Collty has introduced a new AI Engine that changes how projects are started on the platform. Instead of manually searching for teams and assembling a structure from scratch, users can now describe their project in plain language and receive a complete team configuration in seconds – including roles, workload distribution, and calculated cost. This is not a generative assistant that produces abstract suggestions. The system combines LLM-based understanding with a deterministic execution layer built on real team structures. The AI interprets the request, maps it onto existing team configurations, and then adapts those configurations based on scope and complexity.
System flow
When a user submits a request, the first step is interpretation. The LLM processes the input and extracts structured signals: what kind of service is required, what domain the project belongs to, how complex it is, and what type of outcome is expected. This transforms an unstructured sentence into a clear representation of the task. The system then evaluates available teams using a weighted scoring model: fitScore = w_1 \cdot service_match + w_2 \cdot industry_match + w_3 \cdot role_specialization + w_4 \cdot context_relevance This model ensures that teams are ranked not simply by keyword overlap, but by actual capability fit. Service alignment, industry experience, depth of roles, and contextual relevance all contribute to the final score, producing a more accurate representation of how suitable a team is for a given project.
Adapting real teams, not generating hypothetical ones
Once relevant teams are identified, the system does not generate a new structure from scratch. Instead, it takes existing team configurations and adapts them. Each role has a baseline workload, which is then scaled according to the scope and complexity of the project. suggestedHours = baseHours \cdot scopeFactor \cdot complexityFactor \cdot roleWeight This step is where the setup becomes specific. A broader project increases the scope factor, a more complex task raises the complexity multiplier, and different roles are weighted depending on their importance within the context. The result is a workload distribution that reflects how real teams operate, rather than a generic estimate.
A shift from marketplaces to execution systems
Traditional platforms are built around discovery – helping users find individual specialists or teams. Collty moves beyond that model by focusing on execution structure. Instead of presenting options, the system produces a starting point that already behaves like a project plan. This reduces the friction at the most critical stage of any project: defining scope, assembling the right team, and understanding the expected cost. By automating these steps, the AI Engine allows users to move directly from idea to execution-ready configuration.
What comes next
The current version of the AI Engine already delivers consistent and actionable setups, but it represents only the first stage of development. Future iterations will improve contextual understanding, refine workload adaptation, and expand cross-functional scenarios. The long-term goal is to make project structuring instantaneous and reliable – turning a simple description into a fully defined execution model without manual effort.