth — but each role gets the slice they care about.

Plus Living Documents — PRDs, briefs, and release notes generated straight from your graph, and flagged the moment they go stale.

Act 1 · It remembers

Watch one feature — Guest Checkout — enter Specor. You'll follow it down this whole page.

Multiple ways in

Bring what you have — watch your spec build itself

You don't need to start from a blank spec. Paste a meeting transcript, think out loud in chat, drop in a brief, or connect a GitHub repo — Specor turns any of them into a connected, versioned product graph.

Paste any meeting recording transcript or notes. Specor reads the discussion, proposes what should change in the spec, and stages it for your review and one-click commit.

You paste

"We agreed to drop mandatory signup from checkout. Pix payment should expire after 15 minutes — legal requirement. Guest users should be able to buy without creating an account."

Specor proposes

→ review on a draft branch, then commit

Full Version Control

Git for product — without the engineering degree

Branch off to explore a new direction. Merge when you're ready. Roll back anything if you change your mind. See exactly what changed between any two versions — in plain English, not code.

  • Product Branches — Explore a pivot without touching main. Multiple squads work independently, then merge.
  • Sealed Milestones — End-of-sprint snapshots with author stamps. Your product changelog writes itself.
  • Non-destructive Rollback — Reverse any milestone safely. History is never deleted — rollback creates a new record.
  • Full Immutable History — Every field, relationship, and decision — permanently recorded. Time-travel to any past state.

Semantic diff — plain English

+ Feature "Guest Checkout" introduced

+ Rule "Pix expires after 15 minutes" added

~ Feature "Standard Checkout" — status: active → deprecated

- Flow "Legacy Checkout Flow" removed

AI-written impact summary

Affected capabilities: Commerce › Checkout

Affected metrics: Checkout Conversion Rate (likely impacted)

Affected personas: Guest User (new), Returning Customer (unchanged)

Affected flows: 2 modified, 1 added, 1 removed

Every diff is a shareable link you can paste into Slack or a PR.

Act 2 · It reviews itself

Now Specor reads what you wrote. The Guest Checkout spec you just committed gets reviewed on every change.

Product Intelligence

Compliance, security & UX gaps — caught before your auditor does

Every time you seal a commit, Product Intelligence runs a deep, branch-level sweep. It checks your entire product spec against compliance frameworks and UX best practices — without consuming AI Token credits.

  • Compliance & Security — Gaps mapped to SOC2, OWASP, and GDPR — auth rules, audit trails, rate limits, privilege escalation.
  • UX & Flows — Missing login, signup, or password-reset paths. Destructive actions without a confirmation step.
  • Product Completeness — Features without onboarding flows. Personas with no way to reach the product.

Findings live in a dedicated inbox — separate from Spec Quality — with framework badges, article references, and one-click resolution.

app.specor.ai/intelligence

Product Intelligence

Findings from the last commit sweep

2 errors 1 warning
error SOC2 "Checkout" has user personas but no authentication rule

Features involving users must be governed by an auth rule. Missing auth coverage is flagged as SOC2 CC6.1.

SOC2 / OWASP · CC6.1 / A01 · 94% confidence
AI Fix Snooze Reject
error UX & Flows No login or signup flow defined for user personas

Personas exist but there is no feature describing how they gain access to the product.

UX · 88% confidence
AI Fix Snooze Reject
warning Security (OWASP) "Public API" feature lacks a rate-limiting rule

Public-facing features should be governed by an explicit rate-limit rule — OWASP A01 / SOC2 CC6.6.

OWASP / SOC2 · A01 / CC6.6 · 81% confidence
AI Fix Snooze Reject

Runs automatically on every commit seal. Zero AI credits consumed.

Plus six Spec Quality agents reviewing every node as you type

CoverageConflict DetectionFlow ConsistencyProduct ImpactAlignmentNaming & Consistency

Instant per-node checks — "Guest Checkout has no Metric", "these two rules conflict" — surfaced for your approval, never auto-applied. Everything rolls up into a live 0–100 Product Health Score.

