Explore the key innovation points in this roadmap, which reveal the improvements and breakthroughs on the way.
- PyGEAI v0.7.0
- The pygeai.proxy module will be deprecated and moved to a standalone package. You must migrate to the new pygeai-proxy package.The proxy functionality for managing MCP and A2A servers has been extracted from the main pygeai package into a dedicated pygeai-proxy package. This change provides better separation of concerns and allows the proxy functionality to evolve independently.
- PyGEAI v0.6.0
- Responses API with Streaming
- Responses API Support: New get_response() method in ChatClient for accessing the Responses API endpoint
- Multimodal Inputs: Support for sending images and PDF files alongside text input
- Real-time Streaming: Stream responses in real-time for immediate feedback
- Function Calling: Complete support for tools and tool_choice parameters for advanced function calling
- CLI Integration: New geai chat response command with comprehensive parameter support
- Authentication & Access Control
- API Token Management: Complete CLI and client support for managing project API tokens
- Access Control Endpoints: Comprehensive API for managing organization and project memberships, roles, and permissions
- Agent Migration & Import/Export
- Migration Tools: Enhanced migration capabilities for agents, tools, and agentic processes between environments
- Embeddings Enhancement
- Improved Embeddings API: Enhanced embeddings generation with better parameter support
- Additional Parameters: Support for encoding format, dimensions, user tracking, input type, and caching options
- Bug Fixes
- Fixed evaluation tests
- Fixed Docker CLI tests
- Fixed AILabClient credential passing to AdminClient for token validation
- Corrected test signatures and mocks (removed old project_id parameters)
- Fixed AI Lab tests, processes tests, and migration tests
- Improved file response handling in core files module
- Fixed CLI help examples formatting
- Corrected imports in documentation examples
- Enhanced informative messages for failed embedding generation
- Fixed organization and project message validation
- Resolved continuous development pipeline issues
- Fixed configuration loop issue
- When upgrading to v0.6.0, please note:
- Python 3.13 is now the recommended version (minimum 3.10 still supported)
- OAuth 2.0 Authentication Support
- Added access_token and project_id keyword-only parameters to all client classes
- OAuth authentication support in BaseClient, Session, and ApiService
- Automatic injection of Authorization: Bearer {token} and ProjectId headers for OAuth requests
- Backward compatibility maintained with existing API key authentication
- Validation ensures both access_token and project_id are provided together
- Support for OAuth in all clients: AILabClient, AgentClient, ToolClient, AgenticProcessClient, EvaluationClient, SecretClient, and more
- ToolClient refactored to remove manual header setting, relying on ApiService for automatic header injection
- CLI Verbose Mode
- Global --verbose / -v flag for detailed debug logging
- Usage: geai --verbose <command> or geai -v <command>
- Enables DEBUG-level console logging with detailed execution flow
- Logging includes:
- Command identification and matching process
- Option extraction and parsing steps
- Session and configuration information
- Command execution flow and completion status
- Formatted output: YYYY-MM-DD HH:MM:SS - geai - LEVEL - message
- Debug logs sent to stderr to not interfere with command output
- Complete documentation in CLI docs with examples and use cases
- Useful for troubleshooting, debugging, and issue reporting
- Enhanced CLI Error Handling & Validation
- ValidationError exception with structured error context:
- Field-specific validation errors with examples
- Attributes: field, expected, received, example
- Formatted multi-line error output with validation details
- Enhanced validators with comprehensive input validation:
- Integer validation with range checking and helpful error messages
- Float validation with inclusive boundary checks (e.g., [-2.0, 2.0] for penalties)
- Boolean validation with multiple accepted formats
- JSON validation with detailed parse error messages
- URL validation with protocol checking
- All validators provide expected format and examples on error
- Improved error message formatting:
- Consistent error type labels: ERROR [Type]: message
- Types: Validation Error, Unknown Command, Unknown Option, Missing Requirement, etc.
