Why we're writing this blog

This blog is not a content marketing farm. It is a field journal.

PursuitAgent is building proposal-intelligence software. A category that, in 2025, looks like it was built by people who left the industry a decade ago and never came back. The two largest vendors have been shipping essentially the same product since 2015. Their AI features, bolted on in the last two years, generate answers that practitioners describe on public review sites in words like “Magic doesn’t work well” and “it’s an overpriced document repository.” The prices are quote-only. The entry point is a 20K-a-year seat floor. The best engineers on your team spend 100 to 300 hours per response.

We think the category deserves a ground-up rebuild. We’re doing the rebuild. This blog is where we write about what we’re learning along the way.

What this blog is

It is five things.

A record of what we ship. The PursuitAgent engineering team publishes build-log posts under a shared byline, documenting how we chunk proposals for retrieval, how we evaluate our own RAG pipeline against held-out corpora, how we render citations in a way reviewers actually trust. These are not vendor posts with marketing chrome. They are engineering notes.

A record of the craft. Sarah Smith — a composite byline, disclosed on every post — writes about proposal craft: win themes that actually win, color-team review discipline, the unwritten rules procurement leads operate by when they write an RFP. Our house pen for that work is a single voice so the craft posts stay coherent week to week.

Our take on the category. Bo (that’s me) writes on Mondays about what’s broken, what’s changing, and where we’re betting. Category commentary gets a footnote-heavy treatment: if I say Loopio’s pricing floor is a 20K-a-year seat commitment, the citation is in the post.

Data, grounded. The PursuitAgent research team publishes benchmarks, teardowns of public RFPs, and quarterly state-of-the-category reports. Every statistic cites a public source. No invented customer wins.

Short utility posts. Changelogs, reading lists, field notes that don’t hang on a single author — shipped under the “PursuitAgent” byline.

What it isn’t

It isn’t a vendor blog. Vendor blogs sell. This one tries to be useful to the operator, whether they buy from us or not.

It isn’t a content SEO engine. You won’t find “5 Tips for Better RFP Responses” or “The Ultimate Guide to Everything About AI Proposal Software.” Those posts are written for Google. Ours are written for the person on a Tuesday who has to get a DDQ done by Friday. Titles under 65 characters. No listicles. No “ultimate guide.” No handwaving.

It isn’t an announcements feed. When we ship something notable, the relevant /platform/ page updates first. Then a blog post — if we have something to explain — follows. New product launches and the announcement of their feelings about the launch are not the same thing.

Rules we won’t break

These are copied from our internal voice doc. We’re publishing them because they only hold if we can be called on them.

No fabricated specifics. We will not invent customer names, quotes, case studies, or statistics. If a post would be stronger with a concrete customer win, we either get the customer’s permission or we write the post without it. There is no middle path where we ship “one customer cut their DDQ time by 62%” without a named case study behind it.

Competitor claims are publicly verifiable. If we say a competitor’s pricing floor is X, the citation links to their site, a public G2 review, or a filed document. If we can’t find a source, we don’t assert the claim.

Product claims match the marketing site. If a post describes a feature, that feature matches what /platform/* says. If the product changes, the marketing page updates first. The blog never gets ahead of the product.

Opinion is labeled as opinion. First-person, explicit framing, no pretending-to-be-reporting when we’re really arguing a position.

Every LLM-assisted post that demonstrates an LLM cites the model. The blog as a whole is operated by a small team using LLM assistance for drafting. We don’t litter every post with an AI disclaimer — readers are adults. But if a post shows a screenshot of a model output, the model and date are named.

The vocabulary we use

The proposal industry has a rich, specific vocabulary, and we use it without apology.

• RFP, RFQ, RFI, DDQ, security questionnaire — distinct things. We don’t conflate them.
• Bid/no-bid decision — the most important edit you’ll make on any RFP.
• Color-team review (pink → red → gold → white) — the review discipline that separates mature proposal shops from everyone else.
• Win theme — not a value prop. Not a message pillar. A specific, proof-weighted thread that runs through a proposal.
• Compliance matrix — the thing your proposal is graded against. Build it before you draft.
• Executive summary — the shortest high-leverage document in B2B sales.
• SME, capture plan, B&P budget, past performance, color team, red team, pink team, gold team, white glove — Shipley and APMP vocabulary that we use because it is load-bearing.

If a term doesn’t match this glossary, either it’s wrong or we’re learning something new and will say so.

The eight pillars

Every post fits one of eight pillars. They rotate, they don’t silo.

1. RFP Mechanics — the discipline of responding: reading, compliance, drafting, review, submission.
2. Grounded AI — what it means to draft from a corpus you can cite, and what still breaks.
3. Engineering Build Log — how PursuitAgent actually works under the hood.
4. Team & Workflow — SMEs, review cadence, proposal-team structure.
5. Category Commentary — the vendors, the pricing, the market’s direction.
6. Procurement / Buyer Side — the DDQ and security-questionnaire surface from the buyer’s vantage.
7. Craft — win themes, exec summaries, style. The parts that are still art.
8. Research & Benchmarks — data, teardowns, quarterly reports.

What to read first

Three posts land in the next 72 hours that set up the rest of the year:

• How the Grounded-AI Pledge is enforced in code — Engineering, tomorrow.
• The 8-stage RFP response pipeline, explained — Sarah, Sunday. 3,500 words and the canonical long-read for anyone new to proposal work.
• The RFP software category is broken in three specific ways — me, Monday.

A full year of posts lives in our editorial calendar. Every day, for 365 days. That calendar will miss days, and we’ll say so when it does. What it won’t do is run filler.

One honest question

A year of daily posts is a commitment we can keep as long as we keep having useful things to say. If at any point this turns into vendor-blog sludge, tell us. Email is public, our comparative pages invite correction, and the person writing this post is a ten-second Google search away.

Here we go.