Content area
It's so empting to ask a generative AI (gen AI) tool to analyze an investment or compare your company with another. AI tools quickly find insights and patterns in large datasets, answer questions about multiple "documents" uploaded to their platforms and predict the outcomes of recommended actions based on opportunities they see in the data. Shannon De Marco, Crayon marketing manager, states in a blog post, "Better prompts = smarter insights" and stressed that using appropriate prompts and understanding models' capabilities allow users to conduct research, analyze existing material, identify trends, and indicate gaps where further research should be pursued. Each prompt type elicits a specific improvement in the results produced by the AI tool: chain-of-thought prompting improves clarity and completeness, few-shot prompting gives the chatbot examples of desired output and contextual prompting adds company-specific details for better relevance results.
It's sooo tempting to ask a generative AI (gen AI) tool to analyze an investment or compare your company with another. AI tools quickly find insights and patterns in large datasets, answer questions about multiple "documents" uploaded to their platforms and predict the outcomes of recommended actions based on opportunities they see in the data.
Shannon De Marco, Crayon marketing manager, states in a blog post (crayon.co/blog/mastering-prompt-engineering-for-competitive-intelligence), "Better prompts = smarter sights" and that using appropriate prompts and understanding models' capabilities allow users to conduct research, analyze existing material, identify trends, and indicate gaps where further research should be pursued. Each prompt type elicits a specific improvement in the results produced by the AI tool:
* Chain-of-thought prompting improves clarity and completeness.
* Few-shot prompting gives the chatbot examples of desired output.
* Contextual prompting adds company-specific details for better relevance results.
LLMS AS COMPETITIVE INTELLIGENCE TOOLS
Large language models (LLMs) use retrieval-augmented generation to improve the quality of their responses by retrieving information from external sources to help reduce hallucinations and bolster results relevance. This can be particularly beneficial when dealing with specific, real-time, or private information, such as information gleaned from internal company research documents. Free or low-cost models may not consult recent publications or access the web to update their responses.
When prompted, most LLMs provide links to the sources they used to create their responses, which is undoubtably useful for assessing the comprehensive nature and currency of their responses. It's helpful for users to know when a model was built (to determine how dated its training material is) and which models can access newer publications, including the web, to update their responses. When researching a fast-changing industry, the Al systems may miss novel approaches that are beyond the patterns in their training.
SWOT/TOWS
The SWOT analysis is one of the foundational analytic techniques of competitive intelligence (CI). It's one of the easiest techniques to complete, though not always executed effectively. SWOT is often rendered as a bulleted list of the strengths (S) and weaknesses (W) of an organization and a recognition of industry opportunities (О) and threats (T).
SWOT can be an initial step toward gaining an advantage over existing and potential competitors. Users can find canned SWOT analyses in market research studies and business databases such as Business Source Complete, Gale Business: Insights, and ProQuest One Business, but these analyses are frozen in the time they were developed. Organizations change, so SWOT analyses must be dynamic, updated with new information.
I asked six Al tools using only their free models: "Act as a competitive intelligence professional. Conduct a SWOT analysis, synthesizing data from credible, authoritative sources. Present 3-4 bullets for the strengths and weaknesses of
Two additions to the exercise can improve your results:
1. Include the purpose of the analysis and its intended audience in your prompt, indicating what you want the audience to do upon review of the SWOT, such as, "I want to encourage
2. Rather than relying on the output of a single gen Al tool, use the bullets to spur further research to confirm the information in higher-quality research sources subscribed to by your library, since those are beyond the reach of LLMs.
Identifying the current competitive climate is merely a first step. I think that the better analytic technique is TOWS, which rearranges SWOT into threats, opportunities, weaknesses, and strengths. TOWS forces organizations to develop specific strategies, using a strength to capitalize on an opportunity, minimize a threat, and identify how to change weaknesses by mapping out concrete steps to overcome vulnerabilities in order to avoid risks. That would be a constructive follow-up prompt for your next CI chat session with your favorite LLM.
USING AI TOOLS FOR CI
Al tools can be effective with many aspects of CI. If the sector is new, these tools can miss novel contributions absent in the patterns of their training sets. Also, the tools do not deliver the same response each time when asked to create a competitive analysis report. I'd advise setting up custom instructions, such as writing and citation style (e.g., APA), the insertion of text citations, and inclusion of a reference source list at the end.
Here are some prompts that are helpful for CI analysis:
* Describe the current state of an industry, emerging market trends, and current industry challenges.
* Identify the top players and their products, including competitive strengths, weaknesses, and actionable strategies for gaining a competitive advantage in the industry. Include the following:
* Detailed descriptions of each company's organizational culture and leadership style
* An overview of financial health and notable strategic pivots of each firm
* How each organization is responding to current industry challenges.
In general, the replies are helpful, but I'd caution making any investment based solely on these responses. Perhaps each re-sponse should conclude with a disclaimer: "This response is for information purposes only and is not a recommendation for in-vestment or acquisition."
Each reference should be scrutinized for misinterpretation by the AI tool and reviewed to provide additional background information and evidence that the AI tool chose to minimize in its summarization for the CI report. Do the algorithms display any bias that might need to be remediated? You must be able to explain how an AI agent arrived at all decisions. Still, the use of Al can be a timesaver in conducting CI and devising competitive strategies.
