What's new in May

Shipping trust in an AI-first world

In automation and AI, speed is only useful if you can trust what ships. Teams are now building software in conversations, not IDEs, and the feedback loop between idea and impact has collapsed. That puts a spotlight on the unglamorous parts of product quality: instant publishing, previews that reflect reality, and a chat surface that never flakes. This month, we tightened those screws so you can move fast without second guessing.

The problem: Fast is fragile when basics wobble

We hear the same refrain from founders, ops leads, and agency builders: I can describe what I want, but the tools slow me down. You paste a prompt, hit publish, and wait. You open a preview and it shows an old or broken version. You invite a teammate and onboarding snags. You try to upload files and the app sends before they finish. You connect a tool twice and the session freezes. Each little wobble steals momentum and confidence.

Even the leaders in the space feel the pressure to smooth these edges. Teams using Zapier and n8n want clarity on what will run in production. Folks exploring agents in Claude.ai or rapid builders like Lovable expect previews to be accurate and responsive. Product teams inspired by Linear know the bar for UX polish is rising. If the foundation hiccups, trust evaporates and cycles stretch.

What we built: Faster, clearer, steadier

This release focuses on reliability where it counts, from publish to preview to the conversation itself.

• Publish with zero wait: The publish window opens immediately so you can keep momentum and confirm details without staring at a spinner.
• Transparent previews: While your app preview builds, you now see a clear Deploying preview... message. It refreshes automatically the moment it is ready.
• Always the latest view: Previews now reliably show your newest version rather than occasionally surfacing outdated or broken states.
• Smoother first-time setup: Clearer instructions guide new users through initial steps, and we fixed an issue that blocked some from completing onboarding.
• Consistent text highlighting: Selecting text in chat now works across the app, including on dark backgrounds where it used to disappear.
• Calm navigation: The sidebar no longer jumps when you collapse or expand it, so your focus stays put.
• Smarter file sending: If you hit Send before files finish uploading, the app tells you and lets you choose to send without them.
• Tool connections that stay stable: Reconnecting or linking a tool that is already connected no longer causes the chat to freeze.
• Chat resilience under load: We fixed several issues that could crash long conversations, very large responses, or certain image uploads. If a problematic image repeats errors, the app now recovers automatically instead of looping.
• Reliable remixing: Remixing apps and workflows shared by others now works consistently so community examples become jump-starts, not detours.

Why it matters: Reliability is the new speed

The industry is moving from click-by-click builders to conversational creation. That shift changes what speed means. It is not just fewer steps; it is fewer doubts. You should be able to say it, see it, ship it. If a preview lags or lies, you slow down to confirm, test, and re-check. If onboarding falters, teammates drop off. If the chat crashes during a key handoff, work pauses.

Accuracy in previews sets a new expectation for production confidence. Instant feedback on file uploads prevents accidental misses. Recovering from bad assets keeps sessions alive. These are not nice-to-haves; they are the control surfaces of AI-era development. Platforms that nail them become the default canvas for building internal tools, client automations, and lightweight products.

The effect compounds with your ecosystem. Tooling like Composio expands what agents can reach. Orchestrators like Zapier and n8n help stitch data together. But without a trustworthy preview and resilient chat loop, you still end up tapping the brakes. Our focus this month is removing those brakes so every conversation progresses toward a working, shippable system.

Why CodeWords: Real code, zero coding

CodeWords starts from a simple premise: Talk to it like ChatGPT. It builds like a developer. It runs like infrastructure. That means we optimize not just for model quality, but for the full lifecycle across planning, building, verifying, and deploying. The publish flow is instant because flow-state matters. The preview is deterministic because you need to know what will run. The chat holds up under stress because long sessions are where real systems emerge.

You have told us what matters, and we build to those Voice of Customer patterns:

• Minutes, not months: Teams report building in 25 minutes what took a day elsewhere. Every removed delay protects that edge.
• Trust the preview: You want to invite stakeholders without caveats. Always-fresh previews make live reviews safe.
• It just works: Reliability becomes emotional relief. When highlighting, sidebars, and uploads behave, your brain stays on the work, not the tool.

This release raises the baseline for polish in conversational development. Where some tools focus solely on breadth of integrations or raw model prompts, we sweat the stability work that turns demos into dependable systems. It is a different kind of speed—the kind you can bring to clients and teammates without disclaimers.

Get started

If you left a build mid-flight because a preview stalled, come back and try publishing again. Start a long conversation, upload big files, remix a shared app, and watch the surface stay smooth. Connect your favorite tools and push to production with confidence.

Open CodeWords, describe what you want, and ship something real today. Minutes, not months. Real code, zero coding. One place for everything.