nac sic gpt

                                                                   

The COCOO AI Doctrine: A Strategic Model for Analytical Supremacy

This doctrine establishes the protocol for leveraging Artificial Intelligence (AI) platforms such as specialized GPTs (e.g., NACE Classifiers, Marketing Companions), and general large language models (LLMs) like Gemini and ChatGPT. These tools are not primary sources of evidence; they are a strategic layer of analysis applied to the intelligence gathered from all other platforms. Their purpose is to dramatically accelerate the process of sifting through data, identifying patterns, generating narratives, and weaponizing information to achieve COCOO’s core objectives.

1. Core Principles of Interrogation

Our use of AI is governed by the most advanced principles of the COCOO framework. We do not ask simple questions; we assign strategic missions.

  • The Ultimate Noisefilter: The primary function of AI in the COCOO doctrine is to act as the ultimate Noisefilter.1 We will feed vast quantities of unstructured data—such as hundreds of regulatory announcements, thousands of pages of financial reports, or extensive media coverage—into these models to have them instantly extract the critical signals from the noise, allowing our human analysts to focus only on what matters.
  • Force Multiplication: AI allows a single solicitor to perform the initial analytical work of a large team. It can summarize, translate, categorize, and draft at a scale and speed that is otherwise impossible, enabling COCOO to operate with superior efficiency and scope.
  • Narrative Generation and Weaponization: Evidence is only as powerful as the story it tells. We will use AI to rapidly draft compelling narratives for our complaints, Unsolicited Proposals (USP), and media campaigns. By providing the AI with the factual evidence, we can have it construct the initial arguments, frame the public interest angle, and tailor the message to the target audience (e.g., a regulator, a potential government client, the public).
  • Overcoming Platform Limitations: We will use specialized AI tools to overcome the functional gaps in other platforms. The most critical application of this is using a NACE/SIC Classifier GPT to categorize companies identified on OpenCorporates, which lacks this search functionality, thereby enabling targeted sectoral analysis on other platforms like the EC’s competition database or Companies House.

2. Weaponizing the Platform’s Arsenal: Capabilities and “Rules” of Engagement

Unlike structured databases, the “rules” for AI are not about search syntax but about the art of strategic prompting. Effective interrogation requires providing the AI with a clear mission.

  • Key Capabilities:

    • Summarization & Extraction: Condensing lengthy documents (e.g., annual reports, court judgments) into executive summaries or extracting specific data points (e.g., all mentions of “supply chain risk”).
    • Classification & Categorization: Applying labels to data, such as using a NACE Classifier GPT to assign industry codes to a list of companies.
    • Sentiment Analysis: Assessing the tone of news articles, social media commentary, or shareholder communications to gauge public and investor opinion.
    • Drafting & Generation: Creating initial drafts of documents, from formal complaints and USPs to press releases and internal briefings.
    • Translation: Rapidly translating documents from foreign jurisdictions to enable faster analysis.

  • The Rules of Engagement (Strategic Prompting):

    1. Define the Persona: Always begin by assigning the AI a role.
      • Example: “Act as a senior competition law analyst for the European Commission.”
    2. Provide Clear Context: Give the AI all the relevant background information it needs to understand the task.
      • Example: “I am preparing a complaint to the UK’s Competition and Markets Authority regarding a suspected ‘stealth consolidation’ in the UK veterinary services market.”
    3. State the Mission Clearly: Define the specific task you want it to perform.
      • Example: “Analyze the following 20 regulatory announcements from the London Stock Exchange. Identify every announcement related to an acquisition or merger. For each, extract the names of the acquirer and target, the deal value, and the stated strategic rationale.”
    4. Specify the Output Format: Tell the AI exactly how you want the information presented.
      • Example: “Present your findings in a markdown table with the columns: ‘Date’, ‘Acquirer’, ‘Target’, ‘Deal Value’, ‘Stated Rationale’, and ‘Key Risk Factors’.”
    5. Provide Examples (Few-Shot Prompting): For complex tasks, give the AI an example of the desired output to ensure it understands the format and level of detail required.

3. Strategic Interrogation: The Questions We Ask

We interrogate AI platforms with mission-oriented prompts designed to advance our strategic goals.

  • For StealthConsolid & MATOIPO Analysis:

    • “Act as a forensic financial analyst. I am providing the annual reports for three companies: [e.g., VetPartners, IVC Evidensia, Linnaeus]. Analyze the ‘Management Discussion & Analysis’ and ‘Risk Factors’ sections of each. Extract and summarize any language related to acquisition strategy, market consolidation, or competitive pressures. Identify any common themes across the three reports.”
    • “Here are 50 ‘Holding(s) in Company’ announcements. Identify all instances where a single fund, such as “, has increased its stake in multiple competing companies within the same sector over the last quarter. Present this as evidence of potential coordinated influence.”

