Welcome to the Kairntech low-code AI platform product roadmap
— last updated April 29, 2026.
If you have a question, a project in mind, or want to learn more about our platform, please contact us !
Caption for the table on the effort required for development
S
some days
M
some weeks
L
some months
XL
one semester or more
| ID | TOPIC | SUB-TOPIC | EFFORT | DELIVERY |
|---|---|---|---|---|
| 1 | Ingestion pipeline | Make content ingestion pipeline more robust: New event-driven architecture (RabbitMQ…), parallel processing, handling errors, progress bar, retry, recovery, failed data management, store original document… | XL | New Web Service |
| 2 | Connectors | Sharepoint connector v2 (call ingest API + ability to configure in the Studio) | M | Connector & Studio |
| 3 | Veeva connector | ? | Connector & Studio | |
| 4 | Kairntech Studio | Display segments in the document view | M | Improved Studio |
| 5 | Display original documents (PDF viewer, images, video…) | S | Improved Studio | |
| 6 | New features for creating RAG dataset: Question Generator, Synthetic Answer Verifier, Cross-Chunk Contradiction Detector | L | Improved Studio | |
| 7 | New features for creating S2T dataset : audio file playback, text document editing, automatic spell checking, report of the time position in the audio file | ? | Improved Studio | |
| 8 | Improve user onboarding: interactive product tour, use-case based signup, prefilled first project, contextual tooltips, project template library, onboarding assistant, progress checklist, collaborative onboarding, … | M | Improved Studio | |
| 9 | Improve AI/NLP pipeline configuration: Visual pipeline editor, Real-time validation, Run a pipeline step-by-step, Pre-built pipelines, Fault tolerance, Cost-aware execution, Batch vs. streaming modes | L | Improved Studio | |
| 10 | Chatbot’s tools configuration: Let the user decide which AI tools (web search, document retrieval, API connectors, etc.) are active for each individual project’s chatbot. | M | Improved Studio | |
| 11 | Advanced RAG | Loop on questions/documents, evaluator LLM (Relevance, Faithfulness, Completeness…), report (per question, per document, global…) | M | New Web Service |
| 12 | Query transformation & normalization (typos correction and spell checking, business vocabulary normalization, cross-lingual normalization, Intent & filter analysis, Query reformulation & expansion, enriched query object) | M | New Web Service | |
| 13 | Extensive search (not the top n results of RAG): Query decomposition, Iterative passes over the full knowledge base, Confidence scoring as a stopping condition, Full evidence pool output | ? | New Web Service | |
| 14 | Self-auditing answer engine (trust score): Retrieval confidence, Source grounding, LLM self-assessment, Answer coherence checks, RAG fallback pipeline (Image retrieval, Visual context representation, Vision LLM answer generation) | L | New Web Service | |
| 15 | GraphRAG: LazyGraphRAG, KG construction, Hybrid search (fulltext, vector, graph), cross-document entity linking, confidence scoring, explainability, graph optimisation, query routing, cross modal GraphRAG, Agentic GraphRAG, Ontology – OWL | XL | New Web Service | |
| 16 | NLP tasks | Support segment classification (categorization project) | M | New Web Service & Studio |
| 17 | Support anaphora resolution | M | New Web Service & Studio | |
| 18 | Support intent & slot filling for event detection (LLM), generate a structured output | S | New Web Service & Studio | |
| 19 | Support relationship extraction (LLM), generate a structured output | S | New Web Service & Studio | |
| 20 | Ability to link entities to any external KG | M | New Web Service & Studio | |
| 21 | Provide a normalization framework to populate Knowledge Graph | M | New Web Service & Studio | |
| 22 | LLM-assisted active learning (Text Classification, NER) | M | – | |
| 23 | LLM-assisted synthetic data generation | S | – | |
| 24 | Kairntech Chatbot | Metadata filtering: Dynamic / User-Declared Filters, LLM tool-calling to convert natural language to structured filters | M | Improved Chatbot |
| 25 | MCP support | S | Improved Chatbot | |
| 26 | UI: Chainlit framework replacement | M | Improved Chatbot | |
| 27 | Multi-agents implementation (LLM) | M | Improved Chatbot | |
| 28 | Platform Monitoring | Access log | S | – |
| 29 | User behavior metrics. Automatic weekly report | M | – | |
| 30 | Deployment & scalability | SSO-OKTA autentication (Machine-to-Machine) | M | Improved platform |
| 31 | LiteLLM product integration & re-engineering (LLM, embeddings) | M | Improved platform | |
| 32 | Chatbot in K8s | S | Improved platform | |
| 33 | Elasticsearch scalability (vectorization) | M | Improved platform | |
| 34 | Incremental processing pipeline | L | Improved platform | |
| 35 | Sherpa-core in K8s (split Sherpa container: Annotate API vs All-the-rest API, jobs) | L | Improved platform | |
| 36 | Hot deployment of pymultirole component | ? | Improved platform | |
| 37 | entity-fishing scalability | M | Improved platform | |
| 38 | Technical debt | Java upgrade (for k8s) | M | Improved platform |
| 39 | Python upgrade 3.12 (only pymultirole components) | M | Improved platform | |
| 40 | Suggesters components refactoring (python upgrade, updated/clean engines, support of K8s…) | L | Improved platform | |
| 41 | Suggesters new engines | L | Improved platform | |
| 42 | SDK industrialization (documentation, samples, creating & saving new components, deployment…) | L | SDK | |
| 43 | Quality Evaluation | Merge Quality Evaluation Server in Studio (BE & FE) – Support NER & EntityLinking | M | Improved Studio |
| 44 | Support Classification task (confusion matrix…) | M | Improved Studio | |
| 45 | Support RAG Multimodal (Retriever, Answer generator, with standard metrics and LLM-as-a-Judge) | L | Improved Studio |