Fee Proposal Assistant: Synthesis Knowledge Agent Demo For AEC Firms

All right, I'm going to keep going to level 3, which are the process-based assistants. Let's do the fee proposal assistant first. This one came directly from— actually, most of these examples have been inspired directly from clients. This one is like that. And in this example, what our client wanted to do, he's in charge of practice operations, is he reviews all the fee proposals. And he wanted to transfer some of that knowledge out of his head that he uses. Again, I'm going to keep hitting this phrase home, what good looks like. So define like a fee proposal checklist. So if I go and I look at the agent actions for this, and I look at the evaluate fee proposal, this one is a more complicated instruction set. You can see that it reviews the uploaded fee proposal document. It then compares the uploaded fee proposal against the questions on the page fee proposal checklist, which goes through and says like, look, when we put fee proposals together, these are the things that we want to make sure that we have covered. So it's gonna run through that checklist. And this is an interesting thing that maybe we'll get into a little bit later. It will include an obvious disclaimer at the top of the response, making it clear that it's an AI-assisted review and should not replace review by a human, preferably somebody with contract experience. So I actually added this one in. I didn't have this step in the instructions when I got started. And one time when I ran it, the agent just did it and it put a yield sign up at the top and it said like, hey, This was written by AI. So I'm like, great, let's formalize that. So I, you know, iterated on the instructions and added that in. Um, so once it does that, it's going to create a table and it's going to compare each question from that fee proposal checklist to the content in the uploaded fee proposal. And it will then build a table and the columns in the table will be the checklist question, the relevant proposal language, page number you can find it on. Was it sufficient? Needs minor improvement, needs major improvement. And then it recommends an improvement. to make to the fee proposal. And you can see it does a few more things. So great. Let's look at one of those. I'm gonna go ahead and upload it. So I'm gonna run this one live so I can ask questions of the agent. I can use voice mode, I can type, but I can also upload resources. So I'm gonna upload a document from this file. And so I'm gonna do the Sunset Ridge mixed-use development fee proposal. I'm going to kick this thing off, and then I'm going to, you know, jump out. But you'll see basically how it works. All right, so I've uploaded my document, and I'm going to say evaluate fee proposal. So it's reading through the checklist to make sure it has the context. Now you can see that it's writing that AI-assisted review warning that I mentioned. And now you can see that it's going step by step. Through that checklist, each question that we ask. So is the client clearly identified? Kind of. And so it has some recommendations and making sure that we know that that's actually the legal name, you know, of the company. And it's got some minor improvements, some things that are sufficient. Um, there are some here that it's flagged that need major improvement. And again, recommendations for each of those of what to do based on that client, um, the fee proposal checklist. All right. Um, so that's an example of a process-based agent.