  • For USP-to-WTO & Victim Identification:

    • “Act as a trade policy advisor drafting a proposal for the government of Kenya. I am providing data from the Global Trade Alert on a new, potentially illegal EU product standard affecting Kenyan coffee exports, and the financial statements of three major Kenyan coffee exporters. Synthesize this information into a three-paragraph executive summary for a USP, highlighting the economic harm to Kenya and positioning COCOO as the expert partner with the proprietary data needed to win a WTO case.”
    • “Analyze these Spanish-language agricultural news articles and forum posts. Identify and translate all mentions of farmer discontent related to low prices paid by large supermarket chains like [e.g., Mercadona, Carrefour]. This is for a FOC DAM (Find Other Claimants) analysis.”

  • For Public Tender & Competitor Disqualification:

    • “Act as a public procurement risk assessor. Our competitor for a government IT contract is [e.g., Capita PLC]. I am providing their last two annual reports and 15 recent news articles. Summarize any information related to project delays, cost overruns, shareholder dissent, or negative regulatory findings. Frame this summary as a ‘Risk Dossier’ for a government client.”

4. The COCOO-AI Strategic Playbook: A Model for Action

The following playbooks provide standardized workflows for integrating AI into our core operations.

Playbook A: The “Noisefilter” Funnel

  • Objective: To rapidly process high volumes of unstructured data to identify the most critical intelligence for human review.
  • Execution:
    1. Data Ingestion: Collect a large dataset from a primary source (e.g., all LSE announcements in the “Technology” sector for the past month from Investegate).
    2. AI Mission Prompt: Use a prompt based on the “Rules of Engagement”: “Act as an M&A intelligence analyst. Your task is to review the following 200 company announcements. Your sole focus is to identify any announcement that discusses a merger, acquisition, disposal, strategic investment, or the appointment of a new CEO or CFO. For each relevant announcement you find, extract the company name, the date, the headline, and a one-sentence summary of the event. Ignore all other announcements. Present the results in a table.”
    3. Human Analysis: Review the AI-generated summary table. Instead of reading 200 documents, the human solicitor now only needs to perform a deep dive on the 5-10 highly relevant documents flagged by the AI.
  • Strategic Outcome: This playbook dramatically increases the scope and efficiency of our market monitoring, ensuring we never miss a critical MATOIPO trigger or StealthConsolid signal.

Playbook B: The “NACE/SIC Classification Engine”

  • Objective: To overcome the lack of industry code search on platforms like OpenCorporates, enabling targeted, sector-wide analysis.
  • Execution:
    1. Extract Company List: From OpenCorporates, compile a list of companies in a jurisdiction of interest (e.g., all companies containing the word “Logistics” in Germany).
    2. Engage Classifier GPT: Use the specialized NACE Classifier GPT. Prompt: “Here is a list of 50 company names. Based on their names, assign the most likely NACE Rev. 2 code to each. Format the output as a two-column table: ‘Company Name’ | ‘NACE Code’.”
    3. Pivot to Targeted Search: Take the AI-generated list of NACE codes. Use these codes to run a targeted search on the EC Competition Cases database to see if any of these companies have been involved in EU-level antitrust or merger cases.2
  • Strategic Outcome: This playbook creates a direct, powerful synergy between our intelligence tools, using AI to bridge a critical functionality gap and unlock deeper levels of sectoral intelligence.

Playbook C: The “Complaint Drafter”

  • Objective: To accelerate the creation of formal complaints and other legal documents by using AI to generate a structured first draft.
  • Execution:
    1. Compile Evidence: Gather the key facts for a complaint (e.g., data from Violation Tracker showing a pattern of environmental breaches by a water company like “, and data from GOV.UK showing the regulator has missed its own enforcement targets).
    2. AI Drafting Mission: Prompt the AI: “Act as a UK public law solicitor. Using the following bullet points of evidence, draft the ‘Factual Background’ and ‘Grounds for Complaint’ sections of a formal complaint to the regulator, Ofwat. The complaint’s core argument is that the regulator is failing in its statutory duty to address systemic environmental violations by the company. Frame the argument as a matter of significant public interest (WPI).”
    3. Human Refinement: The solicitor takes the AI-generated draft, which provides a solid structure and narrative flow, and refines it with legal nuance, case law citations, and strategic polish.
  • Strategic Outcome: This playbook cuts drafting time significantly, allowing solicitors to focus on high-level legal strategy rather than initial composition, enabling COCOO to launch more interventions, more quickly.

Leave a Reply