Connect Every System Once and Keep Its Permissions.
An AI platform is only as good as the data it can see. Squirro keeps yours current and checks permissions on every query.
Trusted by Industry Leaders
The Full Built-In Catalog.
- 70+built-in connectors
- 24self-service, set up in the interface
- 9source categories
- 7SQL engines through one connector
- Your ownsources, through the Python plugin SDK
71 connectors
Microsoft SharePointSelf-service
Microsoft ExchangeSelf-service
Atlassian ConfluenceSelf-serviceAtlassian JiraSelf-service
Salesforce Sales CloudSelf-service
Salesforce Service CloudSelf-serviceGoogle GmailSelf-service
SQL ConnectorSelf-serviceFigmaAssisted
- RLRAW LabsAssisted
Amazon S3AssistedElasticsearchAssisted
DocumentumAssisted
Adobe CloudAssistedApache SparkAssisted
HadoopAssisted
HortonworksAssistedClouderaAssisted
DropboxAssisted
SymphonyAssisted
Dow Jones NewswiresAssisted
FactivaAssisted
FactSetAssisted
RefinitivAssisted
S&P GlobalAssisted
MorningstarAssisted
LexisNexisAssisted
PitchBookAssisted
PreqinAssisted
MandateWireAssisted
Money-MediaAssisted
FUNDfireAssisted
FundmapAssisted
Fin|SearchesAssisted
StreetAccountAssisted- WWebhoseAssisted
CSV ConnectorSelf-service- XLSExcel ConnectorSelf-service
JSONAssisted
Documents ConnectorAssistedData Loader PluginAssisted
Python SDKAssisted
SquirroAssisted
Squirro Activity LogAssisted
Crawler ConnectorAssistedMicrosoft OneDriveAssisted
Google DriveAssistedRSS or Atom FeedAssisted
European Patent OfficeAssisted
ServiceNowAssistedZendeskAssisted
BMC RemedyAssisted
HP Service ManagerAssisted
FreshserviceAssisted
SamanageAssisted
Power BIAssisted
TableauAssisted
SAP Business IntelligenceAssistedMicroStrategyAssisted
SisenseAssisted
LinkedInAssistedX / TwitterAssisted
FacebookAssisted
InstagramAssisted
RedditAssisted
YouTubeAssisted
Stack OverflowAssisted
Microsoft DynamicsAssisted
NetSuiteAssisted
SugarCRMAssistedSquirro Activity InsightsAssisted
No built-in connector has that name. Ask us about a custom connector.
Every Source Connects in Four Steps.
Self-service or assisted, every connector runs on the same data loader.
-
01
Microsoft SharePointSite: Plant engineeringSigned in with OAuth2
Configure the source.
Name it and sign in.
-
02
FileLeafReftitleBodybodyModifiedcreated_atUniqueIdid
Map the fields.
Point its fields at title, body, date, and ID.
-
03Department: EngineeringPlant: KasselDoc type: 8D reportFiltersPermissionsAnalytics
Map the labels.
Source metadata becomes facets for filtering, permissions, and analytics.
-
0412,480+214+36+9+7106:0006:3007:0007:3008:00
First run loads everything. Each run after loads what changed.
Schedule the job.
Set how often it refreshes, then monitor it on the Data Sources page.
Three ways in.
Self-service, built in.
24 connectors a project creator sets up alone, in the browser. No ticket, no custom code, no release cycle.
Exchange
Confluence
SalesforceGmail
SQL
CSV
One-click OAuth2.
SharePoint, Exchange, Confluence, Jira, Gmail, and both Salesforce clouds, revocable at the source at any time.
SharePoint
Exchange
ConfluenceJira
Gmail
Salesforce
Assisted and custom.
Our team sets up Hadoop, ServiceNow, BI tools, and premium data. Anything not listed becomes a data loader plugin.
Hadoop
ServiceNow
Power BI
FactSet
S&P GlobalPython SDK
One connector, seven databases.
- Microsoft SQL Server
- MySQL
- PostgreSQL
- Oracle
- SQLite
- Sybase
- Firebird
Where Connectors Fit in Squirro.
Each connector brings a source into the AI context layer once, with its access rights. Search, agents, and the apps your developers build all draw on that one index, so a source connected for one use is ready for the next.
What Holds Up in a Security Review.
Visit the Security and Trust Center
Q3 margin review.pptxShared withFinance leadership
Access list kept:Finance leadership
Maria, Finance leadershipThe file is in her answer.
Tom, EngineeringSame question; the file never reaches his answer.
No stored passwords.
Enterprise connectors sign in through OAuth2. On Azure, SharePoint can run on managed identity instead.
Source permissions carried forward.
SharePoint access lists apply through Microsoft Entra ID, so people retrieve only what they could open at the source.
Incremental loads, not full reloads.
Each run loads only what changed since the last one.
Custom plugins, same treatment.
Data loader plugins get the same mapping and incremental loading as built-in connectors.
Two Ways to Put Enterprise Data in Front of an Agent.
MCP works when you already know where your data is. Ingestion indexes your repositories so your agents can find the rest.
| Compared | Ingested into Squirro | Called live over MCP |
|---|---|---|
| Typical question | “What do we know about X?” | “What is the state of X right now?” |
| Where the data lives | Indexed in Squirro | Stays in the source system |
| Preparation | Cleaned, enriched, and classified | None, raw data at query time |
| Scale | Proven at 10+ TB and 10,000+ users | Bounded by the context window |
| Access control | Bound at ingestion, applied at query | Enforced by the source system |
| Speed | Instant, already indexed | Live, as fast as the source |
| Best for | Enterprise search and analytics | Live data, and data that cannot be copied |
Many Squirro deployments run both.
Ingest.
Documents, policy, and correspondence. Anything people search for but cannot name in advance.
Call live over MCP.
Transactional records and system state. Anything that changes by the hour.
Decide case by case.
High-volume structured data that is rarely queried.
Administrators declare the MCP servers each project can use, and tools that change something need approval.
Ask Us About a Custom Connector.
Tell us what the system is and how it exposes its content. You will hear what it takes to bring it in.
Frequently Asked Questions.
Do source permissions carry over?
Squirro filters every query by the access each project grants. Where a connector carries the source system’s permissions, those apply per document as well, as SharePoint’s do through Microsoft Entra ID. On Squirro Cloud, SharePoint permissions need sign-in through Microsoft. Your engineer confirms this for each of your sources.
Does Squirro keep a copy of our data?
Squirro indexes the content it ingests wherever you run it, in Squirro Cloud, your VPC, or your premises. Where policy forbids a second copy, an agent can call the system live over MCP instead, and Squirro stores nothing. See the Deployment Options
Do new agents use the sources we already connected?
Yes. Search, chat, and agents work on the projects, content, and index your platform already has, so every source you connected is there for them, with the same permissions.