Skip to main content
Legal

AI Transparency Policy

Last updated: March 15, 2026

This AI Transparency Policy ("Policy") describes how Skode Technologies ("Skode," "we," "us," or "our") uses artificial intelligence ("AI") within our CRM and Flow platforms, what data is processed by AI systems, the safeguards we have in place, and your rights regarding AI-powered features. We are committed to responsible, transparent, and ethical use of AI technology.

1. Our Commitment to AI Transparency

Skode believes that users have the right to understand how AI is used in the products they rely on. We are committed to:

  • Transparency: Clearly disclosing where and how AI is used across our platform.
  • Accountability: Maintaining human oversight over AI-driven decisions and providing mechanisms for review and appeal.
  • Data Minimization: Processing only the data necessary for each AI feature to function effectively.
  • No Training on Customer Data: We do not use your data to train AI models. Your data is used solely to provide you with AI-powered features within the Services.
  • User Control: Providing you with the ability to opt out of AI processing features.

2. AI Features in Skode CRM

Skode CRM includes the following AI-powered features:

2.1 Voice AI (Whisper Transcription)

Skode's Voice AI feature allows users to input CRM data using voice commands. Audio recordings are processed by OpenAI's Whisper model for speech-to-text transcription. The transcribed text is then processed by GPT-4o-mini for field extraction and structured data entry with confidence scoring. Audio is processed in real-time and is not stored after transcription is complete. Transcribed text is stored as CRM data within your account.

2.2 38+ AI Analytical Tools

Skode CRM includes over 38 AI-powered analytical tools that provide insights across your sales pipeline, customer data, and business operations. These tools use AI to analyze patterns, generate summaries, and provide actionable recommendations.

2.3 Lead Scoring

AI-powered lead scoring analyzes lead attributes (demographics, firmographics, engagement behavior, source channel) to assign a numerical score indicating the likelihood of conversion. Lead scores are used to help sales teams prioritize their outreach. Lead scoring uses a combination of rule-based logic and machine learning models.

2.4 Deal Predictions

AI-powered deal prediction analyzes deal attributes, historical win/loss patterns, engagement signals, and pipeline stage progression to estimate the probability of a deal closing. Deal predictions are presented as a probability percentage with contributing factors.

2.5 Sentiment Analysis

AI-powered sentiment analysis evaluates the tone and sentiment of customer communications (emails, notes, chat messages) to classify them as positive, neutral, or negative. This helps sales and support teams gauge customer satisfaction and identify at-risk accounts.

2.6 Email Drafting

AI-assisted email drafting generates suggested email content based on context (recipient information, deal stage, previous communications). Generated drafts are always presented to the user for review and editing before sending. No AI-generated content is sent without explicit user approval.

3. AI Features in Skode Flow

Skode Flow (our messaging platform) includes the following AI-powered features:

3.1 AI Chatbot

An AI-powered chatbot that can handle initial customer inquiries, answer frequently asked questions, and route conversations to human agents when needed. The chatbot is trained on your business-specific knowledge base and configured FAQs — not on general internet data or other customers' data.

3.2 Auto-Responses

AI-powered suggested responses for human agents, providing recommended replies based on conversation context. Agents always review and approve or modify suggestions before sending.

3.3 Message Classification

AI-powered message classification automatically categorizes incoming messages by topic, intent, and urgency to assist with routing and prioritization.

4. Data Processing

When you use AI features, the following data may be sent to AI processing systems:

  • Voice AI: Audio recordings (real-time processing, not stored), transcribed text for field extraction.
  • Lead Scoring and Deal Predictions: Lead/deal attributes, engagement metrics, historical conversion data within your account.
  • Sentiment Analysis: Text content of emails, notes, and chat messages selected for analysis.
  • Email Drafting: Recipient information, deal context, and previous communication snippets.
  • Chatbot and Auto-Responses: Conversation messages, knowledge base content, and configured FAQ data.
  • Analytical Tools: Aggregated account data relevant to the specific analysis being performed.

All data sent to AI systems is encrypted in transit (TLS 1.2 or higher). Data is processed for the sole purpose of providing the requested AI feature and is not retained by AI sub-processors beyond the processing session, except where noted.