- Clear actionable suggestions with examples
- Preserved backward compatibility with exit codes
- UnknownArgumentError improvements:
- Proper typed attributes: arg, available_commands, available_options
- Type-safe exception initialization
- Better IDE support and error detection
- Enhanced CLI documentation:
- New "Error Handling" section with exit codes and message format
- Examples of error messages with suggestions
- Complete list of error types and their meanings
- Extended test coverage:
- Added 133+ test commands to Docker CLI test suite
- Validation tests for all parameter types
- Analytics Module
- Complete Analytics API integration for monitoring platform usage, costs, and performance
- New pygeai.analytics module with AnalyticsClient, AnalyticsManager, and response models
- 35 analytics endpoints covering lab metrics, request metrics, cost metrics, token metrics, and user/agent activity
- CLI commands (geai analytics) with 9 subcommands:
- agents-created (ac) - Get agents created and modified counts
- requests-per-day (rpd) - Get daily request counts with error tracking
- total-cost (tc) - Get total cost for a period
- average-cost (ac) - Get average cost per request
- total-tokens (tt) - Get token consumption metrics
- error-rate (er) - Get overall error rate percentage
- top-agents (ta) - Get top 10 agents by requests
- active-users (au) - Get total active users count
- full-report (fr)
- Comprehensive analytics report with CSV export
- Full Report feature combining all metrics in a single comprehensive report:
- Lab metrics (agents, flows, processes created/modified)
- Request metrics (total requests, errors, error rate, avg time)
- Cost metrics (total cost, avg cost per request)
- Token metrics (total tokens, avg tokens per request)
- User & agent metrics (active users, agents, projects)
- Top performers (top 10 agents by requests/tokens, top 10 users by requests/cost)
- CSV export functionality with --csv option
- Date range defaults: All commands default to previous month when dates not specified
- Inline endpoint documentation with request type and description comments
- Manager methods for all analytics endpoints:
- Lab: agents/flows/processes created and modified (total and per day)
- Requests: total, per day, errors, error rate, average time, per user
- Cost: total, per day, average per request/user
- Tokens: total, per agent/day, average per request
- Activity: active users/agents/projects, usage per user, averages
- Top performers: agents by requests/tokens, users by requests/cost
- Lab
- Improved Tools capabilities in the Lab so you can integrate external services faster.
- Import OpenAPI 3.x to auto-generate tools from grouped API methods.
- Attach multiple credential types (API Key, OAuth Client Credentials, OAuth Flow).
- Console
- Model Configuration Controls: Organizations will gain the ability to define which LLMs are enabled, improving governance and cost management.
- Station
- Enhanced publication flow from Lab to Station, clarifying visibility scopes and capturing solution metadata consistently to preserve privacy settings and maintain moderated organization-wide visibility.
- Tenant-isolated SaaS deployment option for Station, provisioned in a client’s cloud or on-prem environment, with IdP integration (Azure AD/Entra, Okta, SAML/OIDC) and in-tenant admin moderation for approvals.
- AI Pods client environments in Station, configurable per client organization, including IdP setup and an administrator role to moderate and publish only approved solutions.
- Improved confidential solution setup and usage in Station, enabling definition of execution privacy from Lab and consistent privacy indicators when solutions appear in Station.
- Security
- New centralized Credential Management module that lets you securely create and manage ApiKey and OAuth credentials at organization or project level, attach them to ToolPlugins or Tools, and resolve them securely at runtime.
- Store credentials in secure backends, validate OAuth scopes, and toggle active/inactive status.
- Support static (ApiKey, Client Credentials) and interactive (OAuth Flow) authentication.
- Use a UI to list, create, update, toggle, and delete credentials.
- Resolve and decrypt credentials at runtime only in secure backend services.
- Attach credentials to ToolPlugins or individual Tools with future-ready features (expiration/rotation metadata, versioning, tagging).
- LLMs
- Incorporation of Claude Opus 4.6, Anthropic’s flagship model optimized for advanced coding and agentic, multi-step workflows with tool use. Available now in production via Anthropic, and coming soon via AWS Bedrock and Google Cloud Vertex AI.
- Integration of GPT-5.2-Codex, an upgraded GPT-5.2 variant optimized for agentic coding tasks, available via the Chat API or the Responses API.
- Claude Haiku 3.5 and Claude Sonnet 3.7 will be migrated to Claude Haiku 4.5 and Claude Sonnet 4.5, respectively, to take advantage of improved intelligence and enhanced capabilities. For more details on the migration, please refer to Deprecated Models.
- Integration of the GLM-4.6 model hosted on Globant’s DGX H200, improving performance and delivering reliable inference for AI capabilities in Globant's Corp and Clients environments.
This section lists features and improvements still in development, with no confirmed release date.
- API
- New /videos/generations endpoint so you can generate videos via Vertex AI’s Veo 2, with proxy handling for long-running operations and MP4 decoding.
- New token management API so you can create, revoke, and update project-level API tokens using secure, role-validated endpoints (project tokens cannot manage tokens).
- New secure endpoints to list organizations and projects by user email that use OAuth and restrict access to ProvisioningServices or GAM Administrator roles, so you manage roles effectively across platforms.
- Improved portability via CLI and Python SDK export that lets you programmatically export agents, tools, and agentic processes for backup and reuse.
- Improved observability for the Data Analyst Assistant that forwards logs and traces via OpenTelemetry, so you correlate requests using trace_id, span_id, and trace_flags across your monitoring tools.
