Pursuit ReplayPricingLearnWorkbenchAboutThe SprintResultsFAQBook a Pursuit Replay →
‹ The Workbench

Meet your evaluation board.

frwrd.eval runs your draft proposal past two AI panels: an external board seated to read like your actual buyer — the compliance gate, the tired skimmer, the skeptical SME — and an internal color team that reviews it the way you should. Findings come back in evaluator language, anchored line by line in the draft. The debrief, before you submit.

The external board — ten evaluators, ten ways to lose

Real proposals don’t fail one way. They get screened out on a form, skimmed past on page four, or scored down for a claim nobody substantiated. Each persona reads your draft through one of those documented failure modes — built from protest decisions and source selection procedure, not from job titles.

The Gatekeeper

stops reading at the wall

Section L screener

Page limits, fonts, forms, timestamps, amendments acknowledged. Real boards have eliminated proposals for counting the cover page wrong and for arriving one minute late — this evaluator stops reading at the page wall, exactly like they do.

The Hunter

reads their factor, nothing else

Assigned-factor evaluator

Real evaluators are often assigned one factor and open only those pages, searching for Section M’s keywords. Content in the wrong section doesn’t exist — no evaluator is required to piece your intent together from scattered parts.

The Skimmer

8 minutes per volume

Nine proposals and a day job

Reads headings, topic sentences, callouts, and captions — the documented way time-poor evaluators actually read. Either your points are findable in one pass or they don’t exist.

The SME

every flaw is a risk statement

Technical evaluator

Grades the substance of your approach and asks “how?” of every claim. Every flaw comes back as a calibrated risk statement in FAR 15.001’s terms — because a weakness isn’t “didn’t like it,” it’s a risk of unsuccessful performance.

The Points Accountant

scores only what M names

Scores strictly against Section M

Only stated criteria earn credit. Boards have had awards overturned for crediting features the solicitation never asked for — so this evaluator gives your unrequested brilliance exactly nothing, and tells you where it’s parked.

The Amnesiac

the page is all there is

Incumbent-blind reader

Deliberately forgets everything the customer knows about you. “A technical evaluation is dependent on the information furnished” — incumbents lose recompetes by writing to what’s in the room instead of what’s on the page.

The Past-Performance Hawk

scale is math, not argument

Relevancy auditor

Relevancy is size-and-scope arithmetic, not vibes — a $40M reference against a $241M requirement fails the test no matter how similar the work sounds. Tests every reference against the stated standard, no partial credit.

The Cost Realist

narrative vs. number

Price-technical coherence

Reads the technical volume with the price in hand: staffing your rates can’t support, salaries too low to keep the people you named, line items that contradict the narrative. One of them is wrong, and it gets written up.

The Chairperson

runs the room

Consensus enforcer

Real consensus meetings compress individual generosity — a strength one evaluator loved can legally vanish at the table. This persona runs that meeting, flags the outlier scores, and shows you what consensus ate.

The Competitor

your draft, their war room

Reads it to beat it

The black-hat seat. Reads your draft the way the incumbent’s capture team would: where you ghost, where you’re ghostable, and which of your claims they’ll turn against you.

Why these ten:each persona is built from the record — protest decisions where real proposals died, the government’s own source selection procedures and rating scales, and published research on how reviewers actually score. That research says experienced reviewers of the same document agree far less than anyone admits — so the board disagrees on purpose, the consensus pass shows you what the disagreement was, and a finding the whole board lands on is the one that deserves your night-before hours first.

Seated for your buyer

A state DOT does not read like a federal agency, and a county procurement office reads like neither. Every run opens by identifying the customer from the solicitation — federal, state, local, education, or a prime’s flowdown — and seating a board tuned to that buyer.

The seating changes more than tone. A state board scores against published point weights with price as formula arithmetic; a DOT qualifications board is barred from looking at price at all; a Florida panel scores knowing competitors can sit in the room and the sheets go public; a city board answers to a council vote. Even the findings switch dialects — federal work gets debrief language, a county RFP gets score-sheet language, because using the wrong one is itself a realism bug.

Two things never change with the seating: the finding taxonomy and the report structure. That’s deliberate — it’s what makes run four comparable to run one. And the report’s first page always says which board was seated and why, so a misread customer gets corrected before you trust a word of it.

The internal panel — the reviews you’d run on yourself

The external board reads your draft as the buyer. The internal panel reads it as the color team you may not have the people or the calendar to convene.

Compliance review

Every question answered fully and coherently, in a way that fits the requirements and the evaluation criteria — not just present, but responsive.

Red team

The adversarial scoring pass: the draft graded cold against Section M, the way a review board would grade it in the room — and it scores, it never copyedits. That drift is the classic way real red teams fail.

Editorial review

Claims that perform without proving, contradictions across volumes, and the sentences an evaluator has to read twice. Same commitments, made checkable.

Is this good business?

The bid-economics read: margin, terms, staffing reality, opportunity cost. Sometimes the finding is that the draft is fine and the deal isn’t.

What comes back

Six artifacts per evaluation, in the vocabulary of FAR 15.001 — because a mock eval that doesn’t speak debrief is just feedback.

The compliance screen

Every Section L requirement graded met, at risk, or failed — with the instruction quoted and the page cited. The Gatekeeper runs first because real boards screen before they score.

Read traces

What each evaluator saw, in order, and where they stopped reading. The Skimmer’s trace tells you how many of your win themes are findable in eight minutes. Usually it’s fewer than you think.

Evaluation worksheets

Findings per factor in the language of a debrief — strengths, weaknesses, significant weaknesses, deficiencies — each tied to the sentence or the silence that caused it.

Line-by-line comments

Each reviewer marks up the draft itself — comments anchored to the exact passage, one reviewer per color, like a review copy handed back. Findings you can fix in place, not a report you have to translate.

The consensus debrief

The panel’s combined read: a rating per factor and a fix list ranked by score impact, so the night before the deadline goes to the five changes that move points, not the fifty that don’t.

The run-over-run dashboard

Findings opened and closed between runs, screens cleared, findability trending up. The re-run after fixes is the point, and the dashboard is where you watch it work. Ships with subscriptions.

Every artifact is shown in full, on a constructed pursuit, at the sample debrief— inspect exactly what you’d receive before any of your material is involved.

How a run works

01

Give it the pair

One draft volume and the solicitation’s Sections L and M. The board never scores a draft without the rules it will be judged by.

02

The board reads

The Gatekeeper screens first. Then the seated evaluators read the draft, each through their own failure mode, and write their worksheets independently — the Chairperson runs consensus last and shows you what it compressed.

03

You get the debrief

Screen, traces, worksheets, consensus, ranked fixes — in minutes. Fix, re-run, watch the findings close. The re-run is the point.

What the board can’t tell you

The panels measure what’s checkable: compliance, findability, whether an evaluator could score what you wrote. They do not know your customer, the incumbent’s history, or your price-to-win — that judgment stays human, and pretending otherwise is how AI proposal tools get people thrown out. A mock evaluation is a measured read of a draft under stated conditions, not a prediction of award.

The human version of this is the Compliance Club— $750 a month, a person reads one submission at a time and tells you what to do about what they find. frwrd.eval is the scalable alternative underneath it: more runs, faster, cheaper, no human judgment. Plenty of teams will want both; be clear about which you’re buying.

Get the debrief before you submit.

Your first evaluation is free — one draft volume through the internal compliance review, one per organization. Self-serve subscriptions are coming, and the prices will be published like everything else here. Until then, send the draft or book a call and we run it with you.