Talking to your senior partner about AI: the conversation that actually works
Junior partners and senior associates often see the AI opportunity before the senior partner does. Here is the conversation that actually works to bring a skeptical senior partner along, including the three concerns that come up every time.
TL;DR
The bottleneck on AI adoption at most Indian law firms is not the technology and not the cost. It is one or two senior partners who have not yet agreed. Junior partners and senior associates regularly come to us with the same question: how do I have this conversation? Here is the structure that has worked at every install we have done, including the three objections that come up every single time.
Why this is hard
The senior partner who is hesitant about AI is rarely a fool. He is usually someone who has been right about a lot of things for a long time, and his instincts about technology hype have been correct more often than not. He saw the firm sink money into a knowledge management system in 2018 that nobody used. He saw a litigation analytics dashboard the firm bought in 2021 that the analytics vendor stopped maintaining. He has reasons to be skeptical, and most of them are good reasons.
The mistake most junior partners make in this conversation is treating the senior partner's skepticism as a knowledge gap. It is not. It is a pattern-matching exercise based on past disappointments. Saying "but this is different, the technology has improved" does not work, because every vendor has been saying that for 30 years.
What works is a different kind of conversation.
The structure that works
Set up a single 30-minute meeting. No deck. Bring two documents printed on paper.
Document one: the firm's last 10 Section 138 reply notices, or the firm's last 10 NDAs, or the firm's last 10 of whatever document type the partner cares about most. Get them from the matter management system. Print them out.
Document two: drafts of the same 10 documents produced by an AI workflow with the firm's corpus as context. We can produce these for any firm in about 90 minutes, no engagement needed. Most other vendors can too.
Lay both stacks on the table. Hand the senior partner a red pen. Ask him to mark which ones are the firm's actual drafts and which are the AI-generated ones. Do not tell him in advance.
The first time we did this, the senior partner correctly identified 6 out of 10. The second time, 7 out of 10. Across 14 installs, the average is 6.5 out of 10. Random chance would be 5.
The point is not to humiliate the partner. The point is to establish in 30 minutes that the AI output is at the firm's quality bar. Once that is established, the conversation changes shape.
The three objections that come up every time
Even after the document exercise, three objections will come up. They are always the same three. Be ready for them.
Objection 1: "What about confidentiality?"
This is the first objection 9 times out of 10. The senior partner is right to ask. The answer has three parts.
Where the data goes. All processing happens inside the firm's existing M365 or Google Workspace tenant. Nothing leaves. The firm already trusts Microsoft or Google with client data. The AI workflow uses the same trust boundary.
What the model retains. Under the firm's enterprise contract with the LLM provider, the model does not retain inputs beyond a 30-day abuse-monitoring window, and does not train on the firm's data. We get a copy of the data processing addendum showing this in writing.
What the audit trail looks like. Every output is logged. If a regulator or a client asks "how was this drafted," the firm has a complete record.
If the senior partner wants more detail than this, he is asking the right questions, and we should be in the room. But this answer is enough to get from "no" to "let me hear more."
Objection 2: "Will this make our work look generic?"
This is the second objection. The senior partner has spent 25 years building a writing style for the firm. He is not interested in seeing the firm's drafts come out sounding like every other firm's drafts.
The answer is the corpus. The AI workflow only sees the firm's own past work. The output reads like the firm because it was trained on the firm. Not on a public legal database, not on a generic contract repository. On the firm's drafts.
The document exercise above is what makes this concrete. If 6 out of 10 of the AI drafts pass the senior partner's eye-test, the writing is not generic. The writing is the firm's writing, just produced 30 times faster.
Objection 3: "What about the juniors?"
This is the third objection, and it comes from the better senior partners. The concern is not "will my juniors lose their jobs." It is "will my juniors get worse at lawyering."
This is a legitimate concern. The answer is in two parts.
The first part is that the workflows we install never produce a finalised output. They produce a starting point. Juniors still review, edit, refine, send up the chain. They do not go from "drafting from scratch" to "rubber-stamping the AI." They go from "drafting from scratch" to "editing a first draft." That is roughly the same skill, applied to a more useful starting point.
The second part is that the firm's training has to evolve. We tell every senior partner this. The old training was "here is a blank page, draft a Section 138 reply, partner reviews." The new training is "here is the AI's draft. Find what is wrong with it. Why is that wrong. What would you change. What would the partner change." That is actually richer training than the old way, because it teaches the junior to read critically rather than write reactively.
The senior partners who buy this argument tend to be the partners whose juniors do best in the new world.
What you do not say
Three things to leave out of the conversation.
Do not promise specific time savings. "We will save 12 hours per lawyer per week" is the kind of consultant promise that triggers the senior partner's skepticism. The savings are real, but the partner needs to see them himself, not be told about them. The 30-minute document exercise is more persuasive than any productivity number.
Do not name competitors. "Khaitan is doing this, AZB is doing this." The senior partner is not making decisions based on what other firms are doing, and bringing it up reads as social pressure rather than business case.
Do not argue with him. If the senior partner is not convinced after the document exercise, do not push. Schedule a second conversation a month later. Bring a small new piece of evidence. Most senior partners come around in 2 to 3 conversations. None of them come around in 1 conversation if they were skeptical at the start.
What happens after he agrees
Most of the time, the conversation ends with the senior partner saying something like "alright, show me what a first install looks like, and let me think about it." That is a yes in disguise. The next step is a teardown call where we (or whoever you choose) sits with the partner for an hour and walks through what the first 30 days would actually be.
That meeting is what we call the teardown. The senior partner usually leaves the teardown ready to commit to a Phase 1 install. Three weeks after that, the firm has its first workflow live, and the partner who was the last "no" is the first one to send the workflow to colleagues at other firms.
That is the path. It is not fast. It is not easy. But it works, and it works the same way at every firm.
Frequently asked
Not for the first conversation. Vendors trigger the senior partner's defensive instincts. The first conversation should be between you and the senior partner, with you having done your homework. The vendor (us, or whoever) comes in for the second conversation, after the senior partner has agreed to explore.