Flowgear ITWeb Preview
ITWEB-STYLE PREVIEW
Preview source: clients/flowgear/itweb/ITWEB_CASS_GOVERNED_AI_DRAFT.md. Taylor content pass + CI sweep 20 Jul 2026. No publish date; Pash picks the lead and timing.

ITWeb draft - governed AI integration demo

Status: Draft for Pash review (Cass-voice pass, 20 Jul 2026)

Recommended lead candidate: Yes - strongest narrative bridge from demo to market positioning

Source base: Cassandra Wallace demo script; flowgear.net messaging; existing Builder MCP draft

Editorial note for Pash

Rewritten 20 Jul after Dan/Pash feedback that the first draft read too AI-heavy. Goal: keep the demo facts and IDE positioning, but sound closer to how Cass walks the room through it - concrete steps, less manifesto language.

This piece uses Cass's client demo as the proof point. Flowgear is not selling a chatbot bolted onto integration. It is showing an Integration Development Environment where an assistant can help build, test, and later call workflows, while Flowgear stays the place credentials, approvals, and execution live.

Suggested hero asset: a clean still from the assistant or MCP tab in the demo. Product screenshot preferred over headshot.

YouTube: Cass demo recording exists at https://www.youtube.com/watch?v=2PCUg3et538 - do not cite in the submitted copy until Pash confirms ITWeb can reference it.

Draft copy

Build Integrations at The Speed of AI

A client demo walks through Salesforce leads to OneDrive briefs inside Flowgear, with the assistant doing the build work and the platform keeping control of connections, tests, and runs.

JOHANNESBURG - In a recent demonstration, Flowgear walked customers through a familiar sales ops job: a new lead lands in Salesforce, the team needs a structured handoff brief, and that brief has to land in Microsoft OneDrive where people will actually use it.

Cassandra Wallace, a Flowgear software engineer, showed the path end to end. Flowgear's built-in assistant helped stand up the first workflow, then tightened it so each upload left a clear outcome and a status trail you can query later. Nobody pasted credentials into a chat window. The assistant asked which approved connections it could use, assembled the canvas, ran a test, and waited for a human to accept the change.

The point is to show that AI can speed up the build, but Flowgear still owns permissions, connections, testing, execution, and the audit trail. That lets an organization move faster while keeping credentials and data access governed.

"AI can now help teams get to a working integration faster, but speed is only useful if the result can be inspected and operated," said Daniel Chilcott, co-founder and CEO of Flowgear.
"Flowgear gives assistants a governed place to build, rather than asking teams to move business logic into disconnected scripts."

AI Assistant Capabilities

In Cassandra's walkthrough, the assistant is not just a bot that can answer questions in chat.

It can read the workflow canvas, find the nodes and templates on offer, ask which connections are fair game, put a workflow together, test it, and explain what it wants to change before anyone hits accept.

That matters when the work touches production data. A generic coding assistant can spit out scripts quickly. Integration work still has to respect credentials, ERP and CRM dependencies, exception handling, and evidence for auditors. Flowgear's model is narrower on purpose: the assistant only gets the tools Flowgear exposes, not blanket access to every secret in the tenant. You stay in review, and your data stays governed.

Builder MCP and Workflow MCP

The same pattern shows up outside the Flowgear Console through Model Context Protocol (MCP). Flowgear demonstrates two different uses.

Builder MCP is for creating and testing workflows from an external agent or IDE. The agent can drive Flowgear's build surface. The integration logic still lives in Flowgear.

Workflow MCP is for calling a finished workflow as a tool. In the demo, a chat assistant pulls live Salesforce lead context through a Flowgear workflow that someone deliberately exposed - not through a free-range API key.

Short version: Builder MCP helps you build the integration. Workflow MCP lets an assistant use an integration you already approved.

Thin Layer versus Durable Layer

Flowgear frames the choice as a thin layer versus a durable layer. A thin layer is the brittle glue an agent can knock together in an afternoon: fast to start, hard to operate, easy for credentials and one-off scripts to leak into the internet. A durable layer is where AI reaches business systems through scoped tools, approved connections, logs, diagnostics, and a review step before anything ships.

AI tools can speed up the early stages of building an integration, but you still need a system that owns and manages the finished solution. Flowgear's approach is that the Integration Development Environment should be that owner: it lets AI speed up your work, while still keeping all logic, credentials, and control within the platform, where IT teams have full visibility.

Why South African Businesses Care

Plenty of South African organizations run mixed estates - Microsoft stacks, ERP and finance platforms, logistics, local payment rails, custom databases, customer apps. When those systems do not communicate, people fill the gaps with spreadsheets and chase-ups.

AI can cut some of the build friction. It can also multiply shadow automation if every team invents its own backend. The demo argument is practical: give teams a faster way to build, and keep the controls you need once the workflow is in production.

"The future is not a choice between AI speed and enterprise control," said JJ Milner, co-founder of Flowgear. "The winning pattern is to let teams build faster through a governed integration environment, so the business gets momentum without losing the system of record."

About Flowgear

Flowgear is an Integration Development Environment for building, running, and governing business integrations across cloud, hybrid, and on-premises systems. The platform combines a visual workflow designer, pre-built connectors, AI-assisted build experiences, Builder MCP, Workflow MCP, and observable runtime infrastructure so teams can integrate systems quickly while keeping control over credentials, approvals, and execution. Founded in South Africa in 2010, Flowgear supports customers and partners across multiple markets.

Edit log