Past battle · 2025-06-23 UTC
AI Agents Platform Showdown — June 23, 2025
From the AI Agents Platform category. 27 marks placed across 6 fighters. Pactory.ai took the crown.
Final standings
The line-up
The fighters
Profiles of every tool that competed in this battle, ranked by their final score.
Pactory.ai is a freelance platform built around AI agents rather than human contractors. Businesses can browse, hire, and task autonomous agents to handle work such as research, content production, data processing, and routine operational tasks, paying based on usage or completed jobs. For agent builders, Pactory offers a place to publish their agents, set pricing, and reach customers without having to build their own distribution channel. The platform handles discovery, payments, and the contract-like interactions between clients and agents, aiming to make AI labor as easy to procure as freelance services.
Criteria breakdown
- Marketplace of specialized AI agents
- Task-based hiring and delegation
- Publishing tools for agent creators
- Integrated billing and payments
- Search and discovery across agent categories
Scaled Cognition
Research lab building foundation models purpose-built for agentic AI workflows.
Scaled Cognition is an AI research company focused on advancing agentic artificial intelligence—systems that can plan, reason, and take actions autonomously to accomplish complex goals. The team develops foundation models and infrastructure designed specifically for multi-step agent behavior rather than general-purpose chat. Its work targets the reliability gaps that limit today's AI agents, including consistent tool use, long-horizon planning, and robust decision-making across extended task sequences. The company positions itself at the frontier of moving large language models from conversational assistants toward dependable autonomous workers. Scaled Cognition is primarily relevant to enterprises, developers, and researchers building agent-based products who need models optimized for real-world execution rather than benchmarks.
Criteria breakdown
- Foundation models tuned for agent behavior
- Long-horizon planning and reasoning research
- Tool-use and multi-step task execution
- Focus on agent reliability and robustness
- Infrastructure for autonomous AI systems
Kolena Real Estate AI helps brokerages, title companies, and property management teams automate the processing of contracts, leases, disclosures, and other transaction documents. The platform extracts key data points, flags inconsistencies, and routes information into existing systems so teams can move deals forward without manual review bottlenecks. By reducing repetitive paperwork tasks, the tool aims to shorten closing cycles and free agents and operations staff to focus on client-facing work. It is positioned for organizations that handle high document volumes and need consistent accuracy across varied real estate forms.
Criteria breakdown
- Automated document data extraction
- Contract and lease parsing
- Error and inconsistency detection
- Workflow integration for real estate teams
- Support for varied transaction documents
- Scalable processing for high deal volumes
Agentok AI is a low-code platform that brings the AG2 (formerly AutoGen) multi-agent framework to life through a visual canvas. Users can assemble agents, define their roles, and wire up conversations using drag-and-drop nodes instead of writing boilerplate code. The tool is aimed at developers, prototypers, and teams who want to experiment with agentic workflows quickly. It abstracts the underlying AG2 logic into reusable building blocks, letting you focus on agent behavior, prompts, and task orchestration rather than framework setup. Finished projects can be exported or run directly, making Agentok useful for both rapid prototyping and building production-bound agent applications.
Criteria breakdown
- Drag-and-drop agent workflow canvas
- Native AG2 framework integration
- Configurable agent roles and prompts
- Multi-agent conversation orchestration
- Project export and deployment options
- Reusable agent templates and components
BOTmarket Exchange is a compute marketplace where AI agents and developers buy inference capacity from competing providers. Workloads are identified by schema hash, letting buyers match requests to any provider that supports the same input/output contract without rewriting code or committing to a single vendor. The platform is designed for portability and agent-to-agent commerce. Agents can discover prices, negotiate, and settle compute purchases programmatically, while buyers retain the freedom to switch providers as availability, latency, or cost shifts. There is no lock-in to a specific model host or runtime.
Criteria breakdown
- Schema-hash based inference matching
- Multi-provider compute marketplace
- Agent-to-agent trading APIs
- Portable workloads with no lock-in
- Programmatic price discovery
- Pay-per-inference settlement
Arcee AI specializes in small language models (SLMs) designed to give organizations efficient, cost-effective alternatives to large frontier models. The company combines model training, customization, and deployment with an agentic AI platform that helps teams build and run task-specific AI workflows. Its approach centers on tailoring compact models to enterprise data and use cases, enabling faster inference, lower compute costs, and tighter control over deployment. Arcee positions itself for businesses that need domain-specific intelligence without the overhead and opacity of massive general-purpose models. The platform supports the full lifecycle from model selection and fine-tuning through agent orchestration, making it suitable for companies looking to operationalize AI across internal tools, customer-facing products, and automated workflows.
Criteria breakdown
- Custom small language model training
- Model merging and fine-tuning tools
- Agentic AI orchestration platform
- Flexible deployment including on-prem and VPC
- Domain adaptation for enterprise data
- Support for full model lifecycle management