To get the most accurate answers from Caplena’s Insight Agent—especially when querying your data via Slack, Microsoft Teams, or ChatGPT, the AI needs to understand the blueprint of your project.

We’ve rolled out a few structural updates to reduce AI guesswork and make the Isight Agent even more powerful:

• 10x More Project Context: We’ve increased the character limit for your Project Context from 500 to 5,000 characters. You now have the space to thoroughly brief the AI on the specific nuances, background, and goals of your dataset.
• Default Date & Score Columns: You can now explicitly define which columns represent your primary timeline (e.g., Review Date) and primary metric (e.g., Rating). By setting these defaults once, the AI immediately knows how to accurately track trends and weigh sentiment without needing extra prompting.

Where to find it:
You can update these anytime in the Data View, or simply confirm them when setting your project to Live to connect with everyday tools like Slack, Microsoft Teams, or ChatGPT.

Sometimes you need the blueprint of your project, not just the raw feedback data. We’ve added a new export option that allows you to download the structural overview of your Caplena projects directly to Excel.

This is especially useful for maintaining internal documentation, auditing your data architecture, or sharing project setups with stakeholders who don't need access to the granular text responses.

What’s included in the Excel export?

• Project Overview: High-level details and settings of your current project.
• Data Sources: A clear log of where your feedback is originating from.
• Columns: The exact structure, naming conventions, and metadata of your dataset.

How to use it:

1. Navigate to the Data view in your project.
2. Click the Export icon (or trigger an export for your filtered rows).
3. In the export modal, select the new Project Metadata tab.
4. Hit Export to download your file.

We’ve separated the global Report Master from your custom Views for a cleaner, more intuitive workflow. Plus, sharing just got a major upgrade. Let's dive in:

• 🧱 Edit Report Master: Think of it as your global foundation: Manage insight elements & sections per drag & drop, create segments and apply filters. The settings will automatically cascade down to all of your Views.
• 🔎 Configure Views: Select segments, apply filters, and lock them exactly as you are used to. Keep the Views sidebar perfectly organized by easily dragging & dropping your Views into any order.
• 🚀 Share: You can now instantly share a single View, a custom selection of Views, or the entire Report.

Powerful for Experts, Effortless for Everyone — now where work happens. Insight Agent is now available in Slack, Microsoft Teams, Copilot, and ChatGPT (MCP), so anyone can ask questions and get reliable insights without leaving their daily tools.

This is a big step toward democratizing insights across your organization: experts move faster with on-demand analysis, while non-research teams self-serve clear, actionable answers — no dashboards, training, or back-and-forth required. Less chasing data. More turning feedback into decisions. 💪

How it works

• Create a shared channel (e.g., #customer-feedback-insights)
• Ask questions by tagging the agent with @Caplena in 1:1 messages or public channels
• Here is an example question: @Caplena What topics had the biggest impact on the NPS change in the last 4 weeks?

Advantages

• Help teams discover what the bot can do (and copy great prompts)
• Learn from each other’s use cases over time
• Discuss insights together — and jump in if you disagree
• Keep decisions and learnings accessible for everyone

👉 Set up the integration in your Integrations Settings and get started.

We’ve launched a new integration with Trusted Shops, one of Europe’s leading trust and review platforms for online shops. You can now automatically import and analyze your Trusted Shops customer reviews in Caplena. All verified reviews are processed with AI-powered topic and sentiment analysis, transforming raw feedback into structured, actionable insights within minutes.

👉 Set up the integration in your Integrations Settings and get started.

We’ve introduced a new permission that gives you granular control over credit deduction. Admins can now define which roles are allowed to trigger actions that consume credits — adding an extra layer of governance and cost control across your workspace.

This ensures that powerful AI capabilities remain accessible, while maintaining full transparency and budget oversight.

We’re excited to introduce our new native integration with Decipher / Forsta — making it easier than ever to bring open-ended survey data into Caplena and turn it into structured, client-ready insights.

If you’re running studies in Decipher, you can now seamlessly connect your project and start analyzing verbatims in minutes — no manual exports, no formatting headaches.

👉 Set up the integration in your Integrations Settings and get started.

We’ve rolled out a major upgrade to AI Advanced Output (Phrase Extraction) across all projects — powered by our new LLM.

