Eddy AI

Master this essential documentation concept

Quick Definition

Document360's proprietary AI suite that powers content creation, multi-language translation, and knowledge base assistance within the platform.

How Eddy AI Works

graph TD A[Document360 Platform] --> B[Eddy AI Suite] B --> C[AI Writer] B --> D[AI Translator] B --> E[Eddy Assistant] C --> C1[Draft Articles from Prompts] C --> C2[Rewrite & Summarize Content] D --> D1[Multi-Language KB Translation] D --> D2[Locale-Specific Publishing] E --> E1[Reader Q&A Chatbot] E --> E2[Contextual Article Suggestions] C1 --> F[Published Knowledge Base] C2 --> F D2 --> F E1 --> G[End User Self-Service] E2 --> G

Understanding Eddy AI

Document360's proprietary AI suite that powers content creation, multi-language translation, and knowledge base assistance within the platform.

Key Features

  • Centralized information management
  • Improved documentation workflows
  • Better team collaboration
  • Enhanced user experience

Benefits for Documentation Teams

  • Reduces repetitive documentation tasks
  • Improves content consistency
  • Enables better content reuse
  • Streamlines review processes

Making Eddy AI Knowledge Searchable Across Your Team

When teams onboard to Document360, walkthroughs of Eddy AI's capabilities — how to trigger content suggestions, configure translation workflows, or use the knowledge base assistant — are often recorded as screen-capture videos or live training sessions. These recordings capture useful context, but they create a practical problem: when a writer needs to remember how to invoke Eddy AI for a specific language output at 2pm on a Tuesday, scrubbing through a 45-minute onboarding video is rarely a realistic option.

The challenge compounds as your team grows. New documentation professionals joining mid-project have no way to quickly locate the segment explaining how Eddy AI handles terminology consistency across translated articles, or which content types support AI-assisted drafting. Knowledge stays locked inside timestamps rather than becoming part of your team's shared reference layer.

Converting those training recordings into structured documentation changes the dynamic entirely. A video demonstrating Eddy AI's multi-language translation workflow becomes a searchable, linkable article your team can reference in context — right when they're inside the editor and need a quick answer. You can surface specific Eddy AI behaviors, flag known limitations, and keep the documentation updated as the feature evolves, without scheduling another recording session.

If your team relies on video to document internal tool workflows like this, there's a more sustainable approach worth exploring.

Real-World Documentation Use Cases

Rapidly Documenting a New SaaS Feature Release Across 6 Languages

Problem

A product team ships a new dashboard feature and needs documentation published in English, French, German, Spanish, Japanese, and Portuguese within 48 hours of release. Technical writers lack the bandwidth to manually translate and localize six separate articles without introducing inconsistencies or delays.

Solution

Eddy AI's AI Writer generates the initial English article from a structured prompt describing the feature, and Eddy AI Translator then produces all six localized versions simultaneously, preserving terminology consistency across every language variant.

Implementation

['Use Eddy AI Writer in Document360 to draft the English feature article by entering a prompt that includes the feature name, key behaviors, and user actions required.', 'Review and approve the AI-generated draft, making targeted edits to align with brand tone and technical accuracy before saving it as the source article.', 'Trigger Eddy AI Translator on the approved English article, selecting all six target locales from the translation settings panel in Document360.', "Publish all translated versions simultaneously using Document360's version control, then notify locale-specific user segments via the platform's notification settings."]

Expected Outcome

All six language versions of the feature documentation are published within 6 hours of the English draft approval, reducing translation turnaround from 2 weeks to under a day and ensuring terminology consistency across every locale.

Deflecting Tier-1 Support Tickets for a Complex API Integration Product

Problem

A developer tools company receives over 400 monthly support tickets asking how to authenticate with their REST API, configure webhooks, and handle rate limit errors—questions already answered in their knowledge base. Support agents spend 60% of their time answering repetitive queries instead of resolving complex issues.

Solution

Eddy Assistant is embedded in the developer portal as a conversational chatbot that interprets natural-language questions, surfaces the exact API authentication article, webhook setup guide, or error-handling section, and provides a direct answer with a link to the full documentation.

