Can AI help a firm win more work?

What actually loses a firm the work?

Ask a partner why a pursuit was lost and you will usually hear something about price or about the incumbent. Ask the buyer and you often hear something duller. They were slow to come back. The document did not answer the questions in the order they were asked. The examples were from a different sector. Someone promised a follow-up note and it arrived nine days later, by which point the decision had drifted.

None of that is a failure of persuasion. It is a failure of logistics. Firms lose work they were well qualified to win because the machinery around business development is slow, manual and always competing with chargeable deadlines. That is the part AI can change, and it is a bigger part of the picture than most firms admit.

Which parts of business development are really admin?

Separate the pipeline into two columns. In one, put the things only a person can do: reading the room, judging whether the budget is real, deciding what to charge, saying no to a bad fit, telling a prospect something true that they did not want to hear. In the other, put everything else.

The second column is longer than people expect. Sorting new enquiries and working out who should own them. Checking for conflicts. Finding the three most relevant pieces of past work out of hundreds. Pulling team biographies and trimming them to the length the form allows. Assembling insurance certificates, policies and accreditations. Rewriting a methodology that has been written forty times before. Formatting, proofreading, chasing the partner who owes a paragraph.

Every one of those steps is necessary, none of them is why the client chooses you, and most are being done by people whose time is expensive. That is the opportunity.

Where does AI genuinely help?

Four places, in rough order of how quickly a firm feels the difference.

The front door. Enquiries arrive by email, through the website, from a referrer, sometimes through a procurement portal. In most firms they land in an inbox and wait for someone senior to triage them between meetings. A rebuilt process reads what has come in, pulls the obvious facts, routes it to the right team and drafts an acknowledgement that actually says something. Speed is the whole game here, because the first firm to give a useful answer tends to shape how the buyer frames the problem. Enquiry qualification and email triage are where that starts.

Finding your own material. Firms rarely lack relevant experience. They lack a way to find it under pressure. The scheme that answers the buyer's question sits in a folder named after a client who left years ago, described in a document nobody remembers writing. Knowledge retrieval turns that into something you can ask a question of, which matters most in the week a bid is due.

Assembly. Proposals and tenders are largely reassembly of existing content, shaped to a new set of questions and a new word limit. Rebuilding proposal and bid production and tender responses means a first draft exists before the bid meeting, not after it, so the meeting is about strategy rather than who is writing section four.

Follow-through. The note after the pitch, the revised scope, the answer to the clarification question. Small pieces of writing that get delayed by bigger work, and the delay reads as indifference.

What has to stay personal?

Everything that carries risk, price or belief.

The decision to pursue. A firm that pursues everything because pursuing became cheap will dilute its win rate and exhaust its people. If drafting gets faster, qualification has to get harder, not softer. Someone senior still says no.

The fee. What you charge reflects appetite, risk and what you know about the client. Fee estimating can do the arithmetic and assemble the assumptions; a partner still owns the number.

The promise. Anything you say you will do becomes a contractual commitment and, later, a professional indemnity question. A named person reviews and signs off anything that goes to a client. AI drafts; people approve.

And the conversation. Buyers are choosing who to trust with something consequential, and they are reading the person. There is no version of this where that gets delegated to software, which is the argument we make in the work that stays human.

Why does AI-written outreach backfire?

Because fluency is no longer evidence of effort. A message that is well written, mildly flattering and vaguely relevant used to signal that someone had spent twenty minutes on you. It now signals the opposite, and the moment a buyer recognises the rhythm the message becomes a small deduction from your credibility.

The same logic applies inside the bid. A methodology section that could have been written for any firm, for any scheme, scores badly whoever produced it. The failure mode is not that AI writes badly. It is that AI writes plausibly about nothing in particular, and a buyer reading twelve submissions is looking for the one that is about them. On whether to raise the subject with the buyer directly, see mentioning AI in a pitch.

Does faster actually mean more work won?

Only if you use the time properly. There are two ways a firm turns speed into revenue. One is capacity: answering more of the right opportunities without adding bid staff, which is the comparison set out in automation versus hiring. The other is quality: spending the saved hours on the parts that win, which means more time with the client before the deadline and more time arguing about the approach rather than the formatting.

The firms that get nothing from this are the ones that leave the freed hours where they were lost: scattered, unrecorded and quietly absorbed. Buyers already expect more, as we set out in the turnaround squeeze, so the advantage does not last long unless you take it deliberately.

Where should a firm start?

Take the last ten pursuits. For each, write down how many hours went into it, how many of those were assembly rather than thinking, who did them, and how long the buyer waited at each stage. The pattern is usually obvious within an hour, and it is rarely the part people assumed.

Then cost it. Hours lost, multiplied by the number of fee earners losing them, multiplied by your own charge-out rate, multiplied by 46 working weeks. The unbillable hours calculator does the arithmetic, and what to automate first helps you choose between the candidates.

Then rebuild one thing. Not a business development transformation: one process, the one costing you most, live and in use. For a consultancy bidding for framework places that is almost always tender production. For a firm living on referrals and inbound enquiries, it is the front door. The audit will tell you which one is yours.