Skip to main content
Secure enterprise AI gateway for regulated industries

Connect company data to AI
and control every use of AI

From company knowledge search to agents, harnesses and coding tools, every AI request from employees is sent to the model through Syncgate. Syncgate can be installed on air-gapped networks, and its personal data pseudonymization and sealed audit log support the controls that Korean AI security guidelines for financial institutions and public bodies require.

Contact sales
  • SampleChat

    Summarize this year's security audit findings and their deadlines

    Passed — delivered as written · Access checked · 2 excluded

  • SampleChat

    Please refund the card payment 4111-1111-1111-1111[CARD_1] for customer Minsu Kim (kim.minsu@acme-logis.com[EMAIL_1], 010-2345-6789[PHONE_1]).

    Pseudonymized — detected personal data replaced before delivery

  • SampleAgents

    Tool result · prompt injection detected

    Blocked — a policy violation was detected, nothing is delivered

Model management · model types
ChatEmbeddingRerankerImageGuardNEWDiscriminative
System One API connections
JJevTypeSafeProvider API
JJevVercel AI GatewayGateway
JJevDigitalOceanCloud
OOpenJevCodivHosted
OOpenJevvLLMSelf-hosted
KKevOllama · vLLMSelf-hosted
New

Discriminative models, starting with Jev

Jev, released by TypeSafe in September 2026, is a new class of model: it generates no text and returns one of the defined options with a probability. Syncgate adds discriminative models as a model type in model management to support it.

  • Decisions with probabilitiesYes/no, choice and graded-score answers come with a probability, and thresholds decide the action.
  • Applied where it mattersGuardrail AI evaluation, auto evaluation and the search reranker use it instead of a chat model.
  • One API, many connectionsConnect TypeSafe, Vercel AI Gateway, DigitalOcean and self-hosted OpenJev or Kev the same way.
Learn more
38Document sourcesWikis, storage, messaging, ticketing and public data
23Model providersVendor APIs, cloud, gateways and self-hosting
13Personal data types detectedIncluding Korean formats such as resident and business registration numbers
13Guardrail enforcement pointsFrom chat input to files, APIs and the code gateway
11Coding toolsIncluding Claude Code and Codex
10DatabasesIncluding PostgreSQL, Oracle and Snowflake
17Harness templatesIncluding 6 for financial services
361 pagesManualKorean, English and Japanese

Figures reflect the product as of October 2026.

About Syncgate
Syncgate is an AI work platform you install on your company's servers or on an air-gapped network. Employees get answers based on company documents and data, and automate repetitive work. Administrators and security teams decide which models and data each user can use, and usage is recorded.

Architecture

Every AI request from employees
is sent to the model through Syncgate

See the platform overview

Used from

ChatAgentsHarnessesMicrosoft TeamsWeb widgetAPIMCPCoding tools11
Syncgate
The gateway that handles every AI request
Permission checkOnly data within the user's permissionsGuardrailsChecks for personal data and risky requestsModel gatewayApproved models · usage limitsLoggingTraces · audit log

Connects to

Document sources38Databases10Tools and MCP serversModel providers23

Administration and operations

Administration and auditSSO · SCIMInstallationOperator console

Select an item to go to its feature page.

How it works

Connect, search,
automate and control

  1. Connect data and tools

    Connect 38 document sources, 10 databases and tools from your company systems. Documents from Confluence, Jira, Slack, Box, Dropbox and others can keep the permissions of their source.

    Document connectorsDatabasesTool connections

  2. Search company knowledge

    Employees ask in chat and get answers with a source cited for every sentence. The glossary and knowledge graph tell the AI how company terms and data relate.

    Knowledge searchGlossaryKnowledge graph

  3. Automate repetitive work

    The ways of working that skilled AI users built up on their own become harnesses the whole team uses. Assign a task in chat, with no command line, and the harness produces the result files and checks them.

    HarnessesAgentsScheduled tasks

  4. Control AI use

    Syncgate checks permissions on every request and applies guardrails, and the model gateway handles model calls. Each request is recorded in the audit log.

    GuardrailsModel gatewayAdministration and audit
20 features

The features you need, from work automation
to installation and operations

See every item in detail

Work automation5

Hand repetitive work to AI and get the results as files.

HarnessesWork environments your team shares, taking each task through to result filesAgentsBuild AI assistants for each departmentChat and unified agentFinds the documents and agents that fit each question, then answersSkillsSave how a task is done as instructions and share themMemoryRemembers the knowledge you share about your work and uses it in answers

Control and security5

Decide who can use which models and data, and check and record what is sent to and from AI.

GuardrailsMasks personal data and blocks risky requestsPermissions and share reviewAnswers within each user's permissions and shares only after approvalModel gatewaySet who can use each model, and its limitsCode gatewayApplies company policy to developers' coding toolsEvaluation and tracingScores answer quality and records how each request is processed

Company knowledge and data5

Connect scattered documents and business data, and answer with the evidence cited.

Document connectorsBrings in and indexes documents from 38 document sourcesKnowledge search and cited answersAnswers with a cited source for every sentenceDatabase queriesQueries business databases in natural language and answers with tables and chartsGlossaryTells the AI what your company's terms meanKnowledge graphAnswers based on the relations between data and policies

Integration and operations5

Integrate with company systems, and install it on your own servers and run it yourself.

Tool connections and MCPConnects company systems as tools the AI can useDistribution channelsUse it in Teams, a chat widget or the API, and manage the agents you already runCollaboration and scheduled tasksWork together in projects and run scheduled tasksAdministration and auditManage users, sign-in, audit logs and retention periodsInstallation and operationsInstall and update on company servers and air-gapped networks
Solutions

Ready to use in financial and public-sector work

Industry · Financial services

Control and record AI use in line with financial security guidelines

Pseudonymization of personal (credit) information and a sealed audit log support the controls that Korean financial-sector AI security guidelines require.

GuardrailsModel gatewayAdministration and auditInstallation and operations
See financial services solutions Industry · Public sector

Public-sector AI that meets security classification and air-gapped network requirements

Classifies documents into Korea's N2SF Classified, Sensitive and Open levels and answers in Hangul (HWPX) documents. Installation has also been verified on a network with no internet connection.

See public-sector solutions
Departments

Browse by department

Compliance and internal controlInformation security and auditIT and development
Korean financial and public-sector AI security guidelines

Financial and public-sector security
requirements, built into the AI gateway

See security and compliance

Personal (credit) information detection and pseudonymization

FSC guideline check 7 · FSS AI RMF privacy · NIS guidebook M13

Layered defense against prompt injection and jailbreaks

FSC guideline checks 6, 9 · NIS guidebook M13, M15

Identifying security classification levels (C · S · O)

N2SF Model 2 “AI integration system” · IF-14

Log retention and tamper evidence

National Cybersecurity Basic Guidelines Art. 55 · NIS guidebook M09 · Compensating controls for financial network separation

Approved models only

NIS guidebook M17 · FSC guideline check 12 · N2SF Model 2

MFA, least privilege and human approval

FSC guideline check 24 · Basic Guidelines Art. 76 · NIS guidebook M05, M30, A-M12

See it with
your own documents

In a consultation, we connect a few of your documents so you can see the answer quality for yourself. Leave your question below and our team will get back to you.

Inquiry type

We collect your name, company, email, phone number and message to answer your inquiry, keep them for one year after we reply, and then delete them. Without your consent we cannot receive the inquiry.

Product videosReal product screens · Example company ACME Logistics · 16 videos