Implementation

["Enable Eddy Assistant in Document360 and embed the chatbot widget on the developer portal's API reference and getting-started pages using the provided JavaScript snippet.", "Train Eddy Assistant's context scope to prioritize the Authentication, Webhooks, and Error Codes categories within the knowledge base to improve answer precision for developer queries.", "Configure fallback behavior so that when Eddy Assistant cannot find a confident answer, it presents a pre-filled support ticket form with the user's question already captured.", "Monitor Eddy Assistant's unanswered question logs weekly in Document360 Analytics to identify documentation gaps and prompt Eddy AI Writer to generate missing articles."]

Expected Outcome

Tier-1 support ticket volume drops by 38% within 90 days, with Eddy Assistant successfully resolving authentication and webhook questions in under 30 seconds, freeing support engineers to focus on complex integration failures.

Standardizing Inconsistent Legacy Documentation Migrated from Confluence

Problem

After migrating 800 articles from Confluence to Document360, a technical writing team discovers that article structures, tone, heading hierarchies, and terminology vary wildly across content authored by different teams over five years. Manually rewriting each article to meet the new style guide would take months.

Solution

Eddy AI Writer's Rewrite and Summarize capabilities allow writers to select legacy articles and apply a consistent rewrite prompt that enforces the company's approved tone, heading structure, and terminology glossary, transforming inconsistent content at scale.

Implementation

["Define a rewrite prompt template in Eddy AI Writer that specifies the target tone (e.g., 'concise, active voice, second-person'), required heading levels (H2 for major sections, H3 for sub-steps), and key terminology substitutions (e.g., replace 'end-user' with 'customer').", "Batch-select the highest-traffic legacy articles first using Document360's analytics to prioritize the 20% of content that drives 80% of reader visits, and apply the Eddy AI rewrite prompt to each.", 'Have technical writers perform a focused review of AI-rewritten articles, checking only for factual accuracy and edge-case terminology rather than restructuring the entire piece.', "Use Document360's comparison view to diff the original and AI-rewritten versions, approving changes and archiving the legacy drafts to maintain a clean revision history."]

Expected Outcome

The team standardizes 300 high-traffic articles in 3 weeks instead of the projected 4 months, achieving consistent heading structure and tone across the migrated content while reducing writer review time per article from 2 hours to 25 minutes.

Enabling Non-Native English-Speaking SMEs to Contribute Accurate Technical Documentation

Problem

A global software company relies on subject matter experts in Germany, Brazil, and South Korea to document internal engineering processes, but these SMEs write in their native languages. The content must be published in English for a global audience, and the current manual translation workflow through a localization agency takes 3 weeks per article and costs $800 per document.

Solution

Eddy AI Translator allows SMEs to submit articles in German, Portuguese, or Korean directly in Document360, with Eddy AI producing an English translation that technical writers then review and refine—eliminating the localization agency dependency for internal documentation.

Implementation

['Set up locale-specific draft categories in Document360 where German, Brazilian Portuguese, and Korean SMEs submit their articles in their native language without requiring English proficiency.', "Apply Eddy AI Translator to each submitted draft, selecting English as the target language, and route the translated output to the technical writing team's review queue in Document360's workflow.", 'Technical writers perform a focused accuracy and fluency review of the English translation, using the side-by-side source and translated view in Document360 to flag any mistranslated technical terms.', 'Publish the approved English article and optionally use Eddy AI Translator to back-translate it into additional locales needed for the global knowledge base.']

Expected Outcome

Article translation turnaround drops from 3 weeks to 48 hours, per-document translation costs fall from $800 to near zero for internal documentation, and SME contribution volume increases by 55% because experts can now write in their native language without bottlenecks.

Best Practices

âś“ Write Specific, Context-Rich Prompts for Eddy AI Writer to Minimize Revision Cycles

Eddy AI Writer produces significantly more accurate and on-brand first drafts when prompts include the article's purpose, target audience, key concepts to cover, and any terminology constraints. Vague prompts force the AI to make assumptions that require extensive human correction, negating the time savings. Investing 5 minutes in a detailed prompt typically reduces revision time by 60% compared to a one-line prompt.