84

Product Health

Available agents vary by plan. Custom agents built to your vocabulary on Enterprise.

Act 3 · It ships in sync

The spec doesn't stop at the spec. The same Guest Checkout definition is what your engineers' AI tools read, what every merge is tested against, and where real metrics flow back in.

MCP Integration

Your engineers' AI tools read your product definition live

Specor ships an MCP server (Model Context Protocol) that connects to Cursor, Claude, VS Code Copilot, and any compatible AI coding tool. Once connected, the developer's AI reads your feature definitions, business rules, acceptance criteria, and user flows — automatically, in the background.

  • Implement a feature

    AI reads your rules, ACs, and flows — writes code that matches what product intended.

  • Write tests

    AI reads your acceptance criteria — generates one test per AC, grounded in your definition of done.

  • Debug a regression

    AI reads recent milestones — surfaces the rule or decision that changed and caused the break.

  • Onboard to a new area

    AI reads the domain overview — full product architecture before writing a single line.

Configure once in your editor

// ~/.cursor/mcp.json

{
  "mcpServers": {
    "specor": {
      "url": "https://mcp.specor.ai/",
      "headers": {
        "Authorization": "Bearer spk_live_...",
        "X-Workspace-ID": "your-workspace-uuid"
      }
    }
  }
}

After that — nothing. The AI assistant calls Specor automatically every time a developer starts implementing a feature. No copy-pasting docs, no interrupting PMs, no guessing.

Read the MCP docs →

Merge Confidence

AI writes the test cases for every merge — before you confirm it

When a merge request is open, Specor reads its real semantic diff and generates Gherkin-style (Given/When/Then) test cases — one set for what actually changed, plus coverage for the "blast radius": nearby graph nodes that often regress without anyone touching them directly. A developer or QA manually verifies the real implementation against each case and marks it pass or fail.

  • Grounded in the real diff — Test cases are generated from the merge's actual structural diff, not a generic checklist — they name the rule, feature, or flow that changed.
  • Blast-radius coverage — Graph neighbors up to two hops from the change get a lighter-weight regression check, with a "why" explaining the dependency.
  • Pass/fail merge gate — Branch protection can require every test case to pass before a merge confirms — no silent merges past unverified changes.
  • Markable from your AI tool — List, review, and mark test cases pass/fail straight from Cursor or Claude via MCP — editors and admins only; viewers stay read-only.
app.specor.ai/merges/guest-checkout

1/2 test cases passed

Required to merge

Guest can complete checkout without an account

Direct change

Covers the new "no mandatory signup" rule on this feature.

Given a guest user with items in their cart

When they reach checkout without signing in

Then they can pay and receive an order confirmation

Passed

Pix payment link still expires after 15 minutes

Regression · 2 hops

Regression check: this Rule depends on the changed Checkout flow.

Given a guest checkout generates a Pix payment link

When 15 minutes pass without payment

Then the link is rejected as expired

Pass Fail

GitHub Integration

Link a GitHub PR to a Specor MR — see where code and spec diverge

When you open a Specor merge request, attach the GitHub PR that implements it. Run a drift analysis and get two answers in one place: code that was written but has no spec node, and spec nodes that should be covered by the PR but aren't.

Code → Spec

Specor reads the diff, identifies logic that has no matching feature, rule, or flow — and proposes the missing spec nodes directly on the branch.

Spec → Code

Specor checks every spec node in scope for this PR. Nodes with no matching code surface as test cases on the MR: "Is this feature actually in the diff?"

Auto re-analysis on new commits

When the PR gets new commits, the drift analysis re-runs automatically. The Specor MR stays current without any manual action.

