Field notes

The end-of-year reading list for proposal operators

Fifteen links outside our bubble — books, long essays, research papers, category-adjacent writing. Not a proposal-industry reading list; the reading that makes proposal work feel new when you return to it in January.

The PursuitAgent research team 5 min read Research

This is the annual end-of-year reading list. It is deliberately not a proposal-industry reading list — the proposal industry’s writing is mostly tactical, and by late December the team has read enough of it. What follows is fifteen pieces from adjacent disciplines that reward a proposal operator’s attention and come back as mental inventory in January.

Links have annotations. Reading time estimates are honest; the long pieces are long. Not every link is equally weighted — some are books, some are papers, some are forty-minute essays. Pick three or four that match your January.

Retrieval, grounding, and AI research

1. Hallucination-Free? Assessing Legal RAG Tools — Stanford HAI. The foundational paper for the “citations are not grounding” argument. Commercial legal RAG tools hallucinate 17–33% of the time despite retrieval. If your proposal function is adopting AI-drafting tools in 2026, this is the paper to hand your legal team. Est. 45 min, PDF.

2. The issue of hallucinations won’t be solved with the RAG approach — Hacker News thread. The long discussion under the thread is more useful than the post itself. Practitioners arguing over span-level verification, economically viable defense patterns, the difference between retrieval grounding and claim-level grounding. Est. 30 min.

3. Mayo Clinic’s reverse-RAG approach — Hacker News thread. Per-claim evidence verification as the practical defense. The comments debate cost economics, which is where the real argument is. Est. 20 min.

4. The retrieval-evaluation chapter in the BEIR benchmark paper. Not a direct link (many mirrors) — search “BEIR benchmark retrieval.” The methodology for evaluating retrievers across domain-shift is directly applicable to how you should evaluate a proposal KB’s retriever when you add a new vertical. Est. 40 min.

Classification, decision-making, and the psychology of evaluators

5. Thinking, Fast and Slow — Daniel Kahneman, chapters 10–14. The anchoring and framing material. A proposal evaluator is a System-1 reader before they are a System-2 reader. Your win themes live or die in System 1. Est. 3-4 hours for the relevant chapters.

6. Superforecasting — Philip Tetlock. Not a proposal book, but the best available framework for how to make bid/no-bid decisions under uncertainty. The base-rate reasoning in chapters 4-6 transfers directly. Est. 6-8 hours full book; chapters 4-6 in 90 minutes.

7. “The Cognitive Style of PowerPoint” — Edward Tufte. Short, cranky, and full of arguments that port to proposal template design. An executive summary in slides is not a shorter executive summary; it is a different argument. Est. 30 min.

Writing and craft

8. On Writing Well — William Zinsser. Chapter 2 on “Simplicity” is the best 15 pages ever written on the prose discipline a proposal draft requires. If your team writes, you should have a copy of this. Est. 15 min for chapter 2; 4-6 hours full book.

9. “Politics and the English Language” — George Orwell. Free online, short, cutting. The six rules at the end apply verbatim to proposal writing. Rule six — “break any of these rules sooner than say anything outright barbarous” — has saved many responses. Est. 20 min.

10. The Elements of Style — Strunk and White. Not new, not original, and still correct. Omit needless words. Use definite, specific, concrete language. The entire discipline of a proposal response is hiding in this 85-page book. Est. 2 hours.

Operations, systems, and teams

11. The Phoenix Project — Gene Kim. A novel about a failing IT operation. The three ways of DevOps are the same three ways a proposal shop needs: flow, feedback, and continuous learning. The proposal-team retro maps to the third way directly. Est. 8 hours.

12. Team of Teams — Stanley McChrystal. The argument that specialized command-and-control cannot move fast enough against networked threats. The proposal function that wants to respond to a 30-day RFP with a 30-day process will lose to one that has compressed decision latency. Est. 8-10 hours.

13. “The Calculus of Service Availability” — a USENIX paper by Treynor, Dahlin, Rau, Beyer. Short and technical. The framing of error budgets is useful for proposal quality — a team that tries to ship zero errors spends infinite time; a team with a named error budget ships more on schedule. Est. 30 min.

Procurement, category, and markets

14. The annual SAM.gov procurement data drop. Not a book; an analysis task. The federal procurement data is public, downloadable, and under-analyzed by most of the proposal industry. A January spent with the 2025 data produces category insight your 2026 pipeline can use. Est. 3-4 hours for a first pass.

15. The Goal — Eliyahu Goldratt. A novel about a manufacturing plant applying the theory of constraints. The bottleneck analysis is the most useful framework available for figuring out which step in your proposal pipeline is actually rate-limiting. Most teams find their bottleneck is not where they think it is. Est. 6-8 hours.

Not on this list

Category blogs, competitor content, new LinkedIn essays. Not because they are bad, but because the team has already read them. The point of the end-of-year reading list is to refill the parts of a proposal operator’s attention that the category’s own content does not reach.

How to use this list

Pick three. Not fifteen. Fifteen is a bragging number; three is a commitment. One AI-research piece, one craft piece, one operations piece is a balanced set. Read them across the week between Christmas and New Year when nothing is landing in the inbox. Come back to work in January with a different frame than the one you left December with.

The ritual compounds. The list is different every year, but the habit of an end-of-year read is the thing that keeps a proposal function from becoming a narrow specialty that never reads outside itself. The category already has enough of those.

Sources

  1. 1. Stanford HAI — Hallucination-Free? Assessing Legal RAG Tools
  2. 2. Hacker News — RAG and the hallucination question
  3. 3. Hacker News — Mayo Clinic reverse-RAG