Forgepoint Raised $9.8M for an AI Quote Desk. Price Speed Against Margin.

On October 14, 2026, Forgepoint announced a $9.8 million seed round for Frank, an AI inside sales rep built for manufacturers. (GlobeNewswire) Frank sits in the work between an incoming request for quotation and a customer-ready quote: it reads emails and part files, flags missing requirements, applies the shop's cost model, and routes drafts for human review. After approval, it sends the quote, follows up and records the exchange.

This is a good test of AI in a sales workflow where speed matters and a bad answer has a price. A slow quote can lose a job before the buyer hears back. A fast quote can still lose money if it misses a specification, underprices a difficult job or promises a delivery date the shop cannot meet.

What was announced

• Round: $9.8 million seed, led by Primary Venture Partners and a16z speedrun, with angel investors from OpenAI, Brex, Vercel and SoFi also participating. (GlobeNewswire)
• Product: Frank, an AI Inside Sales Rep for manufacturers. The product page says nothing goes out until a human signs off. (Forgepoint)
• Who keeps the judgment: engineers, estimators and salespeople keep the technical and commercial decisions; Frank handles the administrative work between them.
• Company-reported traction: Forgepoint says Frank reached seven-figure annual recurring revenue within six months of launch, that customers process 30% more quotes per month on average, and that time-to-quote has fallen by more than 55%.

Independent coverage the same day confirmed the raise and the quote-desk focus. (Forbes; SiliconANGLE)

Where the data stops

The ARR, 30% quote-throughput and more-than-55% time-to-quote figures are company-reported. Public material does not spell out customer count, baseline, measurement window, win rate or margin impact. Treat them as vendor averages until a shop can show the same lift on its own jobs, with realized margin after delivery.

More quotes vs better business

Faster response can raise quote volume without improving win rate or profit. A strong result would show which quotes Frank made faster, what the human reviewer changed, which jobs were won, and whether the realized economics held after delivery. Compare similar jobs across product type, complexity, customer and estimator before generalizing early averages.

Six lines for the quote desk ledger

This is the Reversal Ledger applied to one RFQ at a time. Keep the customer request, cost assumptions, human decision and final outcome connected:

1. Buyer request. The original RFQ, attachments and deadline.
2. Specs and gaps. Extracted specifications, plus the missing-detail questions sent back to the buyer.
3. Estimated cost. The shop's cost model applied to the package.
4. Reviewer changes. What the estimator, engineer or sales lead changed before approval.
5. Final quote and response time. What went out, in whose template, and how long it took from request to send.
6. Outcome and realized margin. Won or lost; then purchase order, realized margin, delivery performance, rework, pricing corrections, cancellations and buyer complaints.

Split the ledger by product type, complexity, customer and estimator. A strong average can hide a segment that is fast and wrong. Time saved without that comparison is still a claim.

Who owns which line

• CRO or VP of Sales: which RFQ types an agent may draft for a buyer, and the human-led baseline each type must beat on win rate and margin.
• RevOps: where response time, reviewer changes, outcomes and post-PO issues get logged so they join to pipeline and margin.
• Enablement: the review standard for estimators and sales engineers. Reviewing a draft quote is a judgment job, not a headcount cut.
• Estimating / sales engineering: the cost model, the exception list and who signs off when the agent is unsure.
• Finance: realized margin after delivery, and the spend caps on any quote-desk automation.

What to do this week

1. Pick one product family and pull the last 20 RFQs that reached a quote.
2. For each, log response time, who reviewed the draft, what they changed, win or loss, and any rework, pricing correction or complaint after the PO.
3. Where an AI draft was involved, mark it. Where it was not, treat the human path as the matched baseline.
4. Compare quote volume, win rate and realized margin. Grant more automation only where reviewer changes catch little and margin holds after delivery. Where review catches real problems, keep the reviewer and fix the inputs.

The same test fits a metal service center, a foundry, an injection molder or a machine shop. Count the reviewer changes and the post-PO economics alongside the speed. If a faster quote wins the job and loses money after delivery, volume was not the win.

Related

• Koch Backed Emanate's AI for Industrial Sales. Make Autonomy Beat a Control Group.
• Marklo Raised $5M to Tell Brands What to Market Next. Grade Every Pick on Margin.
• Reversal Ledger

Sources

1. Forgepoint, "Forgepoint raises $9.8M to help manufacturers quote faster and win more business," GlobeNewswire, Oct 14, 2026: globenewswire.com
2. Forgepoint, product page for Frank: forgepoint.ai/product
3. David Prosser, "Meet The Start Up Promising To Automate Manufacturers' Back Office," Forbes, Oct 14, 2026: forbes.com
4. Mike Wheatley, "Forgepoint raises $9.8M to automate custom price quotes for manufacturing firms," SiliconANGLE, Oct 14, 2026: siliconangle.com

If you want the next quote-desk decision broken down the same way, subscribe.