Research note

Why RFP Is Hard To Automate In General?

A systems perspective on why RFP work is difficult to automate. It examines the operational dependencies behind tenders, including knowledge retrieval and workflow coordination, as well as the governance and implementation challenges that remain after AI tools become available.

The Request For Proposal (RFP) is not just a set of tasks, it is a very tedious workflow such as:

Tender arrives → documents ingested → requirements extracted → bid/no-bid analysis → previous approved material retrieved → draft generated → unsupported claims flagged → SMEs assigned → approvals tracked → deadline monitored → final package generated → submission archived → new knowledge incorporated.

That's an operating system. Hence, the proposal teams need support because:

  • new tenders keep arriving;
  • company information changes;
  • employees change;
  • product capabilities change;
  • approved answers become obsolete;
  • integrations break;
  • models change;
  • exceptions happen;
  • quality needs monitoring;
  • deadlines need management;
  • new workflows get added.

A similar example can be why people continue paying accountants even though they understand accounting, cloud providers despite knowing how servers work, and managed IT companies despite owning their computers.


1. So, Why Can't We Simply Use Zapier Or ChatGPT?

Singapore's April 2026 MOM study found that 71.5% of firms still had not adopted AI. Among adopters, the main barriers included implementation cost and lack of in-house expertise. Larger firms specifically reported integration complexity and data-security concerns. Yet 70.7% of AI-adopting firms reported improved worker productivity. (Ministry of Manpower Singapore)

In fact, if I may paraphrase, Singapore government is loudy trying to say:

Tools exist. Value exists. Implementation is still difficult.

Likewise, Singapore has launched a programme intended to help 10,000 enterprises deepen AI adoption over three years. This clearly shows that AI integration still has a long runway ahead but this creates a lot of business and technical opportunities too. (Digital Development Ministry)


2. Which Companies Are Actively Working Towards Solving These Issues?

Here is a list of some companies that are actively working to solve this domain.

CompanyWhat they do
GovDashVertical AI rather than generic AI. It focuses specifically on the painful workflow of government contractors: capture → proposal → contract operations. It announced a $30M Series B in January 2026. (GovDash)
ArphieIt connects company knowledge sources, drafts RFP responses, tracks approvals and maintains knowledge. Its own case study reports one customer's RFP workload dropping from 20 hours to 2 hours. (arphie.ai)
GleanStarted around finding organizational knowledge and has expanded toward enterprise context and workflow automation. Glean says it passed $300M ARR in May 2026. (Glean)
WriterShows where your model could eventually go: company knowledge + integrations + repeatable AI workflows + approvals + governance. Their product explicitly goes beyond isolated prompts toward complete workflows. (WRITER Knowledge Base)

Luckily for me, these examples are also what closely resonate with my personal core strengths such as:

Land on one expensive workflow → deeply understand it → operate it → standardize it → reuse components → become software-like.

A question for the research

Work with complex tenders?

If your team repeatedly runs into a tender or proposal bottleneck that deserves closer attention, I'd be interested to hear about it. You do not need to share confidential information.

Continue reading

More research notes