Back to Blog
Flowly

Flowly Gets Native Apps, a New Look, and Smarter AI

FlowlyFlowlyΒ·30 Mar 2026
Flowly Gets Native Apps, a New Look, and Smarter AI

This is Flowly's biggest update. We're not talking about a button color change β€” we're talking about native apps on every platform, a complete design overhaul, smarter AI under the hood, and a billing system that actually makes sense.

Here's everything that's new.

Native Apps: iOS, Android, and Desktop

Flowly is no longer a web-only platform. You can now manage your AI assistant from:

The mobile apps aren't web wrappers. They're built native β€” SwiftUI on iOS, full push notifications, real-time chat, and instant server management. If your bot gets a message at 3am, you'll know about it.

The desktop app goes further: you can run your entire AI assistant locally on your machine. No cloud server, no monthly VPS bill, no data leaving your computer. Just you and your AI.

Why This Matters

Most AI platforms force you into their cloud. Flowly gives you the choice: deploy to a VPS for 24/7 availability, or run everything locally for maximum privacy. Either way, you manage it all from the same app.

A Design That Feels Right

We rebuilt the interface from the ground up, starting with our iOS app's design language and bringing it everywhere:

  • Capsule buttons β€” Pill-shaped, 56px tall, with subtle inner glow and drop shadows. They feel tactile.
  • Larger corner radii β€” Cards and containers now use 20-28px radii, matching the iOS aesthetic.
  • Bolder typography β€” Headlines use font-weight 900 with -3% letter spacing. They hit harder.
  • iOS-style shadows β€” Deeper, more intentional shadows in light mode. Clean borders in dark mode.
  • New hero background β€” The landing page got a complete visual refresh.

Every button, card, and badge is now consistent across the web dashboard, mobile apps, and landing pages. One design language, everywhere.

Smarter AI, Lower Costs

Anthropic Prompt Caching

We implemented Anthropic's prompt caching system. Your system prompts and recent messages are cached, which means:

  • Up to 90% reduction in token costs on cached reads
  • Faster response times for repeated context
  • Automatic for all Claude models β€” no configuration needed

If you're on a BYOK plan with your own Anthropic key, you'll see the savings immediately.

Multi-Agent Profiles

Run multiple isolated AI instances from a single installation. Each profile gets its own:

  • Memory and conversation history
  • Custom persona and system prompt
  • Skill set and tool access
  • Configuration directory

Think of it as having a work assistant, a personal assistant, and a coding assistant β€” all separate, all on the same machine.

Full-Text Search Over Conversations

Every conversation is now indexed with FTS5 full-text search. You can search across your entire chat history with:

  • BM25 ranking for relevance
  • Snippet extraction with context
  • Support for agglutinative languages (Turkish, Finnish, etc.)
  • Incremental indexing that stays in sync

Your AI can also search its own past conversations before responding, so it doesn't repeat itself or forget what you discussed last week.

Self-Improving Pipeline

Flowly's AI can now review its own conversations and create reusable skills from patterns it notices. It's a background process that runs after a configurable number of turns:

  • Automatic skill creation from repeated tasks
  • Trajectory export in ShareGPT format for fine-tuning
  • Memory nudges to improve long-term context

Security That Doesn't Compromise

End-to-End Encryption

All messages are encrypted in transit and at rest. When you use the desktop app locally, your data never touches our servers. Period.

Prompt Injection Prevention

We scan all bootstrap and persona files for 12 known attack patterns plus 10 invisible unicode injection techniques. Memory writes are guarded too.

Hardened Infrastructure

  • Rate limiting with rolling 5-hour windows and weekly caps
  • crypto.randomBytes for all code generation (no more Math.random)
  • CORS restricted to known origins
  • Desktop bot authentication via gateway tokens

Flowly Bot: Now Complete

The Flowly Bot β€” our lightweight alternative to OpenClaw β€” is officially out of beta and marked as completed on our roadmap. It's faster to set up, uses fewer resources, and connects to more platforms:

  • Telegram, WhatsApp, Discord, Slack, X/Twitter
  • Voice calls via Twilio
  • Speech-to-text via Groq
  • 8 built-in personas
  • Gmail integration

If you want a focused conversational assistant without the full agent capabilities of OpenClaw, Flowly Bot is the way to go.

What's Next

Here's what's on the roadmap:

  • Desktop App β€” Native macOS and Windows apps (in progress)
  • Voice Messages β€” Send and receive voice in chat
  • Team Support β€” Share AI assistants across your organization
  • API & Webhooks β€” Integrate Flowly into your own applications

Try It Now

Flowly is free to start. Deploy your first AI assistant in under 2 minutes:

  1. Sign up at useflowlyapp.com
  2. Choose your bot (OpenClaw or Flowly Bot)
  3. Click deploy
  4. Download the app on iOS or Android

No credit card required. No DevOps knowledge needed.

Everyone deserves a Jarvis. Now you can carry yours in your pocket.

Built by Nocetic. Follow us on X.