Gecko AI Platform

The governed AI Operating System.

Most institutions don’t have an AI ideas problem — they have an AI operations problem. Gecko gives the institution one governed place to build any agent, connect it to trusted knowledge and systems, choose the models it runs on, and deploy it wherever people ask for help.

Book a conversation See the architecture
7model provider families
22+built-in integrations
20workflow step types
32guardrail detectors
The architecture

Every layer an institution needs to run AI in production.

Channels on top, agents and workflows in the middle, your knowledge, models and systems underneath — and one layer of governance and observability running through all of it. Select any part to explore it.

DisplaysChannels
Web embedsChat bubble, inline chat, AI search Teams & SlackThreads, DMs, approval cards VoiceAI receptionist, outbound calls SMS, email & APIMessages, REST API, SDK
ApplicationsAgents & workflows
AgentsPrompt, model, knowledge, tools Prompt libraryFragments, versions, pinning Workflows20 step types, versioned runs AI buildersFrom brief to reviewable blueprint
DriversKnowledge, models & systems
KnowledgeSites, SharePoint, drives, files Models7 provider families, your keys or ours MCPAny server, per-tool grants IntegrationsEllucian, Salesforce, Microsoft 365
KernelGovernance & observability
Access & secretsSSO, scoped roles, encrypted vault GuardrailsFlag, redact or block input Human-in-the-loopApprovals and identity checks Tracing & auditEvery run, step and change
Why “operating system”

Not another chatbot. The layer every AI use case runs on.

An operating system lets many applications run safely on shared infrastructure, with permissions, resource control and a view of what’s running. That is exactly the job.

Applications

Agents and workflows — built by IT, by trained departmental teams, or drafted from a plain-English brief by the AI builders.

Drivers

Integrations and MCP servers that make institutional systems agent-ready — Ellucian Ethos, Salesforce, Gecko Engage, Microsoft 365 and your own.

Kernel controls

Scoped roles, expiring API tokens, input guardrails, approval gates, an encrypted secrets vault and per-tenant model allowlists.

Task manager

One run explorer for every agent and workflow: status, duration, tokens, tool calls, guardrail hits and human-review time.

Multiple displays

One agent on the website, in Teams and Slack and on the phone — same knowledge, same tools, same guardrails.

System log

An audit log of every create, update and delete across the platform, attributed to the person or API key that made it.

Capabilities

Explore each part of the platform.

Each capability is useful on its own. Together they turn pilots into a repeatable institutional service.

Agents & AI builders

Versioned prompts, approved models, scoped knowledge and per-tool grants — or a blueprint drafted from a plain-English brief.

Explore agents

Workflows

20 step types — AI, retrieval, branching, MCP tools, messages, calls, sandboxed code and human checkpoints.

Explore workflows

Knowledge & retrieval

Websites, SharePoint, cloud drives and uploaded files, searched with hybrid retrieval and cited with relevance scores.

Explore knowledge

Models & integrations

Seven model provider families, 22+ built-in integrations and open MCP for anything else — with IT curating exactly what builders can use.

Explore integrations

Channels & voice

Web embeds, Microsoft Teams, Slack, an AI phone receptionist, SMS, email and the API — from one agent.

Explore channels

Observability

Every run, step, token, tool call, guardrail hit and approval — filterable in one explorer.

Explore observability
TrustSecurity, governance & data residency BuildDevelopers — API, SDK, webhooks & MCP
No lock-in, by design

The model is a setting, not a marriage.

The best model changes every quarter. Institutional AI shouldn’t be a one-way door — so here are the receipts.

  • Swap the model behind a live agent from a dropdown — no rebuild, no migration
  • Tools use MCP, the open standard — connectors you build outlive any vendor
  • Workflows export and import as JSON — your process logic is yours
  • No hidden private API — the console runs on the same public API and SDK you get
How model governance works

IT curates the model list. Builders pick from what IT approved. Change your mind next quarter without rebuilding a single agent.

Foundations

Enterprise infrastructure underneath.

Independent regional deployments, queue-driven workers that scale with demand, and an OpenAPI-described platform your own teams can build on.

eu-west-1IrelandOwn database, secrets and domain
us-east-1United StatesOwn database, secrets and domain
ca-central-1CanadaOwn database, secrets and domain

Built to scale

Crawling, embedding and workflow execution run in background workers that autoscale on queue depth.

Hardened by default

SSO-only console, AES-256-GCM encrypted secrets, and outbound requests checked against private and reserved network ranges.

Inclusive console

A multilingual interface, light and dark themes, and an accessible outline alternative to the workflow canvas.

Trust & securityFor developers
PlatformAgents & AI builders PlatformWorkflows SolutionsUse cases by team
See it end to end

Walk through the platform with your own use case.

We’ll show how one agent is built, connected, governed and deployed — and what it would take for your institution.

  • A 30-minute conversation, not a sales deck
  • A scoped first use case and the controls it needs
  • Assurance documents in our Trust Center
Book a conversation