Why now

The platforms decided AI will operate software. None decided how to let it in.

Six announcements in fourteen months, each advancing software operation — and each stopping at a different boundary.

Fourteen months
May 2025 · Microsoft BuildWindows adds typed actions for participating apps

With fallback computer-use tiers. The typed tier is opt-in, app by app.

Nov 2025 · Microsoft IgniteWindows framed as an “agentic OS”

Agents run in a separate, contained workspace — away from the desktop the person is using.

Mar 2026 · AnthropicClaude starts driving real Mac and Windows desktops

In preview, with governance at the session level.

Jun 2026 · Microsoft BuildA separate execution environment for AI agents

Built with OpenAI and NVIDIA. Separate from the desktop.

Jun 2026 · Apple WWDCSiri gains screen awareness

Typed App Intents remain vendor-declared — and it is Siri alone, on Apple’s operating systems alone.

Jul 2026 · OpenAIChatGPT starts acting on local files and apps

Governance at the session level.

Every platform validated the gesture. None combined installed-base reach, shared-desktop control, every desktop OS, and a check at the exact target. That combination is what InvokeFlow delivers, today. We extend these platforms — we don’t compete with them.

What we believe

Agents are ready to work. Your software isn’t ready to be called.

— David Richard and Raj Madugula, co-founders

The efficiency case

Faster tasks. Fewer model calls. A lower token toll.

A peer-reviewed systems study ran the same AI model on the same real desktop tasks twice. In one run the agent had to find and click. In the other it could name what it wanted and let a layer handle the mechanics.

44% → 74%

Task success, before and after the agent could invoke instead of click.

−43.5%

Fewer interaction steps for the same work.

61%

Of successful tasks finished in a single model call.

The AI didn’t get smarter. It got a better way to reach the software. Source: Wang, Li & Chen, EuroSys ’26.

Routine steps cost nothing to reason about

Steps that never needed judgement run deterministically, with no model call. The token toll on those is zero.

Fewer steps when reasoning is needed

Invoking a command is one step; finding and clicking it is several. Every step you don’t take is one you don’t pay for.

One call, not a loop

Most tasks finish in a single model call instead of a look-click-look cycle. Faster for the person waiting, cheaper for the budget.

Start here

Start with one workflow.

Point InvokeFlow at the applications it lives in. It runs governed and on the record from the first command.