Product Enterprise Roadmap Security Docs GitHub ↗
See Speedwave live PL

Roadmap

What we're building

What the Speedwave team is working on now, and what we plan next. Finished work lands in What's New.

Now

4 features

In development right now.

PII tokenization policies

In progress

Replace personal, financial and secret values with placeholder tokens before a request leaves the workstation. The strict, GDPR Art. 32 and EU AI Act Art. 5 templates are available behind the beta toggle, which anyone can switch on.

Meeting transcription

In progress

Record a meeting, transcribe it locally and turn it into structured technical notes. System audio and the microphone are captured as separate channels, and since 0.19 Windows transcribes on the GPU through Vulkan, with models matched to the hardware.

More tools for GitHub, GitLab and Reminders

In progress

GitHub and GitLab get more ways to write to a repository, such as one commit that changes several files. Version 0.19 added editing of macOS Reminders and GitLab issue comments and assignees; deleting reminders, recurrence and alerts are next.

Several chat tabs at once

In progress

Keep several chat sessions open in a tab bar and switch between them while each one keeps working. Today the Desktop app shows one conversation at a time and you resume the others from the conversations panel.

Planned

3 features

Next in line. Scope may still change.

On-prem management console

Up next

A self-hosted app for the organization. It collects telemetry from every Speedwave instance, pushes rules to workstations the way MDM does (PII tokenization policies, for example), and gives admins one place to analyze user logs.

Read-only mode for integrations

Up next

One switch that blocks every write tool across integrations, with a per-integration override, plus a read or read-write mode for Git inside the sandbox. For sessions that should look, not change.

Subagent activity in the chat

Up next

When Claude delegates work to subagents, the chat shows it as a highlighted block and a panel with the status of each one, so you can follow what runs in parallel.

Spike

4 topics

Investigating before we commit.

OpenAI with a subscription login

Exploring

Use an OpenAI subscription (a ChatGPT plan) as a model provider the way the Anthropic subscription works today, instead of a pay-per-token API key. We are checking what OpenAI's terms and login flow allow.

Built-in browser

Exploring

A web browser inside the Desktop app, so you can open and check web pages without leaving Speedwave. It is a Desktop feature, separate from the Playwright automation used for browser testing.

Encrypted credentials at rest

Exploring

Integration tokens, OAuth refresh tokens and the PII tokenization key are protected today by file permissions only. We are deciding how to keep them encrypted on disk, with a passphrase or with the operating system's key store.

Several LLM providers at once

Exploring

Today a project routes everything to one active provider. We are checking how to let you pick a model from any configured provider (Anthropic, OpenRouter, a local server) in the same project and session, with cost shown per provider.

Missing something? Open an issue on GitHub ↗ or tell us directly.