What's New

Run structured safety walkthroughs from start to finish. Build reusable inspection templates with categories, complete inspections in the field, and attach observations to specific categories as you go. Admins can create and publish templates; field teams can run inspections and close them out when done.

When you add an observation during an inspection, the AI automatically places it in the right category and can suggest inspection-specific guidance and tips as you work.

Connect your Procore account to keep locations and work units in sync. Observations can reference the right project and site automatically, and admins can see when each connection last synced.

Organize observations by the specific crew, project, or work package they relate to. Assign a work unit when creating an observation and filter the list by it later.

AI responses now show document citations with sequential numbering and friendlier document names. The overall summary no longer contains raw reference tokens, making responses easier to read.

The observation detail page now combines location and work unit into a single, tidier section so you can scan the context of an observation at a glance.

In-app announcements can now be targeted to specific organizations, and dismissing one only hides it for you — not for everyone else in your company.

When something goes wrong, the app now surfaces the specific application error (like "observation not found" or "not authorized") instead of a generic message, while still hiding internal details.

Record a voice note while creating an observation and Sophie will transcribe it, extract the key fields, and fill in the form for you. A new status banner walks you through the transcription and analysis stages.

Control how the app behaves for you in Settings, including whether voice recordings should auto-populate observation fields. Your preferences sync across devices.

Observation video and audio processing now runs asynchronously on the server, so you can keep working while Sophie analyzes your recordings in the background.

Generate safety posters, signage, and visuals in widescreen aspect ratios, better suited for slide decks and large-format displays.

When you ask Sophie for toolbox talk material, she now proactively offers to generate visual infographics alongside the written content to make your briefings more engaging.

Document citations in chat responses render more consistently and reliably, with improved readability and accurate links back to the source document.

When Sophie identifies hazards, she now keeps recommendations to the most impactful ones instead of dumping a long list — making it easier to actually act on what she suggests.

AI image generation now automatically retries when a provider has a transient hiccup, so transient network blips no longer cause failed generations.

A new quick action card helps you role-play and prepare for difficult safety conversations with workers, supervisors, or clients.

The hazard analysis results inside chat are now easier to scan, with cleaner layout and clearer risk indicators.

PDF reports use a consistent, professionally-tuned font stack across every section, producing cleaner and more branded-looking documents.

Database transaction handling across the platform has been tightened, reducing the chance of stale reads and improving stability during high-traffic periods.

When scheduled maintenance is coming up, a banner in the app gives you advance warning and quietly disappears once the window is over.

The admin CSV import and organization logo dialogs now actually accept drag-and-drop, matching the behavior you already get for document and image uploads elsewhere in the app.

Resolved an issue where the description field could not be edited on some observations. You can now always update observation descriptions, regardless of source.

The Submit and Edit buttons on observation pages no longer shift around as the status updates, preventing accidental clicks during processing.

When you delete an observation, it now vanishes from the list right away instead of occasionally sticking around until the next refresh.

When the analytics agent returns a chart, it now shows the written answer alongside the visualization so you get the full explanation in one place.

Clicking a tooltip or popover inside a dialog no longer accidentally dismisses the whole dialog — a small fix that makes complex forms much less frustrating.

The AI can now generate and share ZIP workbooks and RTF documents directly in chat, alongside the existing PDF and spreadsheet support.

After each AI response, suggested follow-up actions appear as clickable buttons. Tap a suggestion to quickly continue the conversation without typing.

Download images shared in chat as branded PDF documents. Optionally include a signature page for formal documentation and sign-offs.

See whether safety controls were effective before and after an observation. Green indicators highlight energy sources with controls that were corrected on site.

Names of at-fault workers in negative observations are automatically replaced with role identifiers like "Worker" or "Operator" to protect privacy.

Document references now appear as clickable links directly within the AI response text. On iPad, the cited document opens in a side panel alongside the chat.

Positive safety observations without action items are now automatically closed after processing, saving you a manual step.

