LoginSign Up
← Back to all sparks
T

Tiledesk

SUPPORT
Velocity0.0

Open-source conversational AI platform for customer support and lead generation

www.tiledesk.com ↗

Tiledesk's editorial is now 100% agentic AI and MCP — the platform pivot is the story

agentic-aimcp-protocolself-learningrag-architectureopen-source-supportcustomer-service
◆Current state
Nearly every recent post is about agentic AI, MCP-driven actions, self-learning resolutions, and RAG architecture. The cadence is light and content is mostly snippet-level, but the topical concentration is unmistakable: Tiledesk is repositioning from a chatbot platform to an agentic support runtime.
◆Where it's heading
Architectural disclosures across the last year — hybrid RAG engine in summer, self-learning controls in late 2025, MCP playbooks through Q1 — trace a multi-quarter buildup of the agentic capability stack rather than one big launch. The MCP framing is consistent enough to suggest first-class protocol support, not just content marketing.
◆Prediction
Likely next moves are a packaged MCP toolkit or template library, plus self-learning observability (what the agent learned, what humans corrected). Given the MCP repetition, an exposed MCP server or marketplace listing for Tiledesk-built agents is plausible.

◆Recent moves

  1. 1mo ago

    Build AI Agents That Take Action Using MCP: 5 Practical Business Use Cases

    MCP business use-case roundup. Demonstrates Tiledesk's commitment to the protocol but is editorial framing rather than a release; supports the agentic-pivot narrative.

    View source ↗
  2. 1mo ago

    From Answers to Outcomes: How AI Agents Reason, Plan, and Act

    Conceptual piece on reasoning, planning, and autonomous actions in AI agents. Sets the vocabulary Tiledesk is selling against; no product disclosure.

    View source ↗
  3. 2mo ago

    How a Self-Learning AI Agent Turns Human Resolutions Into Better Support

    Explanation of how the self-learning agent improves from human resolutions. Implies the capability is shipped, but the post is positioning rather than a release announcement.

    View source ↗
  4. 4mo ago

    How to Build an AI Agent for Lead Qualification with MCP

    How-to guide on building an MCP-based lead-qualification agent. Reinforces the protocol-first framing and pulls a sales-adjacent use case into the agentic-support narrative.

    View source ↗
  5. 5mo ago

    Control what your self-learning AI adds to your knowledge base

    Knowledge-base governance content — how to control what the self-learning agent ingests. Hints at admin-control features behind the agentic surface without describing the change itself.

    View source ↗
  6. 9mo ago

    Tiledesk Hybrid Search RAG Architecture

    Architectural explainer on Tiledesk's hybrid-search RAG approach. Best evidence in the window that there's real engineering substance behind the agent storytelling, even if framed as a blog rather than a release.

    View source ↗