Interlinked.

Comparisons · measured July 2026

A search index lives in one of two places. A server you send code to, or the machine it's already on.

Sourcegraph indexes your repositories on a server and answers code queries across them. Glean does the same for your work (Drive, Slack, tickets, wikis) crawled into one searchable place. Both are genuinely good, and both make the same architectural choice: the index lives on a server, and your content travels to it. That one choice sets three things: privacy, cost, and freshness. Here is the picture, then exactly where a server wins and where a local index does.

TL;DR · key takeaways

The whole story is where the index lives: on a server your code is sent to (Sourcegraph, Glean), or on the machine it's already on (Interlinked). That one choice sets privacy, cost, and freshness.
A server wins for shared, org-wide knowledge: one canonical index the whole team queries. That's not a flaw; it's the cost of that shape.
A local index wins for “what's on this machine, right now”: the unpushed working copy, the non-git 90%, private and offline, fresh ~1 ms after you hit save.
Measured on one machine: a filename in 85 µs (139 µs median across 4.47M files); content search in ~7 to 9 ms where ripgrep took 93.8 s → 16 ms; ~44 MB at rest.
It's not either/or: keep the server for the org's cloud, add Interlinked for the local machine: every agent, one connector. Everything free forever, no card. Hosting starts at $5.99/mo.

The two shapes of search

Cloud index vs local index, drawn to scale.

The only real difference between these two designs is where the index box sits relative to the edge of your machine. Put it on a server and your code has to leave the machine to be indexed, and every query round-trips the network. Put it beside the files and nothing crosses the line. Everything else (privacy, price, how fresh the answers are) follows from that.

SERVER-SIDESourcegraph · GleanTHE INDEXlives on a server(their cloud, or one you host)▲ YOUR CODE LEAVES THE MACHINE① crawl + push② query round-tripYOUR MACHINEYOUR CODEworking copy on diskLOCALInterlinkedNO SERVERnothing lives off your machine✓ NOTHING CROSSES THE LINEYOUR MACHINEYOUR CODEworking copyTHE INDEXright beside itquery: never leaves
PRIVACY
SERVER
Content is copied to a server to be indexed.
LOCAL
Nothing leaves the machine. Fully offline.
COST
SERVER
Per seat, per month, plus servers kept warm.
LOCAL
Free forever. No seats. No server.
FRESHNESS
SERVER
Fresh only at the last push or crawl.
LOCAL
Fresh ~1 ms after you hit save.

The two famous server-side answers

What Sourcegraph and Glean are genuinely great at.

This is not a teardown. Both are the correct tool for the job they were built for: a team. The company's knowledge is shared, so its index should be too: one server everyone queries, kept consistent, permission-aware. The trade-offs that follow are not flaws. They are the cost of that shape.

SOURCEGRAPH
server-side code search

Point it at your organisation's repositories and it builds a cross-repo index: go-to-definition, find-references, and search that spans every repo your team works in, from one place. It runs as a deployment, cloud or self-hosted, and it indexes the committed, pushed state.

BEST FOR · Org-wide code intelligence · review · onboarding
GLEAN
server-side work search

Connectors crawl your company's apps (Drive, Slack, tickets, wikis, repos) into one permission-aware index on Glean's servers. Anyone can ask a plain question and get an answer drawn from everything the company has written down, respecting who is allowed to see what.

BEST FOR · Company-wide knowledge across every tool

One thing both share by design: they see what has been pushed or synced to the server, not the working copy on your disk this second, and never the files that were never in git or in a SaaS app to begin with. That gap is the whole story below.

The sharpest gap: freshness

A server is only as fresh as its last crawl.

You edit files continuously. A server index refreshes in bursts (when you push, or when its crawler next comes around), so between those moments it drifts out of date, and a query can return the version from an hour ago. A local index rides the live disk instead: change a file and it is searchable again in about a millisecond, under 30 milliseconds in the worst case.

SERVER-SIDE: fresh only at each crawlLOCAL: fresh continuouslystalefreshINDEX STALENESSTIME · A WORKING SESSION OF EDITS →↺ push / crawl≈1 ms behind

Every trade-off, row by row

Same question from both sides.

This is not “who is faster”: a server index and a local index answer for different scopes. It is a map of what each one costs you: where it lives, what it can see, and who gets to query it.

