Skip to content
All tool news

Tool desk · Updated daily

Tool news·Bubble·

Bubble Ships Sprint 7 With Image Upload for AI Agent and Numbered Proposal Cards.

Bubble Sprint 7 adds image upload for AI Agent, structured proposal cards with numbered steps, end-to-end permission enforcement, and mobile improvements.

CW

Create With tool desk

1 source checked · 3 min read · 6 sections

ShareLinkedIn
Bubble Ships Sprint 7 With Image Upload for AI Agent and Numbered Proposal Cards

Agent Gets Eyes, Mobile Gets Tappable Inputs

Bubble announced on LinkedIn the release of Sprint 7, bringing the most practical improvement yet to its AI Agent: the ability to upload images. Instead of typing out descriptions of UI or data structures, users can now attach screenshots directly. The update also restructures how the Agent presents its work, transforming text-heavy proposals into numbered plan cards that show exactly what changes are being made.

The release follows Sprint 6's infrastructure overhaul, which introduced faster iteration loops and a redesigned conversation model. Sprint 7 appears focused on usability: making the Agent easier to communicate with and easier to understand.

Image Uploads Let Users Show Instead of Tell

The standout feature is image upload for the AI Agent, now available to everyone. Users can share mockups, screenshots of competing apps, or photos of data tables and let the Agent interpret them visually. The change addresses a longstanding frustration: trying to describe layout, color schemes, or complex workflows in text.

Bubble hasn't detailed the technical implementation, but the move suggests expanded multimodal capabilities in the underlying model. For builders prototyping quickly or iterating on design, attaching a screenshot is faster and less ambiguous than writing paragraphs of specs.

Proposals Now Render as Structured Plan Cards

Edit proposals, previously presented as conversational text, now render as structured cards with numbered steps. The change makes it easier to scan what the Agent intends to do before committing. Each card breaks down individual actions: create this field, update that workflow, adjust this styling rule.

The numbered format also helps with troubleshooting. If a proposal doesn't do what a user expects, they can point to a specific step rather than rereading a full paragraph. It's a small UI shift that speeds up iteration.

Permission Enforcement Goes End-to-End

Sprint 7 implements end-to-end permission enforcement for the Agent. Bubble hasn't published granular details, but the language suggests the Agent now respects role-based access controls throughout the editing process, not just at the surface level. That's critical for teams using Bubble for client work or internal tools with strict access policies.

The update aligns with Bubble's broader push toward enterprise features, including the property editor overhaul and version control improvements rolled out earlier this year.

Mobile Improvements: Deep Linking and Input Tappable Anywhere

On the mobile side, Sprint 7 ships push notification deep linking and makes form inputs tappable across the entire field, not just on the label or cursor position. Both changes improve the experience of apps built with Bubble's native mobile package.

Deep linking, previously added as an action, now works with push notifications, meaning users can tap a notification and land directly on a specific screen or record. The input field change fixes a common complaint: users had to tap precisely on a small target to activate text entry.

These tweaks signal Bubble's continued focus on mobile parity. Native app development on the platform has lagged behind web for years, but recent updates suggest the gap is narrowing.

What This Means for Builders

Sprint 7 isn't a headline-grabbing overhaul, but it addresses real friction points. Image upload makes the Agent more practical for visual thinkers and rapid prototypers. Structured proposals reduce cognitive load when reviewing AI-generated changes. Mobile improvements chip away at the platform's weakest area.

The release also reflects Bubble's sprint cadence: smaller, faster updates instead of monolithic launches. That approach keeps the Agent evolving without long gaps between improvements. For users building on Bubble, the pattern suggests more incremental gains ahead as the company refines its AI tooling and mobile capabilities.

Sources

1 checked

How we cover tool news: Create With's tool desk drafts these reports with AI from the sources listed above and checks them against those sources before publishing.

Worth passing on?

ShareLinkedIn

Go deeper on Bubble

Related reading, watching and going.

Everything on Bubble →

The briefing

26 May 2026

Create With Conference 2026: AI Agent Community Event Brighton

Join Create With Conference 2026 in Brighton for insights on AI agents and vibe-coded software. Partnering with EVOLVE for two days of tech innovation.

Read →

The briefing

12 May 2026

The Ultimate Cheap & Scalable Stack for AI Agent Development

Discover the latest AI agent tools and vibe-coded apps from No-Code Poland. Learn about Bubble's AI assistant, Lovable's security standards, and more.

Read →

The briefing

21 Apr 2026

Claude Design vs Figma Make vs Google Stitch: AI Design Tested

We tested Claude Design against Figma Make and Google Stitch to see which AI design tool wins. Plus: Buildprint launches for Bubble, Lovable's security issue, and what's new in vibe coding.

Read →
39 min

Conference

The Biggest AI Opportunities for Non-Technical Builders in 2026

A closing panel discussion from Create With featuring leaders from Glide, Softr, Bubble, Momen, and TechFoundHer examining how AI and no-code platforms are democratizing software development. The panelists explore what non-technical builders can now create, how AI is transforming service businesses, and where expertise still matters when prototypes become production applications.

Watch now ↗
29 min

Conference

AI Lets You Build Everything—Here’s Why You Still Never Launch

Tom Wesolowski, owner of NoCodeRefactoring.com, delivers a conference talk exposing how AI's ability to build anything has created a new problem: chronic overbuilding that prevents launches. Using a real marketplace project as a case study, he demonstrates how an 80% scope reduction transformed a stalled four-month development into a revenue-generating product launched in two weeks. The talk provides a practical framework for identifying the minimal viable customer loop and postponing everything else until real user demand validates the need.

Watch now ↗
21 min

Keynote

How a Non-Technical Founder Learned to Build Anything with AI

JJ Englert shares his journey from non-technical filmmaker to AI trainer and repeat founder, demonstrating how no-code tools and AI have democratized software development. Starting with a $50,000 failed build in 2017, he discovered Bubble in 2020 and later embraced AI coding tools to build products, grow communities, and help Fortune 500 companies adopt AI—all without a computer science degree.

Watch now ↗

Latest tool news

What else changed this week.

All tool news
VercelDigest

What Vercel Shipped in Late October

Vercel shipped quicker static deployments, four AI Gateway models, more free domain choices and new CLI controls in late October. The changes cover developers testing sites…

Vercel

Vercel Adds Submit_feedback to Its MCP

submitfeedback is a new Vercel MCP tool that lets agents and harnesses send bug reports, friction reports and feature requests directly to the Vercel team. Feedback can include…

The Create With Briefing

Don't watch forty changelogs. Read one email.

Every Tuesday: the tool changes worth knowing, real business use cases, and what's on near you. Free, unsubscribe any time.