Signet field notes
Articles
Every Signet field note, newest first.
- $145,000 in one quarter. The AI sanctions cases are not about AI.At least $145,000 in sanctions for fabricated AI citations in the first quarter of 2026 alone (ComplexDiscovery, April 6, 2026). That is the aggregate that gets forwarded around a firm. The individual cases are more instructive, and they all fail in the same place.
- Nobody expects valuations to rise. Diligence friction is one of the few variables still yours.There were 167 RIA transactions in the first half of 2026, up 13% over what had been the record half. Forty-six percent of consolidators name $1 billion to $5 billion firms as their preferred targets. And on valuations over the next six months, 82% of surveyed participants expect them flat, 18% expect declines, and none expect increases (DeVoe & Company RIA Deal Book, Q2 2026, reported July 27, 2026). Record volume, buyer appetite pointed at firms your size, and no expectation that the multiple moves in your favour while you wait. When the multiple is flat, the price is set by what diligence finds, and most of what diligence finds is not fixable in the quarter before a transaction. Client mix, adviser tenure and revenue concentration are what they are. The operational governance file is one of the few items on the list that is purely a function of whether somebody kept it up.
- Before you automate anything, find out whether your ERP will let youEvery conversation about automating quote intake, order entry or AP matching ends in the same place: something outside the ERP has to read from it and write back into it. If that connection does not exist, is not documented, or is not supported, everything upstream of it is a demo. Settle it before anyone quotes you a number, because the answer changes the price of the work by a lot, and you can answer most of it yourself in an afternoon without being a programmer. Five checks, each a yes or a no, then what is actually known about the ERPs in this segment. It is a scoping question, not a dead end.
- A machine makes chips. Software doesn't. Here is the exception.Your instinct about software is correct and you should keep it. Put $340,000 into a machine and you know what you bought: a spindle, parts, second shift, and a resale value somebody will pay in cash. Put $340,000 into software and you get a login, an implementation partner, and dashboards three people open in the first month and nobody opens in the fourth. The exception is not about technology. It is about which lever you have left. You cannot control input costs and you cannot hire the estimator. What is left is how much of the office load runs without adding a person: margin defense, not innovation. Below is the arithmetic. Run it yourself. If it does not clear, do not do it.
- Your notetaker is producing records. Rule 204-2 does not care that nobody sent them.Rule 204-2 defines a record to include "transcribed information of any type, whether expressed in ordinary or machine language." The rule turns on what a communication is and does, not on who or what produced it, which means a meeting transcript or an AI-generated summary can be a required record even though it was never sent to anyone. The decision that matters is not whether to keep transcripts; it is that the decision has to be made in advance, at the category level, and that retaining everything is not the conservative option. Every retained transcript is a document you may be asked to produce, that you must be able to find, and that has to be consistent with everything else in the file. Two boundaries before we go further: there is no SEC guidance expressly addressing AI-drafted communications under 204-2, and whether a specific transcript at your firm is a required record is your compliance consultant's judgment, not ours.
- When a client's outside counsel guidelines arrive with an AI clauseIt arrives as a routine email. Updated outside counsel guidelines, effective the first of next month, please confirm receipt. Attached is a 22-page document you have seen 19 pages of before. Section 11 is new: notice before generative AI is used on the client's matters, no entry of the client's confidential information into any tool that trains on inputs, a written AI governance policy, and the client's right to audit. Three weeks to confirm.
- The quantity typo that shipped 2,000 partsCustomer POs arrive as PDFs and somebody re-keys them into the ERP between two phone calls and an expedite. Part number, revision, quantity, price, due date, ship-to, terms. One afternoon somebody types 2,000 where the PO said 200. The job runs. You find out at shipping, or the customer finds out at receiving, and now you are eating material, setup, machine time and freight, and having a conversation about your quality system with someone who was happy with you last week. Reading a PO and typing it into the ERP is high volume, rule-bound and zero judgment, which is exactly the shape of work that should come off a person's desk, and exactly the shape a person is worst at. Nobody was careless. 2,000 and 200 look alike at the end of a long line on a scanned PDF.
- Eighty-six percent of you wrote the policy. Fourteen percent have an incident response plan that mentions AI.Both numbers come from the same survey, and the distance between them is the next two years of work at most adviser firms. It is not a drafting distance. Writing a policy is a deliverable somebody can sell you; producing the evidence that the policy was implemented is an operating rhythm, and nobody sold you one. Rule 206(4)-7 requires an adviser to adopt and implement written policies, and to review annually both their adequacy and the effectiveness of their implementation. Adequacy is answered by a document. Effectiveness of implementation is answered only by artifacts: a dated inventory, a completed test, an incident record, a written determination, a review that left a trace. If you already have an acceptable use policy, buying a second one moves nothing, and arguably makes things worse.
- Both Reg S-P dates have passed. Can you produce the record for each service provider?The SEC adopted the Regulation S-P amendments on May 16, 2024. Compliance was required by December 3, 2025 for larger entities and June 3, 2026 for smaller ones; for an investment adviser, a smaller entity is one with less than $1.5 billion in assets under management. Both dates are behind us. Ten trade associations including the IAA asked for a six-month extension on November 19, 2025; it was not granted, and no no-action relief has surfaced. Most firms in the sub-$1.5 billion half adopted the policy and never papered the vendor layer underneath it, because the vendor layer is not a drafting exercise. It is a register, maintained, with a diligence record and a notification arrangement per row, and the population includes tools nobody procured.
- 69% of legal professionals used AI last year. 9% of firms enforce a written policy.Individual use of AI among legal professionals went from 31% to 69% in a single year, while 43% of firms have no formal AI policy at all and only 9% have a written policy that is actually enforced (8am Legal Industry Report, March 5, 2026). Fifty-four percent report no training and no plans to add any.
- The cyber clause in your supplier quality agreementIt did not arrive as a regulation. It arrived as a revision. Your customer's supplier quality agreement came through for signature, Rev C replacing Rev B, and around section 9 there are three paragraphs that were not there before. Safeguarding covered defense information. NIST SP 800-171. Possibly a CMMC level. A sentence about export-controlled technical data, and another requiring you to flow the same terms down to your own suppliers. Purchasing wants it back Friday. It landed on your desk because you own the QMS and a supplier quality agreement is a QMS document: not IT's, not the owner's, yours. You already run the machinery this needs. What you do not want is to sign a representation you cannot evidence.
- What a small law firm should automate first, and what it must never automateOnce the policy exists and somebody owns it, a different question arrives, usually from a partner who has just read something: fine, so what do we actually use this for.
- AI for SMBs: The Strategic Framework That Separates Results From NoiseMost AI implementations at the SMB level fail before they begin, not because the technology is wrong, but because the sequence is. Here is the framework that changes that.
- Doing More with Less: A Practical Guide for SMBs to Start Their AI Journey ResponsiblyAI isn’t just for tech giants anymore. With tools like ChatGPT and Microsoft Copilot, small and mid-sized businesses can now boost productivity, automate tasks, and make smarter decisions, all without big budgets or technical teams. In this blog, we break down what AI really means, why data management matters, and how you can get started with practical, responsible AI strategies tailored for SMBs.
Signet · A division of Circle Square
AI does the typing. A named human makes the judgment calls.
Book an AI Readiness CheckThirty minutes, no obligation. You keep the assessment either way.