Inside the Specor MR

Linked PR github.com/acme/app/pull/312

Code → Spec gaps (2)

⚠ No spec node for POST /api/refunds

⚠ No spec node for rate-limit logic in checkout.ts:84

Spec → Code gaps (1 test case)

? Is "Guest Checkout Flow" implemented in this PR?

PMs review the Specor MR. Devs review the GitHub PR. Both link to each other. When both are merged, code and spec are in sync.

Figma Integration

Link the design to the spec — and catch it when they drift

Attach a Figma frame to any node from the Specor plugin, right inside Figma. The link lives in the spec's history — it shows up in diffs, survives rollbacks, and travels across branches on merge. No OAuth to set up: the plugin reads the open file and pushes what Specor needs, so Specor never calls Figma's API on your behalf.

Design on the node

A live thumbnail sits on the node itself — no Figma login needed to look. Open the exact frame in Figma, or flip on an interactive live preview.

Spec ↔ design drift

Specor compares the pushed design against the spec and flags where they disagree — a rule the screen doesn't honor, a state the spec never mentions.

Design coverage

Features with no linked frame are surfaced by the Coverage agent — so a feature that shipped to spec but never got a design doesn't slip through.

On the node's Design panel

Feature Guest Checkout
Checkout Screen · thumbnail
Open in Figma ↗ Live preview
⚠ Design drift — screen doesn't honor the "guest checkout" rule

Product owns the spec. Design owns the file. The link between them is versioned, so spec and design never quietly drift apart.

Live Metrics

Your success metrics, live — not a number someone typed last quarter

Connect your analytics provider and bind a Metric to a real query. Specor pulls the current value and shows it against your target — right on the node and in a dashboard. It's the first time real-world data flows into your spec: a Feature's success metric reading 4.2% vs a 6% target — off track, not a stale guess.

  • Connect your provider — PostHog and Google Analytics 4 today — Amplitude & Mixpanel next, behind one adapter. Keys are encrypted per workspace and set by an admin.
  • Bind any Metric — Point a Metric node at a real insight. The binding is versioned like every other change — diffable and rollback-safe.
  • On / off-track at a glance — Current value vs. target computed live, on the node and in a metrics dashboard, with staleness flagged if a provider hiccups.
  • Mix providers freely — Different metrics can read from different tools. One dashboard, every source — no lock-in.
app.specor.ai/graph/guest-checkout

Metric · Live

Checkout Conversion Rate

PostHog
4.2 % ● Off track

Target 6.0% · higher is better

— target 6.0% updated 2 min ago · refreshes automatically

Live Examples

See what a real product structure looks like

We've modelled six real-world products in Specor — features, rules, flows, metrics, and decisions, all connected. View any of them live in the app — no signup required.

No account required — view any example instantly, right in the app.

These examples are illustrative product structures created by the Specor team for demonstration purposes only. They are not affiliated with, endorsed by, or representative of the actual products or companies depicted. All trademarks belong to their respective owners.

Not another doc tool

Every tool below tracks things. None of them know how those things relate.

Specor Notion / Confluence Jira / Linear ProductBoard / Aha!
Structured semantic graph ✓ Connected & versioned ✗ Unstructured docs ✗ Tickets and tasks ~ Some relationships
Immutable history, time-travel & safe rollback ✓ Every field, forever ~ Last edited by ~ Ticket log ✗ Limited
AI detects spec gaps & conflicts ✓ 6 Spec Quality agents ~ Basic copilot
Compliance & security analysis (SOC2, OWASP, GDPR) ✓ Product Intelligence
Semantic branching & business-readable diffs
AI coding tools read it live (MCP)
Bootstrap spec from codebase + PR drift detection ✓ GitHub integration
Live metric values from analytics ✓ PostHog, GA4 & more ~ Manual KPIs
Sprint & ticket execution tracking → Jira / Linear ~ Basic boards ✓ Core strength ~ Roadmap-level

Simple, transparent pricing

Free for your first editor. $99/month for teams. AI credits are included per plan, and viewer seats are always free.

