Tenderiser – AI-Powered Tender Discovery & Qualification for Cohaesus Group

Overview

Cohaesus Group is a dynamic collective of specialist agencies dedicated to empowering digital agencies to achieve accelerated growth while reducing risk. Since its inception in 2008, Cohaesus has evolved from a technical partner into an ambitious global group with a scaling mindset.

Our Challenge

Tender management within the Cohaesus Group was becoming increasingly complex due to several challenges. Tender discovery was fragmented, with relevant opportunities spread across multiple sources like Contracts Finder, TED Europe, and Stotles. Without a centralised mechanism, this decentralisation led to duplicated efforts, missed deadlines, and lost opportunities.

Furthermore, tender qualification was time-consuming, requiring careful review to assess relevance. The diverse scope and services across the group made it challenging to determine which company should apply. The triage process required domain knowledge, alignment with past work, and contextual understanding, all of which were difficult to access quickly.

Coordination within the group was also strained. Without a centralised system, coordination often relied on informal communication through Slack, resulting in missed opportunities and last-minute efforts. Additionally, the absence of structured analysis led to strategic misalignment, as the group sometimes pursued tenders with low success potential while overlooking those better aligned with their expertise. To overcome these challenges, Cohaesus needed an automated solution to centralise tender discovery, streamline qualification, and intelligently recommend which company should respond.

Our Mission

We set out to create a Proof of Concept (PoC) for Tenderiser - an AI-powered platform designed to automate the tender discovery and qualification process. The goal was to reduce manual effort, standardise evaluation, and ensure that only high-potential opportunities were pursued.

Our Solutions

Our approach combined intelligent automation, AI-driven insights, and seamless integration with existing collaboration tools. The solution centred around creating a multi-source aggregation system. Tenderiser directly connects to key procurement portals, including Contracts Finder, TED Europe, and Stotles, running scheduled checks to retrieve newly listed tenders. The system then extracts basic metadata and any attached documents, consolidating them into a single, centralised location.

To enhance efficiency, we integrated AI-powered tender analysis using a blend of OpenAI and Anthropic LLMs. Tenderiser parses each tender’s description and documentation, identifying key attributes such as service areas, sector relevance, budget range, timeline, and specific requirements. These insights are then fed into a scoring engine.

The platform also supports custom scoring per company. Each Cohaesus company can upload a tailored knowledge base outlining their services, focus areas, past projects, and preferred bid types. Using this data, Tenderiser assesses each opportunity against the company’s capabilities. A multi-factor scoring model evaluates relevance across dimensions like domain fit, complexity, estimated effort, and strategic value. This scoring system ensures relevance and minimises wasted effort, giving each tender a dynamic, company-specific score for faster and more confident qualification.

Tenderiser also features automated notifications and a shared tender board. When a tender meets the suitability threshold for any company, stakeholders are instantly notified via Slack. The opportunity is then pushed to a shared Trello board, creating a centralised, visual overview of high-potential tenders across the group.

We built the system using n8n, a flexible low-code automation platform that allowed rapid prototyping and seamless integration with tools already used within Cohaesus. Google Workspace provided secure document handling, while Slack and Trello enabled smooth team collaboration without the need for extensive onboarding. The entire PoC was delivered in just three weeks, involving a lean team of one developer and one product owner, with input from key stakeholders from each Cohaesus company.

Extracted tender details

Loading details into vector store

Success story

The Tenderiser PoC transformed how Cohaesus handles tender opportunities. The most significant impact was the drastic reduction in manual qualification time, which previously took hours of cross-company analysis and internal discussions. With Tenderiser, triage and qualification now take just minutes, as relevant tenders are surfaced instantly while irrelevant ones are filtered out without manual review.

Strategically, the platform brought increased clarity. By combining AI with each company’s specific expertise, Tenderiser enabled more focused bid decision-making. Teams now only see tenders that match their strengths, freeing up time for quality responses instead of pursuing misaligned opportunities.

The solution received strong positive feedback from stakeholders, who appreciated the user-friendly interface and the accuracy of the scoring results. The shared tender board improved visibility and ownership, significantly reducing duplication and communication gaps. Importantly, the PoC laid the groundwork for further development, with internal interest driving the next stages of Tenderiser’s roadmap.

Retrieve tenders from Contract Finder

What’s next: Phase 2 Roadmap

Building on the success of the PoC, the next phase of Tenderiser will include several enhancements. We plan to refine scoring through feedback loops, allowing users to provide input on tender relevance. This will enable the system to fine-tune scores using reinforcement learning patterns.

Additionally, AI-assisted response drafting will leverage historical content and past successful bids to auto-generate draft responses, significantly reducing the workload on bid writers and SMEs. Tenderiser will also include a tender compliance assessment feature, parsing requirements and comparing them with existing company documents (such as ISO certifications and policies) to identify gaps or risks early.

We will further expand AI model use by incorporating additional large language models such as Google Gemini, alongside OpenAI and Anthropic, to improve understanding and response generation. Data storage will also be enhanced with PGVector, allowing richer, context-aware content retrieval. These improvements will transform Tenderiser into a comprehensive tender management assistant, empowering Cohaesus to pursue more opportunities with greater efficiency.

Project facts

• Project Duration: 3 weeks (initial PoC)
• Team Size: 1 Product Owner, 1 Developer, plus stakeholders from each Cohaesus company
• Project management: Agile
• Technology Stack: n8n, Google Workspace, OpenAI, Anthropic, Trello, Slack
• Key Features: Multi-source tender aggregation, AI-powered document analysis and scoring, company-specific knowledge bases, Slack notifications, Trello integration, and custom qualification workflows.
• Outcomes: Over 80% reduction in manual qualification time, centralised visibility of high-relevance tenders, strong internal adoption, and a clear roadmap for expansion into content drafting and compliance.