Update roadmap post 0.0.4 release

ROADMAP.md

@@ -8,7 +8,12 @@ Core features improve the library itself to cater wider range of functionalities
 
 | Name | Description | Status | Release version |
 |------|-------------|--------|-----------------|
-|Output formatter| Ability to templatize output format using pydantic| Yet to start| 0.0.4|
+|Output formatter| Ability to templatize output format using pydantic| Yet to start| 0.0.5 |
+|Resume work| Functionality that lets agents resume from where they stopped| 0.0.5 |
+|To Yaml| Explore the ability to convert code build agents into Yaml| Yet to start| 0.0.5 |
+|Web server| First step towards creating a publishable service to which agents can be saved and re-used| Yet to start| 0.0.5 |
+|Web app| A webapp where agents can be accessed like chat bot/slack| TBD |
+|Model routing| Explore the possibility to use a model router within the agents, instead of specifying every agent models | TBD |
 |Parellel Router| A router to execute tasks or agents in parallel if the tasks are independent | Yet to start | TBD
 
 ## Observability
@@ -18,6 +23,7 @@ These features improve logging and debugging abilities while building.
 | Name | Description | Status | Release version |
 |------|-------------|--------|-----------------|
 |Recursion control| Expose parameters like recursion control to limit recursions and policy in case of recursion etc | Yet to start | 0.0.5
+| Token count | Expose the total tokens used by an agent execution directly through session| Yet to start | 0.0.5
 
 ## Community
 
@@ -36,16 +42,16 @@ The community is welcome to suggest changes to the roadmap, through a pull reque
 
 | Name | Description | Status | Version|
 |------|-------------|--------|--------|
-|Linear Router|A router lets you build agents or teams that execute linearly or sequentially. The current router supervisor works in a hierarchical way where all the children report to one parent|  ✅ | 0.0.3|
-|Reflection| Reflection lets you build a component that can make the AI retrospectively look at the current output and retry or work again on the task at hand|  ✅ | 0.0.3|
-|Delegator| Delegator lets you build a component that can help delegate a flo to a particular agent, by some condition|  ✅ | 0.0.3|
-|Logging Framework|Better logging framework which can be extended to parent application (with log level control)|  ✅|0.0.3|
 | Full composability | Right now teams can only be combined with teams and agents with agents. We want to extend this to team + agent composibility | ✅ | 0.0.4 | 
 | Error handling | Ability to handle errors autonomously | ✅  | 0.0.4|
 |LLM Extensibilty| Ability to different LLMs across different agents and teams| ✅  | 0.0.4|
 |Async Tools| Ability create tools easily within asyncio | ✅  | 0.0.4|
 |Observer| Observer framework for raising agent decision events and other important events | ✅  | 0.0.4|
-|Set up tests| Create a framework for unit-testing flo-ai and its internal functionalities| ✅  | 0.0.4
+|Set up tests| Create a framework for unit-testing flo-ai and its internal functionalities| ✅  | 0.0.4 |
+|Linear Router|A router lets you build agents or teams that execute linearly or sequentially. The current router supervisor works in a hierarchical way where all the children report to one parent|  ✅ | 0.0.3|
+|Reflection| Reflection lets you build a component that can make the AI retrospectively look at the current output and retry or work again on the task at hand|  ✅ | 0.0.3|
+|Delegator| Delegator lets you build a component that can help delegate a flo to a particular agent, by some condition|  ✅ | 0.0.3|
+|Logging Framework|Better logging framework which can be extended to parent application (with log level control)|  ✅|0.0.3|