Collaboration AI

Connects potential collaborators by grouping them and visualising their place within a network according to various qualitative data sources.


  • Align
  • Search

Type of tool

  • team building
  • networking


  • data analysis
  • network analysis


  • Private (for-profit)
Screenshot of Collaboration AI website

Case study


Collaboration.Ai is a platform that aims to help communities of users to understand what members contribute to the community. Using a combination of network theory, artificial intelligence, proprietary algorithms, blockchain and machine learning, Collaboration.Ai gains a system-level view of networks gleaned from custom data inputs. The platform can be used to help build optimal combinations of people for a specific purpose. For example, it could be used to determine who to work with to achieve a particular goal, how to seat people at an event according to shared motivations and social connections, or how to build teams.

Design of the tool

Collaboration.Ai has two products: HumanOS and a free Team Builder tool.

Collaboration.Ai HumanOS processes data from sources that might include open-ended survey questions, social media profile data, or custom data sources such as registration forms, employee performance data, or organisational and personal email records if the user grants privacy rights.

In addition to customising how data is sourced and elicited, users can also customise how the engine weights the data it considers. This allows them to override artificial intelligence in the platform’s algorithms, effectively enabling them to prioritise selection criteria according to their own expert preferences. Users can elect to ignore frequently occurring data to exclude them from analysis, or to emphasize less frequently occurring data by assigning them higher priority. For instance, an instructor on a biology course using Collaboration.Ai to build teams of students could choose to ignore the commonly stated interest area of ‘biology’, and choose instead to prioritise more specific indicators of sub-interests, like ‘synthetic biology’, ‘immunology’, or ‘plants and people’.

Accordingly, teams can be created based on characteristics or interests that emerge from the dataset rather than being imposed beforehand, or filtering can be applied to identify subsets of participants with certain characteristics or interests. Users can also use previous team configurations as exclusion criteria for future configurations to prevent participants from working with the same group of people again. Groupings can also be formed based on one-to-one relationships matching those who express opportunities or problems with those who articulate relevant skills, experiences, or solutions. Users and participants can both contribute to improving the performance of the system. Users can rate the quality of the generated groups, their individual skills, and work output – either immediately or after reviewing how they worked together; they can also feed in performance data. Participants can rate the quality of the teams that result from Collaboration.Ai’s groupings or confirm the new connections they acquired within the network.

Collaboration.Ai’s free version offers fewer options for input customisation, and is designed specifically for creating teams or groups of people.


To help users make the most of the platform, Collaboration.Ai facilitates online user groups with representatives from different clients. From Collaboration.Ai, clients learn how to use the tool better. From each other they learn potential techniques, applications, and ways of integrating the tool into their organisations.

Initially, Collaboration.Ai was focused on statistical outputs, but after discovering that users desired more transparency and control over how the platform prioritised what it considered, they introduced the custom weighting, as well as dashboarding and individual data authorization rights functionalities. Following this development, the platform provides automation but still allows for the re-introduction of human judgment.


A science organisation tackling climate change has used Collaboration.Ai to figure out which scientists should speak with each other in order to come to an agreement to move work forward more quickly. The US Air Force used the tool to find a new combination of actors within their existing supplier network who could contribute their services separately to quickly create a cheaper alternative to an existing helmet design.

Collaboration.Ai has also been used at events, such as TED Conferences and the World Economic Forum’s Forum of Young Global Leaders. Here, the platform was fed data about participants’ connections, career interests, and hobbies, as well as ‘softer’ information about friendships, families, faith, and personal initiatives. Data was gathered from registration profiles and an event-specific participant questionnaire. It was analysed to create groups of people who seemed to have the highest likelihood of creating meaningful work to achieve the World Economic Forum’s mission of ‘improving the state of the world’. These groups convened for structured icebreaker activities to facilitate networking at various points throughout the event. Based on feedback from participants after each group session, the platform could again be used to generate subsequent, new groups.

Comparison with other sectors and tools

Strengths: Collaboration.Ai is highly customisable in terms of data inputs and analysis. It also provides opportunities for users and participants to feed back into the platform to improve its performance over time.

Weaknesses: In the same way Collaboration.Ai augments human ability, humans have to augment the platform’s outputs in order to ensure its benefits are realised. Collaboration.Ai may help people form teams and connect with others in a network, but it does not guarantee that once relationships are initiated, they will last, or result in innovation or other desired outcomes. Although it can be customised to consider a variety of factors, matching people to build productive teams and establish valuable relationships is complex and there is always the chance that in real life, matches suggested by the platform do not work as hoped. This may be particularly relevant in terms of soft skills, working styles, or tacit knowledge that people bring to relationships but are difficult to ask about or express. Collaboration.Ai also relies on users (or organisations that users belong to) granting the platform access to data, making its handling of consent and privacy another potential concern.

Key takeaways

How well the system functions overall for team creation and networking can vary widely, depending on data source and quality, willingness of participants to provide responses or access to social network data, and the application. The platform is likely to be better suited to certain applications and contexts, but for any given user, this may take some trial and error to get right. Because Collaboration.Ai involves collecting data from different individual users and grouping or encouraging connections between them based on this data, how this data is elicited, used, and shared will need to be considered by any organisation thinking about using the tool.