New analytics chart shows observation counts grouped by creator, with an open vs. closed breakdown.

Upload photos taken on iPhone in their native HEIF/HEIC format without converting them first.

Photos now display with correct orientation regardless of how they were taken, fixing issues where rotated images appeared incorrectly.

Attach up to 5 photos when closing an observation to document corrective actions. A green checkmark badge distinguishes closure photos from original images.

Editing action items is now smoother - the app intelligently matches and updates existing items instead of replacing everything, preventing lost work.

Long-press the submit button to save the current observation and immediately start a new one, speeding up data entry when logging multiple observations.

Pull down on the observation list to refresh and see the latest updates.

Rate AI responses with thumbs up or thumbs down and optionally add comments to help improve the assistant.

The AI can analyze spreadsheets and CSV files you upload, create data visualizations, and generate downloadable custom reports as PDF.

New analytics view showing insights about AI conversations: sentiment trends, resolution status, high-risk alerts, frequent topics, and top locations mentioned.

See real-time status when the AI generates images, with animated progress indicator and automatic display when complete. View, zoom, and share generated images directly from chat.

When the AI cites a custom uploaded document, click the reference to view the original file directly within the app.

The chat home screen displays visual cards for common tasks like analyzing photos, asking OSHA questions, creating observations, and generating safety posters.

When the AI references a document, click the reference badge to view the source PDF directly within the app at the specific page being referenced.

Administrators can upload company safety manuals, procedures, and PDFs. AI agents can access these documents to provide answers grounded in your company policies.

Specify when an observation occurred using a new date and time picker. The AI also picks up dates and times mentioned in your description automatically.

The AI assistant can help create draft observations directly from your conversation, automatically including discussed details, location, hazards, and images.

Drag and drop images directly onto the chat page, new observation screen, or attachments area. A visual overlay shows when you're in the drop zone.

Dropdown menus display as full-screen sheets on mobile devices for easier selection.

Assign multiple team members to a single observation. Everyone assigned receives email notifications.

When creating observations via voice or text, say "assign to myself" and the system automatically assigns it to you.

Chat sessions automatically get a descriptive title based on image content analyzed (like "Warehouse Loading Dock" or "Rooftop HVAC Work").

If image analysis fails or produces incomplete results, retry directly from the chat interface without re-uploading.

When you upload a workplace photo, the AI marks hazards with numbered annotations directly on the image, showing risk levels and locations at a glance.

Upload several photos at once and the AI will analyze them together, understanding shared context and deduplicating hazards across images.

Your observation form edits are automatically saved as you type, preventing lost work if you navigate away.

New analytics chart shows pending and completed action items grouped by assignee, making it easy to track workload distribution.

A confirmation dialog appears when editing closed observations to prevent accidental changes to completed records.

Clear notification when you reach the 4 image upload limit, preventing confusion about why additional images cannot be added.

Safety posters and AI-generated images now include your organization's logo by default.

Generate custom safety signage, training materials, and visual content directly through conversation with the AI. Create 4-10 variations in a single request with automatic organization branding. Edit existing images with AI guidance for quick modifications.

Upload images during conversations and get instant AI-powered analysis. The system can identify OSHA violations, company policy breaches, and safety hazards in workplace photos. Analyze equipment compliance and get detailed observations with regulatory references.

When starting a new chat, see suggested questions to help you discover what each AI agent can help with. Click any suggestion to quickly begin a conversation without typing.

A prominent banner highlights AI-suggested action items for observations. Review and approve suggested actions directly from observation cards and detail pages with one-click acceptance.

AI now handles unclear or ambiguous inputs during hazard identification by asking clarifying questions. Improved error handling prevents workflow failures and provides better guidance throughout the process.

Resolved issue where action items appeared multiple times in generated PDF reports. Reports now display each item exactly once for cleaner documentation.

AI agents now display their associated phone numbers in the chat header and sidebar. Easily reference contact information for voice-based interactions with safety assistants.

