How artificial intelligence can be used in bid writing

Learning capabilities of artificial intelligence

The process of writing a tender can often seem daunting. With aspects such as research, tender writing, and managing multiple opportunities simultaneously, the entire process can feel convoluted and time-consuming. However, as AI continues to become more sophisticated, and industry-specific AI becomes more accessible, various learning capabilities of AI could potentially be adapted to support in the bid writing process:

• Natural language processing (NLP) covers the interaction between the human language and technology – such as spell-checking. In the future, this could be used by AI to scan the questions within the tender and understand the key elements of what the tender requires, providing prompts to the author or potentially alerting them to any major omissions in their response.
• Natural language understanding (NLU) is a subset of NLP that identifies the meaning behind words and categorises them. For example, when assessing a sustainability question the system could use terms included within the question (e.g. ‘waste management’) and answer accordingly.
• Natural language generation (NLG) is the use of artificial intelligence (AI) programming to produce narrative from a data set. Typically, this data set must already exist in the form of a bid library or a similar structure. When combined with natural language understanding, an AI could potentially ‘read’ and interpret a tender question, and automatically generate an outline response by synthesising relevant content drawn from an organisation’s bid library.

Uses of artificial intelligence in bid writing

AI could potentially assist in the following areas:

• Research: The first stage of the tendering process is confirming your company meets the minimum requirements, such as minimum financial turnover, accreditations and experience. AI can streamline the process of assessing compliance by using key phrases input by the bid writer to crawl and collate specific data associated with that word. For example, it could crawl the tender document for ‘accreditations’, identify what the requirements are, and then crawl the bid library/database for matches.
• Bid management: An AI-based system could potentially enable bid writers to input the size/scope of the tender and calculate how long it will take to complete, produce tender responses, collate policies/procedures on file, and monitor the activity of each bid. For example, responses would be categorised as ‘in progress’, ‘requires further information not on file’, and ‘completed’. Bid writers would be provided with a detailed overview of the activity and any missing data required to complete the bid. The system can also conduct competitive analysis, providing insight into potential competitors bidding for the same contracts.
• Producing tender responses: following input of data (e.g. access to an existing bid library) AI could use its NLP and NLG capabilities to produce a tender response by:
  • Undertaking context analysis to assess what is required
  • Crawling the bid library to review/understand data and its relevance
  • Structuring documents and responses using key phrases in the tender question
  • Aggregating sentences and paragraphs by pulling relevant information together into a readable narrative
  • Checking for grammatical soundness.

The AI system could then produce a document containing responses which can be edited by the bid writer to meet the requirements set out in the specification. For ease, whole ITT/SQ documents can be uploaded onto the system, enabling multiple responses to be produced at once.

Benefits

Artificial Intelligence could support organisations to tender more efficiently and potentially more successfully:

• Use of AI could streamline the bid writing process, maximising the efficiency of your bid team, and allowing you to tender for multiple opportunities at once.
• Reduction of human error: using the company data set, the system could analyse the question and select appropriate data to match the response requirements, helping to mitigate the risk of an author misinterpreting a question or failing to identify the most relevant content held within the bid library.
• 24/7 availability, enabling constant monitoring of ongoing bids by the team and instant alerts to any missing data.
• Due to the nature of AI, the system is continually learning and evolving, improving its effectiveness and understanding of the language used.

Limitations

At first glance, AI may seem like the answer to numerous issues relating to tendering, such as lack of resource and time management. However, AI in the bid writing context is in its infancy, and it is not yet clear how effective it will be in creating truly competitive tender responses, or indeed how public sector buyers will respond to its use. As such, there are several substantial limitations that, at this stage, would limit your capabilities to produce a winning bid using AI software:

• AI systems lack human perspective, such as prioritisation and judgement, making it harder for the system to produce a nuanced bid that is bespoke to the client’s requirements, often providing generic responses requiring significant editing by the bid team.
• AI systems are not automatically aware of changes in industry or the introduction of new terms and often require manual updates. This runs the risk of the system describing complex methodologies and commitments in a way that is not applicable or appropriate to the tender.
• Often questions within tender documents are written obscurely, requiring interpretation to correctly answer questions. As AI systems use keywords to identify how to respond, in cases of obscure wording there is a higher chance of misinterpretation.
• AI systems can only use data sets on file to produce responses and do not necessarily take into account feedback from previous tenders, potentially leading to weaknesses in past responses being carried over to future bids. This means that the bid team will need to continue manually writing and updating content such as case studies, or responses. Similarly, bid libraries will need to be regularly reviewed and edited to remove no longer relevant information.
• When writing bids, human writers take into account circumstances that the system, if not included within the data set, would not. For example, in the event of extreme weather, the system might discuss the company policy/plan without necessarily taking into account factors such as employees travelling to the site. More broadly, it is unlikely that an AI system could consider the particular preferences and priorities of a given buyer, instead generating ‘one-size-fits-all’ content which is more generic by design.

Although the capabilities of the AI system are continually evolving in line with technological advances, there is still a long way to go. Human bid writers can consider the specific context of each individual service, including local challenges and opportunities, site- or service-specific requirements and the priorities of an individual buyer, producing a bespoke bid highlighting the strengths of your company. For support in any stage of the tender process, contact us free on 0800 612 5563 or email info@executivecompass.co.uk today to discuss how we can support your organisation.