The retrieval trifecta for generative AI. Graph. Vectors. Filesystem.
$ pip install grover
01 — Brand Mark
Grover
Rove through your data
The Mark
The logomark is a directed graph that reads as a tree — roots at the bottom, branching upward. Each node carries a pillar color: ember (entry point), green (graph), violet (vectors), cyan (filesystem).
The dashed horizontal line represents the hidden connections that vector similarity reveals — structure that isn't visible in the explicit graph.
Set in DM Serif Display, the wordmark carries warmth and confidence. The serif conveys craftsmanship; the weight says production-ready.
Grover
Dark
Grover
Light
Grover
Accent
02 — Color System
Forest — Core Palette
Canopy
#1A2F23
Primary background
Understory
#243B2E
Cards, surfaces
Moss
#2D4A38
Elevated surfaces
Fern
#4A7C5C
Secondary accent, graph
Sage
#7BAE8E
Muted text, borders
Mint
#A8D5BA
Code highlights
Meadow
#D4F0E0
Body text
Daylight
#F0FAF4
Headings, emphasis
Signal — Pillar Accents
Ember
#E8734A
Primary CTA, entry point
Violet
#8B6CC1
Vector search pillar
Cyan
#4ABCE8
Filesystem pillar
Amber
#E8B44A
Warnings, highlights
Neutral — Structure
Bark
#1A1A18
Code backgrounds
Stone
#2A2A27
Code blocks
Cloud
#F5F5F0
Light mode background
Snow
#FAFAF8
Light mode surface
03 — Typography
DM Serif DisplayDisplay / Headings
Rove through your data. Grover.
A refined transitional serif with optical precision. Used for the wordmark, headings, and moments of emphasis. The contrast between thick and thin strokes mirrors the relationship between graph structure (heavy, defined) and vector similarity (subtle, continuous).
Instrument SansBody / Interface
Grover combines in-memory graph storage, vector search indexing, and a virtual filesystem into a single retrieval engine. Agents navigate your data the way they already know how — through files and directories — while Grover handles the traversal and similarity matching underneath.
Clean, contemporary, and highly legible at all sizes. The humanist proportions keep docs and UI feeling approachable rather than clinical. Available in weights 400–700.
JetBrains MonoCode / Technical
from grover importGrover, Node, Edge# Initialize with your data
g = Grover()
g.add_node(Node(id="doc-1", text="quarterly report"))
g.add_edge(Edge(src="doc-1", dst="team-eng"))
# Mount as a virtual filesystem
fs = g.mount("/data")
# Agents navigate naturally
results = fs.ls("/data/teams/eng")
similar = fs.search("revenue forecast", top=5)
Display XL72pxGrover
Display48pxThe retrieval trifecta
Heading28pxGraph-native retrieval
Subheading20pxBuild wherever your data lives
Body16pxAgents navigate your data through files and directories.
Caption13pxMIT Licensed · Python 3.10+
Mono14pxpip install grover
04 — The Trifecta
Three retrieval modes. One interface. Each pillar carries its own accent color.
⬡
Graph Storage
In-memory directed graph optimized for traversal. CSR format, NetworkX-compatible. Your data's real structure — not LLM-extracted guesses.
◎
Vector Index
HNSW similarity search over node embeddings. Find what's semantically close, then traverse the graph to find what's structurally connected.
⌸
Virtual Filesystem
Mount your graph as a navigable directory. Agents use ls, read, search — tools they already have. No new query language to learn.
05 — UI Components
Buttons
Badges / Pillar Tags
GraphVectorFilesystemv0.1.0
Code Inline
Initialize with Grover() and mount your graph using g.mount("/data"). Search with fs.search("query").
Form Elements
06 — Light Mode
The forest palette inverts cleanly. Canopy becomes the text color; Cloud becomes the canvas. Same warmth, same hierarchy.
Grover
The retrieval engine that agents already know how to use.
import grover
# Build your graph wherever your data lives
g = grover.from_postgres("postgresql://localhost/mydb")
# Mount and explore
fs = g.mount("/data")
fs.ls("/")
# → ['teams/', 'projects/', 'documents/']
07 — Voice & Tone
Confident. Clear. Never corporate.
Grover speaks like a senior engineer explaining something to a peer — direct, opinionated, and respectful of the reader's intelligence. No hype. No hand-waving.
✓ This
"Grover mounts your graph as a filesystem. Agents use ls and read — tools they already have."
"Vector search finds what's similar. Graph traversal finds what's connected. You need both."
"pip install grover. That's the whole ops story."
✗ Not This
"Grover leverages cutting-edge AI to revolutionize the retrieval paradigm."
"Our enterprise-grade solution enables seamless integration with your existing infrastructure."
"Unlock the full potential of your data with Grover's powerful platform."
08 — Spacing Scale
Based on a 4px grid with deliberate jumps. Generous whitespace is a core brand value — the design should breathe.