I wrote this because the question I hear most from US contractors is simple: what will this actually save me? Here is how I work it out, with every assumption on the table so you can swap in your own numbers.
Where do the hours go on a typical bid?
Every company is different, so build your own version of this table. The structure matters more than the exact numbers.
| Task | Illustrative hours per bid |
|---|---|
| Finding and qualifying the opportunity | 3 |
| Reading the bid package, specs and addenda | 10 |
| Contract and risk review | 4 |
| Takeoff | 12 |
| Soliciting and leveling sub bids | 8 |
| Bid forms, qualification statement, bonds | 3 |
| Total | 40 |
At a $75 loaded hourly rate, that is about $3,000 per bid in staff time, before overhead. A team that submits 8 bids a month spends about $24,000 a month just to compete.
What does the bid cost when you lose?
Most bids lose. At a 1-in-8 win rate, each won project carries the cost of eight bids: about $24,000 of staff time per win in this example. That is why reading faster and choosing better bids matter as much as pricing.
How much can AI agents save per bid?
BAU AI's savings calculator uses these illustrative time reductions, not yet validated with US data:
| Task | Agent | Illustrative reduction | Hours saved per bid |
|---|---|---|---|
| Finding and reading | Clara | about 85% | 11 of 13 |
| First-pass contract review | Dora | about 90% | 3.6 of 4 |
| Sub bid leveling | Nova | about 78% | about 6 of 8 |
| Takeoff | Theo | you set it; 60% in the calculator | about 7 of 12 |
| Forms and qualification statements | Clara | most of the typing | about 2 of 3 |
In this example, about 30 of 40 hours come back per bid, about $2,250 per bid at $75 an hour. A person still reviews every finding and signs every form, so the remaining hours are judgment, pricing and approval.
What does that mean for revenue?
The saved time can go two ways. You can bid the same volume with less overtime, or you can bid more. If the team keeps its 320 bid hours a month and the cost per bid falls from 40 to about 15 hours, the same hours cover far more bids. In practice the limit becomes how many good-fit opportunities exist and how many your estimators can price well.
Illustrative example: moving from 8 to 12 qualified bids a month at a 1-in-8 win rate adds about 6 wins a year. On an average contract of $2 million at a 5% net margin, that is about $600,000 of added profit a year. Every input here is an assumption; plug in your own win rate, contract size and margin.
How do you work out your own numbers?
- Log the hours on your next five bids by task, using the table above.
- Multiply by your loaded hourly rate to get the cost per bid.
- Divide by your win rate to get the bidding cost per won project.
- Try the savings calculator, which covers all ten BAU AI agents, with your own team size and rates.
If you want to run these numbers on your own bids, join our US founding cohort and we will measure them with you, bid by bid.
Tushar Adsul, Founder and CEO, BAU AI
Frequently asked questions
Does AI replace estimators? No. Agents do the reading and the first pass and cite every finding to the page. Estimators price the work and a person approves every decision.
Are these savings guaranteed? No. The rates are illustrative until there is US production data. BAU AI is live in Europe and onboarding a US founding cohort on their own bid packages.
What about takeoff? Takeoff time depends heavily on drawing quality and scope, which is why the calculator lets you set Theo's time saving yourself.
Sources
- BAU AI savings calculator, bauai.co (illustrative rates).
- Hours per task are an illustrative model for a mid-size commercial bid, not survey data.