- Console
- Prompt Files: It allows you to upload files at the organization and project level so that the Chat Assistant you define can use them to answer questions.
- Quota Alerts: Email notifications will be sent when a project or organization reaches its soft limit, helping teams manage usage proactively.
- Model Configuration Controls: projects will gain the ability to define which LLMs are enabled, improving governance and cost management.
- Evaluation module backoffice.
- Security consent flow so you must read and accept Terms and Conditions in Lab and Console, with identity, date, and time recorded; consent persists across future logins.
- Improved governance with token consumption usage limits so you can set quotas in millions of tokens at organization and project level for tighter control and accountability.
- Improved global navigation consistency that introduces a standardized navigation bar across Console, Workspace, and Lab with unified links, placement, and accessibility.
- Lab
- Options to export/import agentic processes and flows.
- RAG Assistant and API Assistant will migrate from the Console to The Lab:
- This gives Assistants access to advanced configuration, custom tools, and a flexible development workflow in The Lab.
- New RAG Agents so you can connect agents to Knowledge sources (document repositories, vector stores, databases, or graphs) and configure retrieval tools. You link Knowledge to prompts and tools, create agents with templates or via an assisted flow, and start chatting after uploading documents.
- Chat with Database: The current assistants will be fully migrated into The Lab interface, allowing for seamless usage within agent workflows.
- Audit Logs: The Lab will begin tracking user actions related to entity creation, updates, and deletions—strengthening traceability and accountability.
- Entity Version Management: Users will be able to view the version history of any Lab entity and restore previous versions when needed.
- New configuration option in the tool to generate images.
- New per-project avatar prompt templates with dynamic variables so you can differentiate brand and style across projects while preserving agent name and role; the generator compiles the project prompt with safe defaults and runtime variables.
- New prompt files in agents so you can reference uploaded prompt files directly inside an agent’s prompt to enrich context.
- Security guardrails per agent so you can prevent prompt injection and moderate both inputs and outputs with versioned checks validated against supported providers.
- Import MCP remote servers as ToolPlugins with auto-detected authentication and optional auto-registration.
- Improved autosave behavior so you can keep work safe during connection loss.
- Use an offline/local autosave buffer that syncs when the connection resumes.
- See clear status feedback such as “Saving…”, “Last saved…”, “Saved offline – syncing soon”, or “Save failed – retry”.
- Flows
- Improved BPMN compliance in the agentic process engine that enables throw/catch handling for intermediate signal events, so you coordinate process parts reliably.
- New deterministic Gateway conditions that let you define boolean routing rules using process variables (AND/OR), including natural language entry that the system translates into verifiable expressions you can review and edit.
- Improved editing confidence that adds resilient autosave with a local offline buffer and clear status messages (e.g., Saving…, Last saved…), so you avoid losing work during disconnects or session expiry.
- RAG
- New RAG‑enabled agents that let you define Knowledge sources, attach retrieval tools, and chat over documents, databases, vectors, or graphs, with a simple path to migrate from current RAG Assistants.
- Knowledge abstracts your sources.
- Retrieval tools define access, chunking, and prompts.
- Agents tie prompts, tools, and selected Knowledge for seamless RAG.
- LLMs
- Improved cost accuracy with conditional handling based on tokens consumed so you can align billing to provider models. Support includes Vertex AI (e.g., Gemini 2.5 family, including audio tokens), cached-token pricing, and regional pricing for providers such as OpenAI and Azure OpenAI.
- Improved cost accuracy that applies provider‑specific conditional pricing based on token counts (including audio and cached tokens) and regional pricing for providers such as Vertex AI, OpenAI, and Azure OpenAI.
- Station
- New redesigned Station header that centralizes access to Station, Lab, Console, and Workspace, and shows your active organization, project, and user to improve context awareness and navigation.
- New RunSpace execution area in Station that gives you a dedicated place to run, revisit, and organize AI solutions.
- Explore the execution UI before first run.
- Run, re‑run, and test solutions (including temporary runs).
- Find and continue previous runs or start fresh conversations with the same solution.
- Access grouped run histories by solution and recover past outputs.
- Mark favorites for quick access and share read‑only demos or context‑scoped runs.
- Execute and test Lab solutions before publishing to Station.
- New first‑use walkthrough that guides you with interactive tooltips on initial sign‑in and can be reopened whenever you need a refresher.
- Workspace/Playground
- VoiceAgents: Real-Time Conversational AI with Speech
- VoiceAgents introduces real-time, voice-based interactions with AI agents, enabling natural, two-way conversations through speech. Powered by advanced audio transcription, natural language understanding, and text-to-speech synthesis, this feature allows users to speak directly with AI agents and receive immediate spoken responses, bringing human-AI interaction to a whole new level of fluidity and accessibility.