The trade-offServer-side · Sourcegraph · GleanLocal · Interlinked
Where the index livesa server: their cloud, or one you hoston the machine, on your disk
What it can seethe pushed / synced statethe live working copy + the non-git files
Freshnessthe last push or crawl cycle~1 ms after a save · under 30 ms worst case
Privacycontent is copied off the machine to indexnothing leaves the machine, works offline
Cost shapeper seat, per month, plus servers kept warmfree forever, no seats, no server to run
Footprinta deployment to scale and maintain~44 MB at rest · under 1% of the drive
Who can query itthe org, through its own surfacesevery agent on the machine: 19 clients
Best forshared, org-wide knowledgeyour machine, right now, and your agents

Where each one wins

It comes down to one question. Whose answer is it?

If the answer has to be the same for everyone, it belongs on a server. If the answer lives on your disk and has to be right this second, the index belongs on your disk too.

A SERVER WINS WHEN

the question is “what does the company know?”

·Shared across many people, one canonical index
·Spans machines: nothing depends on whose laptop it is
·The pushed monorepo the whole team sees (Sourcegraph)
·Every SaaS app the org runs (Glean)
·Code review, onboarding, cross-team discovery
A LOCAL INDEX WINS WHEN

the question is “what's on this machine, right now?”

·The working copy you have not pushed yet
·The non-git 90%: configs, downloads, that one PDF
·Code that must stay private and offline
·An answer that has to be fresh this second
·The agents on your machine that need it in microseconds

Glean for the org's cloud. Interlinked for the local machine. Same idea (search everything) pointed at the one place a server can't reach: the computer in front of you. Not a competition; different layers of the same stack.

The local index, concretely

Everything a server can't reach, indexed where it already is.

Interlinked keeps an always-fresh index of every file on the machine (names and contents) across every repo and the non-git 90%, and hands it to whatever agent you use through one connector. It never leaves the machine, and there is no model anywhere in the box: every result is an index lookup, not a guess. Here is what that costs and delivers, measured.

85 µs
to find a single file: 139 µs median across all 4.47M files
~7 to 9 ms
content search: a query that took ripgrep 93.8 s here returned in 16 ms
~1 ms
from saving a file to it being searchable, under 30 ms worst case
~44 MB
at rest in Task Manager · the index on disk stays under 1% of the drive
19
AI clients auto-configured in one install: every agent, one connector
26,958
relationship pairs seeded on install from 12/12 repos, 8,455 git events replayed in 1.7 s

A single filename lands in 85 µs; Windows Search takes 67 seconds median on the same machine, about 480,000× slower. And the vault seeds a fact no server-side index can represent: that a folder of separate repos actually ships together. Methodology: Ryzen 9 9950X3D · 64 GB DDR5 · NVMe · Windows 11 · 4.47M files. Full write-ups: the ripgrep comparison and how the vault seeds itself.

GLEAN FOR THE ORG'S CLOUD. INTERLINKED FOR THE LOCAL MACHINE: LIVE, PRIVATE, AND FOR EVERY AGENT.

Common questions

The questions people ask, answered straight.

Q
Does Interlinked replace Sourcegraph or Glean?
A

No. They index the org's shared, pushed knowledge on a server; Interlinked indexes the machine in front of you: the unpushed working copy and the non-git 90%. Keep both: they're different layers of the same stack.

Q
Does my code ever leave the machine?
A

Not for search, and not unless you publish a project yourself. The index lives on your disk beside the files, every query stays inside the machine, and it works fully offline. There is no indexing server to send content to, and no model anywhere in the box.

Q
How fresh is a local index next to a crawl?
A

A server index is only as fresh as its last push or crawl, so it can hand back an hour-old version. A local index rides the live disk: a saved file is searchable again in ~1 ms, under 30 ms in the worst case.

Q
Is it actually faster, or just closer?
A

Both. On the same machine a single filename lands in 85 µs (139 µs median across 4.47M files), about 480,000× faster than Windows Search's 67 s, and a content query that took ripgrep 93.8 s returned in 16 ms.

Q
What does it cost?
A

Everything (including agent access, the one connector that wires up all 19 AI clients) is free forever, no card. Hosting starts at $5.99/mo.

Index the machine a server can't reach

Keep Sourcegraph. Keep Glean. Add the machine in front of you.

One signed installer indexes every file on your machine (names and contents, every repo plus the non-git 90%) and wires up the 19 AI clients you already use. No cloud account, no keys, no JSON editing, and nothing ever leaves the disk. Your team's server index keeps doing its job; every agent on your machine gets a fast, live, private search alongside it.

Everything local is free, forever: the app, whole-machine search, MCP. Hosting starts at $5.99/mo. Download for Windows · One repo vs the whole machine · Why finding files is the bottleneck

Measured July 2026 · Ryzen 9 9950X3D · 4.47M files · Windows 11← All research