Starter

Small teams — get started

$0 / 1st editor

+$25/additional editor · up to 5 editors total

  • 1 workspace, up to 5 editor seats
  • 100 AI credits/month
  • Coverage agent
  • Product Intelligence (compliance & UX checks)
  • Full history, no limits
  • MCP server access
  • All core features
Start free
Most popular

Team

Product teams shipping together

$99/org/mo

First editor free · +$25/additional editor · unlimited editors

  • Up to 10 workspaces
  • Unlimited editor seats
  • 1,000 AI credits/month
  • All 6 Spec Quality agents
  • Product Intelligence (compliance, security & UX analysis)
  • Semantic search across your product graph
  • Live metrics integration (PostHog, GA4 & more)
  • Priority support
  • Slack integration
  • GitHub integration
Start free

Enterprise

Multi-product organizations

Custom

Custom contract · volume AI credits or BYOA

  • Unlimited workspaces & seats
  • SSO / SAML (Okta, Azure AD)
  • Audit log & compliance exports
  • Self-hosted option
  • Bring your own AI (BYOA)
  • Custom AI agents for your ontology
  • Data residency (EU available)
  • 99.95% SLA + status page
  • Dedicated Customer Success Manager
Book a walkthrough

Example: a 5-person product team on Team pays $99 + 4 × $25 = $199/month (the first editor seat is free), with 1,000 AI credits and all 6 agents included.

Monthly total = org plan fee + (active editor seats × seat rate). Additional AI credits available at $0.05 / credit (minimum 100 credits per purchase), no expiry. Proactive agents are event-driven and never consume credits.

Why we built Specor

“Every product team I've worked on lost the same thing: the why. The decisions, the rules, the trade-offs — scattered across six tools and the heads of people who eventually leave. We built Specor so a product can finally remember itself.”
Daniel Ciolfi, Founder of Specor

Daniel Ciolfi

Founder, Specor

Questions, answered

The things teams ask before they start.

Does Specor replace Jira, Notion, or Linear?

No — it sits alongside them. Jira and Linear track the work; Specor holds the definition of what the product is and why. Notion and Confluence hold prose that goes stale; Specor holds a structured, connected, versioned graph that stays consistent. Most teams keep their ticket tracker and use Specor as the single source of truth it points back to.

How long does it take to get started?

Minutes. You don't start from a blank graph — paste a meeting transcript, a PRD, or a plain-English description and Specor generates a structured spec you can review and commit. You can model your first feature in a single sitting.

Do I need to understand Git or version control?

No. Branches, commits, diffs, and rollback are presented in plain English — no command line, no merge syntax. If you can use a doc tool, you can use Specor.

Is my product data secure?

Every workspace is isolated with row-level security, and the full event history is immutable and auditable. Enterprise plans add SSO/SAML, MFA, audit-log exports, EU data residency, and a self-hosted option where your data never leaves your infrastructure.

What happens to my data if I stop paying?

Nothing is deleted. Your history is preserved and read access stays free — viewers never need a paid seat. You can export your full product state at any time. Downgrading simply limits new edits and seats, not access to what you've built.

How do the AI agents use credits?

The six Spec Quality agents and Product Intelligence checks are event-driven and never consume credits — they run in the background for free. Credits are only used for generative actions you trigger, like generating a spec from a description or an AI chat. Each plan includes a monthly credit allowance.

Give your product a permanent memory

Bring a transcript, a PRD, or your repo — you'll have a living, versioned spec in one sitting.

No credit card Viewers always free Export your data anytime

Enterprise

Book a team walkthrough

We'll walk through Specor with your team, answer your questions, and scope the right plan for your organization. Usually 30 minutes.

We respond within one business day. No sales pressure — just a real conversation about your product workflow.

Best viewed on a larger screen

The Specor graph explorer works best on a desktop or tablet. Copy the link and open it there to see the full example.