âś“ Do: Include the product feature name, the reader's technical level (e.g., 'for a developer familiar with REST APIs but new to OAuth 2.0'), the three main points the article must cover, and any terms from your glossary that must be used.
✗ Don't: Do not submit prompts like 'write an article about authentication' without context—this produces generic content that requires complete rewriting and defeats the purpose of using Eddy AI Writer.

âś“ Review Eddy AI Translator Output Against Your Approved Terminology Glossary Before Publishing

Eddy AI Translator excels at fluent, grammatically correct translations but may not automatically apply your product's proprietary terminology, brand names, or locale-specific regulatory terms. Publishing AI-translated content without a glossary check risks confusing readers with incorrect product names or non-compliant phrasing in regulated markets. Maintaining a Document360 glossary that writers reference during translation review ensures consistency across all languages.

âś“ Do: Create a terminology glossary in Document360 for each target language listing approved translations of product names, UI labels, and technical terms, and use it as a checklist when reviewing Eddy AI Translator output.
âś— Don't: Do not publish Eddy AI Translator output directly without review for articles covering legal terms, compliance requirements, or product feature names that have locale-specific approved translations.

âś“ Scope Eddy Assistant to Relevant Knowledge Base Categories to Improve Answer Precision

By default, Eddy Assistant searches across the entire knowledge base, which can surface tangentially related articles instead of the most relevant answer for a specific user context. Scoping Eddy Assistant to the categories most relevant to where it is embedded—such as limiting it to API Reference articles on a developer portal page—dramatically increases answer accuracy and reduces user frustration. Category scoping is configurable in Document360's Eddy Assistant settings without requiring engineering changes.

âś“ Do: Configure Eddy Assistant's category scope to match the documentation section where the chatbot widget is embedded, so a user on the Billing FAQ page receives answers only from billing-related articles.
âś— Don't: Do not leave Eddy Assistant with unrestricted knowledge base access on highly specialized pages, as it will return loosely related results from unrelated categories and erode user trust in the chatbot.

âś“ Use Eddy Assistant's Unanswered Questions Log as a Weekly Content Gap Audit

Every question that Eddy Assistant cannot confidently answer represents a gap in your knowledge base that is actively frustrating real users. Document360 logs all unanswered or low-confidence Eddy Assistant queries, giving technical writers a prioritized, data-driven list of articles to create or expand. Treating this log as a weekly editorial input rather than a support metric transforms Eddy Assistant from a reactive tool into a proactive content strategy driver.

âś“ Do: Schedule a weekly 30-minute review of Eddy Assistant's unanswered questions log in Document360 Analytics, group similar queries by topic, and use Eddy AI Writer to draft articles addressing the top five recurring gaps.
✗ Don't: Do not ignore the unanswered questions log or treat it solely as a chatbot performance metric—each unanswered query is a documented instance of a user failing to find help, which directly increases support ticket volume.

âś“ Preserve Human Editorial Review as the Final Gate for All Eddy AI-Generated Content

Eddy AI Writer and Eddy AI Translator are force multipliers for documentation teams, not replacements for subject matter expertise and editorial judgment. AI-generated content can contain plausible-sounding but factually incorrect technical details, especially for niche product behaviors or recently released features that post-date the model's training context. Establishing a mandatory human review step in Document360's workflow before any AI-generated article is published protects documentation accuracy and brand credibility.

âś“ Do: Configure Document360's article workflow to require at least one technical writer or SME approval step for all articles tagged as AI-generated, and use the platform's review assignment feature to route drafts to the appropriate domain expert.
✗ Don't: Do not enable auto-publish for Eddy AI Writer drafts, even for lower-stakes content like FAQ updates—a single factually incorrect AI-generated article published without review can undermine reader trust in the entire knowledge base.

How Docsie Helps with Eddy AI

Build Better Documentation with Docsie

Join thousands of teams creating outstanding documentation

Start Free Trial