DEDICATED CI TOOLS OR GENERIC LLMS?
CI professionals can choose a platform designed explicitly for CI purposes (e.g., Comintelli) or opt for a generic LLM with deep research capabilities combined with multiple tools that have AI layered on to enhance their translation, transcription, and other capabilities.
Insight? (insight7.io) helps users discover unexpected connections between concepts shared in discussions. It is an AI-powered CI platform designed for audio or video transcription. It analyzes individual transcripts, or groups of transcripts, summarizing conversations; identifying themes, patterns, and sentiments; and extracting actionable insights from unstructured data. The tool helps its users scrutinize conversations, pinpoint opportunities, establish priorities, and make recommendations based on the evidence contained in conversational data. These capabilities are essential for CI.
Insight7 works well for CI purposes, although its main focus is more on analyzing knowledge from conversational data gleaned from customer calls. A free trial entitles users to upload three files for analysis, obtain three transcriptions daily, and ask up to nine questions about the transcripts. The $19-per-month plan includes analysis of 10 "documents, one project analysis, 15 questions, and 20 hours of transcripts.
AI-enabled audio and video translators, transcription tools, and insight platforms are available elsewhere, but you'd have to use a mix of several to match the functionality of Insight7.
TRANSCRIPTIONS AND SPEECH RECOGNITION
If you've recorded interviews, you'll have to get a transcript of the recordings for analysis. There are plenty of AI notetakers and transcribers, such as Notta (notta.ai), Auris AI (aurisai.io), Sonix (sonix.ai), and OpenAl's Whisper (openai.com/index/whisper). There's even a transcript comparison tool you can use to compare the results from multiple tools (ulrikelanger.github.io/transcript-comparison-tool).
Having a transcript is only the first step. Next, you'll want to use AI tools to help you identify some insights that a human reading through the interviews may miss. The insight journey that uses Al to theme responses to open-ended questions depends on qualitative data analysis tools, such as NVivo (lumivero.com/products/nvivo).
Examples of AI-enabled audio and video translator and transcription tools include Google Pinpoint (journaliststudio.google.com/ pinpoint/about), Grain (grain.com), and Trint (trint.com). Tools using OpenAI Whisper's multilingual automatic speech recognition capabilities to transcribe and translate audio files include writeout.ai and Riverside (riverside.com).
Voice of the Customer tools and insight platforms employing AI to expand their capabilities include Kraftful (kraftful.com), Speechmatics (speechmatics.com), Castmagic (castmagic.io), and Siena Insights (siena.cx). These tools and platforms may find another use case in CI.
TOOLS AND PROMPTS
AI tools are adding modalities, such as deep research, that introduce existing functionalities of the Al models to the user interface. Xavier AI (xavier.ai) uses Al to construct slide decks for a Digital Strategy, Blue Ocean Strategy, or Porter's Five Forces strategic analysis for many industry sectors. While these are the three "projects" highlighted on the homepage, there are others, including Product Comparison and Market Analysis.
After creating a free account, I selected Porter's Five Forces project. Porter's looks at the external forces that impact an organization's ability to compete: suppliers, buyers, new entrants to the market-place and substitutes that threaten an organization's products/ services. Through a Five Forces analysis, organizations can determine the intensity of rivalry among their competitors and develop strategies to gain and maintain their advantage within an industry.
The system asked for an industry (Small Modular Reactor) and included an option to identify a company (NuScale Power). Xavier AI spent less than a minute thinking before delivering a 14-slide deck, Navigating Porter's Five Forces for SMRs. The screen has the appearance of PowerPoint (PPT), allowing the user to change the font or color, add text boxes, etc. There is a zero-learning curve for users. The right rail, Provenance, indicates the source of each element on the screen, including links. While my slides are in English, there were sources in other languages. Users can download the PPT or a PDE Unfortunately, the sources are not included in the PDF, which Xavier AI should include as a reference list on the final slide.
In another CI analytic technique, known as the McKinsey 7S Framework, the S's stand for strategy, structure, systems, shared values, skills, styles, and staff. A McKinsey 7S audit can help an organization recognize misaligned strategies inhibiting its growth. TAAFT (theresanaiforthat.com) crafted a detailed and complex prompt for using Al as a strategic business analyzer with the McKinsey 7S Framework (taaft.notion.site/The-Strategic-Business-Analyzer-1dbed82cbfd3802fa7d6ce960112eac4).
AGENTIC MAGIC
Can you use AI for CI? Sure. Should you? Probably. How (and how much)? More than you think. Tools designed for purposes adjacent to CI should be watched for their potential to incorporate into your organization's CI workflow.
Whatever you decide, recognize how quickly things change. Capabilities that appear in one tool are suddenly available in many; features that were available only in higher-priced versions of LLMs are now commonplace in free or low-cost ones.
Monitor the AI space and experiment. What would be useful is an Al assistant for CI (similar to SciSpace): You write a research task and SciSpace Agent chooses "the best Al Models, Tools and Data to complete it for you" (scispace.com/chat). Now that's progress!
Copyright Information Today, Inc. 2025