It’s now enabled by default and no longer comes with additional cost! 👏

Why this matters? 
Advanced AI output adds reasoning and phrases so you can see which parts of the text are relevant for each topic.

We’ve added a brand-new Caplena Broad Color Theme to give your reports more visual flexibility and a richer color palette.

The new theme expands the range of topic and chart colors, making it easier to:

• Clearly distinguish categories and segments
• Create more expressive, client-ready reports
• Maintain visual clarity even in complex analyses

You can manage and assign the Caplena Broad Color Theme from Settings → Themes, just like any other theme. Once applied to a project, it will automatically reflect across all associated reports for report viewers.

The response to our Insight Agent has been incredible! This surge in usage recently pushed our LLM providers to their limits, occasionally leading to resource constraint errors or timeouts.

We’ve been working around the clock to build a more resilient and contineously improving system for you. We are excited to announce a major infrastructure upgrade: advanced multi-region datacenter load balancing (that’s just fancy dev-speak for: "the Agent is now much harder to break, even when it’s busy" 🤓).

• What this means: The Insight Agent now intelligently routes requests to the most available datacenter, bypassing provider bottlenecks and ensuring stable performance even during peak hours.
• Performance: You'll notice significantly fewer errors and smoother analysis for large datasets.
• Compliance: Most importantly, our EU-only data residency remains unchanged. All data processing and storage continue to stay strictly within the European Union.
• Continuous Learning: A quick reminder that your Thumbs Up/Down ratings help the Agent refine its logic to better suit your specific tasks, project goals, and needs. We appreciate the feedback!

We’ve heard your feedback: analyzing a deep-dive customer review requires a different "brain" than analyzing a list of 500 slightly misspelled brand names. To give you the best of both worlds, we are introducing Semi-Open Mode, a specialized algorithm designed for precision in short-text and brand-awareness data.

🧠 Choosing the Right Tool for the Task

Not all text data is created equal. Depending on your source—whether it's an open-ended "Tell us more" or an unaided brand recall question—you now have the power to choose the AI logic that fits best.

1. Open Mode: Deep Semantic Understanding

• Best for: Complex feedback, full sentences, and emotional nuances.
• How it works: This mode looks at the meaning behind the words. It understands negations ("not happy"), synonyms, and the underlying sentiment of multiple sentences.
• Use case: "The check-in was slow, but the staff was incredibly friendly and made up for it."

2. Semi-Open Mode: Precision for Short Text

• Best for: Unaided brand awareness, single words, or short word groups with many duplicates.
• How it works: This mode focuses on spelling similarity and clustering. It is optimized to catch typos and group variations of the same brand or product name without over-analyzing for "emotion."
• Use case: "Nike," "Niky," "Nike Inc," or "Adidas."

🛠️ How it Works in Your Workflow

We’ve integrated this choice directly into the Topic Assistant setup. In many cases, Caplena will even auto-detect if your data looks like a brand list and suggest the Semi-Open algorithm to save you time.

At Caplena, our goal has always been to turn the "noise" of open-ended text into clear, actionable insights. To keep pushing the boundaries of what AI can do for your feedback analysis, we are focusing all our energy on Caplena v3.

As part of this transition, we will be officially switching off the Caplena Legacy App on March 31, 2026. To ensure your workflows remain uninterrupted and to take advantage of our most powerful features yet, we encourage all users to start using v3 over the coming weeks.

Why the Move?

Caplena v3 isn't just a fresh coat of paint; it’s a completely re-engineered core designed for the era of Large Language Models (LLMs). Here is what you get in the new platform:

• Integrated LLM Power: Out-of-the-box human-level coding quality and your personal AI agent to explain the "why" behind your data.
• Smart Columns: A powerful new way to categorize and map data using custom logic and AI, giving you more flexibility than ever before.
• Advanced Reporting & Views: Create multiple views with unique sets of filters and segments—one for each stakeholder, region, or product team.
• Workflow Integrations: Connect seamlessly with Qualtrics, Medallia, Slack, Microsoft Teams, and 20+ other integrations.
• And much more! 🚀

How to Transition

All your projects are already in Caplena v3—no migration needed, and your historical data and custom codebooks are fully preserved. Just follow the steps below:

1. Go to the Caplena v3 project home
2. Choose your project(s)
3. Create new reports with one click to visualize results and ask questions using the Insights Agent.

Need a hand with the move? Our support team is standing by to help with your migration.