By Jatin Modi, founder of Renaissance Group · Published 28 September 2026 · Updated 2 November 2026
How does AI prepare a quote or proposal from past projects?
It starts from the proposals that already won. Hastings reads the new brief, finds the closest winning proposals, applies your pricing logic and writes a full first draft in your template, with the source of every claim marked. A person reviews the draft and the sales director sets the final price.
| Material | What the draft takes from it |
|---|---|
| The brief | The customer's requirements and deadline |
| Past proposals | Structure and wording from proposals that won |
| Pricing logic | The line items and the basis of the calculation, taken from past projects |
| Case studies | Proof points relevant to this customer |
| Datasheets | Technical specifications for the products in scope |
| Your template | The layout of the finished document in Word or Google Docs |
How should a team prepare its proposal library?
A draft is only as good as the material it retrieves. Seven checks turn a folder of old proposals into a source the job can draw on.
Keep each winning proposal as a final document, with its price sheet and the discount rules used.
Label case studies with sector, year and the product involved.
Remove superseded specifications from the datasheets folder.
Note why proposals were lost, so the draft avoids the same weak sections.
Record which customers have agreed to be named as references and which have not.
Write down the pricing logic in plain terms: which line items repeat, what drives each one, and who may discount.
Name one approver for pricing and one for technical claims.
What does a pricing logic sheet look like?
A tool can only reuse pricing logic that exists in writing. The sheet below lists the fields, with no figures, because the figures are yours. Fill one row per line item that appears in your quotes.
| Line item | What drives it | Where it comes from | Who may discount |
|---|---|---|---|
| Base scope | Product or service chosen | Price sheet, with its date | Sales director |
| Volume tier | Quantity or number of users | Past quotes at comparable volumes | Sales director |
| Delivery or installation | Site, distance or complexity | Closed projects of similar shape | Delivery lead and sales director |
| Support term | Length and service level | Standard terms document | Sales director |
| Exceptions | Anything the past quotes handled by hand | A note in the closed project file | Named approver in each case |
What goes wrong when a draft is built from past wins?
Four failures recur whenever a draft is assembled from earlier documents, and each has a check that works with any tool.
| Failure | What it looks like | Check |
|---|---|---|
| Stale prices | A line item carries last year's price sheet | Every price line shows the date of the sheet it came from |
| Another customer's details | A name, volume or discount from a different project appears in the draft | The draft flags any customer name that is not on your agreed-references list |
| Winner's bias | A new segment has no winning proposal, and the tool borrows the nearest one | The tool says when no comparable exists and leaves the section for a person |
| Out-of-date wording | A description matches an old datasheet | Each technical claim cites the current datasheet version |
What do the benchmarks say about where proposal time goes?
Loopio's 2026 RFP Response Trends & Benchmarks Report, a survey of more than 1,500 teams, found that teams spend 33 hours per RFP, down from 35 the year before, and that on average nine people contribute to each one (Loopio, 26 March 2026). It also found that 79% of teams use generative AI, up from 68%, that organizations respond to 55% of the RFPs they receive, down from 63%, and that only 13% of teams attribute losses to proposal quality.
Loopio sells RFP software, so treat the figures as one vendor's survey of formal bids, which differ from many sales quotes. The reading for a buyer is that drafting speed is available to every competitor. The advantage comes from the evidence in the draft and from the choice of which opportunities to answer.
What should you ask any vendor of AI proposal drafting, and what does a good answer look like?
| Question | What a good answer looks like |
|---|---|
| Which of our documents does it read? | A named list: past proposals, price sheets, case studies, datasheets and the template, with permission walls you can inspect |
| How is the quote built? | From the line items and logic of comparable past projects, shown line by line |
| Does every claim show its source? | Yes, and a reviewer can open the case study, datasheet or earlier proposal behind a statement |
| Who sets the final price? | A named person on your team, and the tool says so in its output |
| What happens when no past project fits? | The tool says so and leaves the section for a person, with no invented comparable |
| Where does the draft land? | In the tools your team already uses, in your template |
| What happens in week one? | A defined setup: sources connected, library loaded and a first live opportunity named |
In a Hastings engagement, Hastings runs the agents and your sales director owns the review and the price.
How do you run a blind test before you buy?
This is the method we suggest, and it applies to any vendor. No results from it appear on this page.
Pick three proposals you won, of different sizes, and take them out of the library the tool can read.
Give the tool the original brief for each and ask for a first draft in your template.
Put each draft next to the proposal you actually sent, with the names removed, and ask a senior reviewer which sections they would keep.
Check every claim in the draft against its marked source, and count the claims with no source.
Compare the draft's price logic with the price you sent and ask the sales director where it differs and why.
Record the time from brief to a draft a reviewer can edit.
What does a sales team get in the first week?
The setup scope on the pricing page covers connecting your CRM, email, calendar and market sources, building the knowledge base and configuring the first jobs. For this job the knowledge base is your proposal library: past proposals that won, the pricing logic behind them, case studies, datasheets and your template.
In the pilot calendar, weeks 1 to 2 load that library and weeks 3 to 12 run the job against live opportunities. The quality of the first drafts follows the quality of the library, so a team with a tidy folder of winning proposals starts faster.
Where does the job stop?
A new offer with no precedent has no winning proposal to learn from, so the sales director writes the first version and the library grows from it.
Claims without a source in your material stay out of the draft, and the team adds them with their own evidence. The final price belongs to the sales director, and the draft proposes the structure of the quote.
For tenders that arrive as formal questionnaires, the tender and RFP desk applies the same approach to public bids. Case studies keeps the proof points in the library current, and the presentation builder turns the approved proposal into a customer deck.
Related pages
Choosing between approaches: AI SDR tools or hired SDRs, Hastings and ChatGPT Enterprise and Hastings and an internal AI team. Other buyer checklists: keeping the CRM current and the meeting brief. Plans and prices: pricing. All jobs: use cases.
Questions buyers ask
Can AI prepare quotes from past projects for a B2B sales team?
Yes, when it has your closed projects with their prices and the logic behind them. Hastings builds the quote from the pricing logic of comparable past projects and marks the source of every line. The sales director reviews the draft and sets the final price.
Can AI draft sales proposals from our previous winning proposals?
Yes. The draft reuses the structure and wording of proposals that won, adds the relevant case studies and datasheets, and writes in your template. Every claim carries its source, so a reviewer checks it against the original material.
What should I check before buying AI that drafts proposals from past wins?
Start with five of the seven questions in the table above: which of your documents it reads, whether every claim shows its source, who sets the final price, what it does when no past project fits, and where the draft lands. Then run the blind test, which turns the vendor's answers into evidence from your own proposals.
How fast is a first draft?
Hastings states the aim as proposal time dropping from days to hours, and the film shows an illustrative brewery group receiving a full first draft two days before its Friday deadline. For scale, Loopio reports 33 hours of team time for an average formal RFP (Loopio, 26 March 2026). Measure your own speed in the blind test by timing the step from brief to a draft a reviewer can edit.
Who sets the price?
The sales director reviews the draft and sets the final price. The quote logic is a starting point built from comparable past jobs, and the person accountable for the margin owns the number that goes to the customer.
Where does the draft land?
In Teams or Slack, and as a document in Word or Google Docs, where the team edits it.
Sources
Loopio, RFP statistics and win rates, reporting the 2026 RFP Response Trends & Benchmarks Report (26 March 2026; read 2 November 2026).
Hastings, pricing (setup scope and pilot calendar, read 2 November 2026) and the proposal first draft use case.
Disclosure: Hastings sells the job described on this page. The checklist and the blind test apply to any vendor.