4.1 Pseudonymisation applied before data leaves Skode

When the AI Connector returns CRM records to ChatGPT or Claude, Skode masks personal identifiers server-side before they leave our infrastructure. The specific rules applied today are:

  • Names (first, middle, last, full):first 3 characters retained, remaining characters replaced with asterisks — e.g. Jonathan → Jon*****.
  • Email addresses:first 2 characters of the local part retained; domain retained — e.g. john.doe@acme.com → jo****@acme.com.
  • Phone and mobile numbers:country code retained; last 2 digits retained; middle digits replaced — e.g. +91-98765-43210 → +91-***-***-10.
  • Street address and postal code: fully replaced with ***.
  • Tax identifiers and custom fields flagged sensitive: first 2 characters retained, remainder replaced with asterisks.
  • Avatar URLs: not included in AI-facing responses.

The following fields are not masked, because they are the business context an AI assistant needs to be useful: status, priority, lead score, lead source, industry, budget, expected close date, timestamps, city, state, country, company, job title, website, deal amount, invoice total, and free-text description fields.

4.2 Current limitations we disclose honestly

We believe a transparency policy should be honest about what it does not yet do. The following limitations are present today and are being addressed on our engineering roadmap:

  • Free-text note bodies are not scanned for embedded personal data.If a user writes “Call John at +91 98765 43210” inside a lead note, that text flows to the AI assistant unmasked. We recommend that teams treat lead notes as shareable with the connected AI provider. Inline redaction of unstructured note content is on our roadmap.
  • Contacts flagged “Do Not Contact” are not yet filtered out of AI queries.Masking protects the specific fields listed in § 4.1, but the lead record itself may still appear in an AI Connector response. Record-level opt-out is on our roadmap; until it ships, treat the AI Connector as operating on your entire workspace regardless of DNC status.
  • AI Connector API keys auto-expire after 90 days.Each workspace user must re-issue a new key every 90 days in Settings → AI; existing keys are revoked automatically.
  • AI activity history is readable via our GDPR Article 15 endpoint but is not yet self-serve deletable. To request deletion of AI activity older than the 30-day auto-purge retention, email privacy@skode.ai.

5. AI Sub-Processors

Skode uses the following third-party AI providers to power AI features. Each is engaged only when a customer or user enables the corresponding feature:

  • OpenAI (San Francisco, USA): Provides Whisper (speech-to-text) and GPT-4o-mini (field extraction, text generation) for in-product AI features, and powers the ChatGPT AI Connector when a customer enables it. OpenAI processes data under a Data Processing Agreement that prohibits the use of customer data for model training. We send the X-OpenAI-Training: disallowheader on every AI Connector response so that content cannot be used for model improvement, and we rely on OpenAI's zero-retention API mode so input and output data is not stored beyond the processing request.
  • Anthropic (San Francisco, USA): Powers the Claude AI Connector through the Model Context Protocol (MCP) when a customer enables it. Anthropic processes data under its Commercial Terms and Data Processing Agreement; customer API inputs and outputs are not used to train Anthropic models.

Planned:Google Gemini is on our AI-provider roadmap but is not in use today. We will publish 30 days’ prior notice on our Sub-Processors page before activating any new AI sub-processor, as described in our Data Processing Agreement.

A complete list of all sub-processors (including non-AI sub-processors) is available on our Sub-Processors page.

6. Human Oversight (GDPR Article 22)

Under GDPR Article 22, individuals have the right not to be subject to a decision based solely on automated processing, including profiling, that produces legal effects or similarly significantly affects them. Skode ensures human oversight of AI features in the following ways:

  • AI-generated lead scores and deal predictions are advisory tools. They do not automatically trigger actions without human review and approval.
  • AI-drafted emails and suggested responses must be reviewed and approved by a human user before sending.
  • AI chatbot conversations can be escalated to a human agent at any time by the customer or automatically based on configured rules.
  • Users may request human review of any AI-generated output or decision by contacting our support team.

7. Automated Decision-Making Disclosure

Skode uses automated decision-making in the following contexts:

7.1 Lead Scoring

Lead scores are generated automatically based on lead attributes and engagement data. These scores influence the order in which leads appear in priority queues and may be used by automated lead assignment rules configured by your administrators. The logic considers factors such as company size, industry match, engagement recency, email open rates, website visit frequency, and form completion. Lead scores are advisory and can be overridden by users at any time.

7.2 Deal Predictions

Deal win probability is calculated automatically based on deal attributes, stage progression speed, historical patterns, and engagement signals. This prediction is displayed alongside the deal and may influence pipeline forecasting reports. Deal predictions do not automatically change deal stages or trigger actions without human intervention.