User selection dropdowns now support keyboard navigation with automatic search focus. Navigate and select users faster when assigning observations or action items.

The analytics agent now provides direct links to dashboards and saved custom charts. Navigate to specific analytics views faster without manual searching.

Control whether AI-generated enrichment is included when generating observation reports. Choose between quick basic reports or comprehensive enhanced documentation based on your needs.

Updated sentiment analysis with more nuanced categories and definitions. Better insight into user satisfaction and conversation quality across support interactions.

Multi-select dropdowns now include cancel and clear options. More easily manage selections with the ability to undo changes or clear all at once.

Improved validation ensures only active organization members can be assigned to observations. Better search functionality when filtering by assignee with support for job roles.

Conversations are automatically enriched with intelligent tags including topics, sentiment, risk level, and resolution status. Filter and analyze conversation history to identify trends and urgent issues.

Administrators can view analytics across their entire organization with six new pre-built charts: Conversation Volume Trends, Frequently Asked Topics, Urgent Conversations, Sentiment by Topic, Resolution Status Breakdown, and Topics by Resolution Rate.

Upload and manage your company logo for consistent branding across generated reports and documentation. Logos appear in PDF reports, organization picker, and navigation elements.

Hazard identification now suggests stronger controls beyond basic recommendations, with status tracking for each suggestion. Visual icons distinguish control types (elimination, substitution, engineering, administrative, PPE).

New analytics view on iOS provides visual insights including average close time, observation statistics, risk level distribution, top energy sources, and OSHA violations. Filter by custom date ranges.

Generate and download PDF reports for individual observations or batches directly from your iPhone. Customize report titles and manage reports through a new status tracking view.

Eliminated UI freezes when loading multiple cached images. Images now load asynchronously in the background for smooth progressive loading. Improved app launch with pre-loaded database and file cache.

New panel displays AI reasoning and technical details for chat messages. See how the assistant arrived at its responses for better understanding of safety recommendations.

Reports automatically filter out irrelevant safety regulations that don't match detected energy sources. Cleaner documentation with modern Phosphor icons for better readability.

Never lose your work when documenting safety observations. Save incomplete observations as drafts when exiting, with automatic sync when back online.

New filtering options include date range selection, minimum days open, and immediate attention flags. Active filters display as removable chips for easy refinement of observation lists.

New menu on observation pages provides quick access to PDF generation and deletion. Common actions consolidated in a convenient dropdown for faster workflows.

Select multiple observations and generate comprehensive PDF reports with executive summaries, detailed findings, and risk breakdowns. Single reports download instantly while batch reports (up to 1,000 observations) process in the background with email notification. Track report status with real-time updates.

Control effectiveness now displays as clear categories (Missing, Weak, Acceptable, Strong) instead of percentages. This makes it easier to quickly assess risk controls at a glance and aligns with how safety professionals think about control strength.

Observation filters now collapse on mobile devices for a cleaner interface. Open filters when you need them, then hide them to see more observations on smaller screens.

New FAQ answers 13 common questions about Stepo capabilities, implementation, security, and ROI. Includes clear comparisons with ChatGPT and GitHub Copilot, explaining what makes Sophie different for safety.

Added mission section to homepage explaining Stepo's vision and commitment to workplace safety. Features elegant design with background imagery and clear messaging about our purpose.

Enhanced validation when submitting observations ensures all required fields are properly filled. Better error messages guide you through the submission process with clearer feedback.

Get answers to complex data questions in half the time. Our analytics agent now handles intricate queries more efficiently, cutting average response time in half.

Quickly locate the exact observations you need with new filtering options. Filter by assignee, creator, or unassigned status. Your filter preferences are automatically saved.

Find OSHA answers tailored to your role. Filter safety standards by job type and discover the most popular questions other teams are asking.

Pages load faster and work better on your phone. We optimized images and streamlined the site for instant access to safety information.

Sophie is now faster and more capable. She handles conversation pauses naturally, responds quicker, and understands multiple languages better.