]project-open[ Artificial Intelligence Roadmap

This page summarizes our plans for implementing AI based technologies in ]project-open[, our open-source project and project portfolio management software. If you are interested in details, updates, or if you want to pitch your idea or collaborate with us on roadmap items, please send a mail to info@project-open.com.

 


Slack Integration with Conversational AI

Chat "channels" can be associated with a project or another business object (program, customer, user, ...). Channel conversations are mapped to ]po[ forum discussion topic objects indexed by the full-text search engine. The plan is to include a chatbot  in the interface, in order to process certain natural language phrases in addition to backslash commands.

Status: Our plan is to include this package in ]po[ V5.1. A prototype is working with "backslash" (command-line) interface. First package available: [intranet-slack].

Integration: This package also supports Mattermost , the open-source competitor of Slack and potentially most other enterprise collaboration/chat products.

Disruption Potential: Low. Actually, we are not sure about the added value of a chatbot in the loop. People seem to get used to backslash commands.

 

 


Intelligent Project Planning in Specific Contexts

Project schedules are usually developed using a Gantt editor that calculates the duration of every activity based on resource assignments and uses finish-to-start and similar constraints to calculate a critical path etc. In specific contexts, optimum project schedules can represent a considerable value to an organization. For these cases, ]project-open[ offers a simple and open [REST API] that can facilitate the development of scheduling plug-ins.

Status: The [REST API] is available and stable. However, the ]po[ team will develop specific schedulers only when requested by a customer.

Disruption Potential: Low. This is possibly high value, but only for a single customer each.

 

 

Intelligent Resource Leveling

Conflicts in the assignment of resources to projects as part of the resource management discipline could be handled by an intelligent agent that would (similar to a resource manager) propose to re-allocated resources between projects.

Status: Idea

Alternatives: Integration with TaskJuggler . TaskJuggler performs automatic resource leveling using a "classical" algorithm (no AI at all), but performs pretty well in certain use cases.

Disruption Potential: Low. We haven't seen a single organization where such decisions would be taken by an algorithm, rather than a manager of a group of managers.

 


Intelligent Project Status (Warnings)

Train an AI to detect when projects deviate from their course.

Status: Idea

Training Data: Very limited. We believe that every "type of project" requires separate training, because the critical points to look at may vary across types.

Alternative: ]po[ allows to create rule-based notifications for individual tasks or the entire projects using the rule engine. We could pimp this package and implement some more advanced rules, which could be called (symbolic) AI.

Disruption potential: Low. Every project should have a PM looking at the status in a way a lot more intelligent than any possible AI might act, particularly based on a limited amount of training data.

 


Application Intrusion Detection System

]po[ is designed to be directly exposed to the hostile Internet. Apart from standard security measures, it already includes a number of sensors that report suspicious behavior. together could be called intrusion detection system . Adding some intelligence to this system would greatly improve detection sensitivity and reduce false positives.

Status: Idea

Disruption potential: Low. AI would be very suited for this task, but an IDS  is definitely not the core component of a project management system.

 

 

Types of Artificial Intelligence

Above, we have used the term AI in a slightly fuzzy way. Let's state clearly the available technologies that are frequently subsumed under this term:

  • Machine Learning and Connectivist ("Deep") AI:
    These statistical tools require a large body of high-quality training data in order to classify data in a certain way that may lead to decisions.
  • Conversational AI:
    Chat bots and NLP (natural language processing) technologies allow to extract (part of) the meaning of human verbal communication.
  • Symbolic AI:
    Not very fashionable these days, these techniques apply mathematical logic to problems, rather than statistics. There are valid uses of symbolic AI in scheduling projects with specific constraints, for example.

 

AI vs. Human Decisions

Project management is different from many areas where AI is currently applied successfully (classifying photos, matching users with products, ...).

  • In PM, a dedicated project manager is available to look at project data:
    This is in stark contrast to the mass-data environment found in social media and advertising where transaction costs (=human involvement) has to be limited to the maximum.

  • PM decisions usually have high financial impact:
    Again, this is different form mass-data environments, where the cost of an error might range in fractions of a EUR or USD.

  • AI needs loads of excellent training data:
    However, these data are not available in a PM context, because every project is usually different (by definition...). You would need to execute exactly the same type of projects hundreds of times to get reasonable training data.

  • AI is bad with "context":
    Project decisions depend on many contextual factors (type of project, phase in the project, mood of the customer, ...). However, AI completely depends on the context in which it has been trained. Any minor deviation will render it's results useless.

 

So a stupid, inflexible and expensive AI has to compete with a readily available intelligent human being in taking high impact decisions. 

Our prediction is that AI will not be up to core PM tasks in the forseeable future. However, there are a number of support areas where AI could be applied successfully. These are the cases that we want to list on this page.



  Contact Us
  Project Open Business Solutions S.L.

Calle Aprestadora 19, 12o-2a

08902 Hospitalet de Llobregat (Barcelona)

Spain

 Tel Europe: +34 609 953 751
 Tel US: +1 415 200 2465
 Mail: info@project-open.com