8. Opt-Out of AI Processing

You have the right to opt out of AI processing features. To opt out:

  • Feature-Level Opt-Out: Individual AI features (Voice AI, lead scoring, deal predictions, sentiment analysis, email drafting) can be disabled in your account settings under Settings, then AI Features.
  • Account-Level Opt-Out: You may disable all AI features for your entire account by contacting support@skode.ai or through your account settings.
  • Data Subject Request: You may submit a formal objection to AI processing through our Data Subject Request process.

Opting out of AI features will disable the corresponding functionality in your account. Core CRM and Flow features that do not rely on AI will continue to function normally.

9. AI-Generated Content Labeling

Content generated or suggested by AI within Skode is labeled to ensure transparency:

  • AI-drafted emails are clearly marked as "AI-generated draft" until edited and approved by a user.
  • AI chatbot responses are identified to the end customer as coming from an automated assistant.
  • AI-suggested replies for agents are visually distinguished from manually composed responses.
  • Lead scores and deal predictions are accompanied by a notation indicating they are AI-generated.

10. Voice AI and BIPA (Illinois)

The Illinois Biometric Information Privacy Act (BIPA) regulates the collection and use of biometric identifiers, which may include voiceprints. Skode's Voice AI feature:

  • Does not create, store, or use voiceprints or any biometric identifiers. The Voice AI feature converts speech to text only.
  • Does not perform speaker identification, voice recognition, or voice authentication.
  • Audio recordings are processed in real-time by OpenAI Whisper and are not stored after transcription is complete.
  • No biometric data is collected, stored, or shared with third parties.

If you have concerns about voice data processing in the context of BIPA or similar state biometric laws, please contact us at privacy@skode.ai.

11. EU AI Act Compliance (August 2026)

The EU AI Act (Regulation (EU) 2024/1689) establishes a risk-based framework for AI systems. Skode is preparing for full compliance with the EU AI Act provisions that take effect in August 2026. Our preparations include:

  • Risk Classification: We have assessed our AI features and classified them under the EU AI Act risk categories. Our CRM AI features (lead scoring, deal predictions, sentiment analysis, email drafting) are classified as limited-risk AI systems, subject to transparency obligations.
  • Transparency Obligations: This Policy and our in-product labeling fulfill the transparency requirements for limited-risk AI systems under Article 50 of the EU AI Act, including notifying users when they interact with AI systems.
  • AI Literacy (Article 4): We provide documentation and guidance to help users understand the capabilities and limitations of our AI features.
  • Record-Keeping: We maintain records of AI system design, data processing, and performance as required by the Act.

12. Colorado AI Act (June 2026)

The Colorado AI Act (SB 24-205) requires developers and deployers of high-risk AI systems to provide transparency and manage risks of algorithmic discrimination. Skode's compliance includes:

  • Providing this transparency notice describing our AI systems, their purposes, and the types of data processed.
  • Implementing risk management practices to identify and mitigate potential algorithmic bias in lead scoring and deal prediction models.
  • Providing users with the ability to opt out of AI processing and to appeal AI-influenced decisions.
  • Regularly reviewing and testing AI models for fairness and accuracy.

13. India IT Rules 2026 (Chatbot Disclosure)

Under the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, as amended, and the Digital Personal Data Protection Act, 2023, Skode complies with disclosure requirements for AI-powered chatbots:

  • The Skode Flow AI chatbot clearly identifies itself as an automated system in its initial message to every conversation.
  • Customers interacting with the chatbot are informed that they may request to speak with a human agent at any time.
  • Data processed by the chatbot is handled in accordance with our Privacy Policy and applicable Indian data protection laws.

14. Profiling Opt-Out Mechanism

Skode performs profiling in the context of lead scoring and deal predictions. You may opt out of profiling at any time:

  • Within the Platform:Navigate to Settings, AI Features, and disable "Lead Scoring" and/or "Deal Predictions."
  • Via Email: Send a request to privacy@skode.aiwith the subject line "Opt-Out of AI Profiling."
  • Via Data Request: Submit a formal objection through our Data Subject Request page.

When you opt out of profiling, lead scores and deal predictions will no longer be generated for your account. Historical scores and predictions will be retained in your data but will not be updated. You may request deletion of historical profiling data through a data subject request.

15. Contact Information

If you have questions about this Policy, our AI practices, or wish to exercise your rights regarding AI processing, please contact us: