Which one are you?
Agencies not using AI, whose clients and competitors are. Agencies using AI with no plan: personal accounts, no approved tools, no data rules, nothing written down when someone leaves. And agencies using AI with infrastructure and a partner: approved tools, data rules in writing, a shared library, and someone whose job is to keep up as the models change. In our read, only the third group is still choosing its clients in five years.
Not using AI.
Their clients are, and their competitors are. At the next client meeting, the client already has an opinion about whether the agency is behind.
Using AI with no plan.
Personal accounts, no approved tools, no data rules, nobody tracking the spend. Every account manager solving the same problem a different way, and nothing written down when they leave.
Using AI with infrastructure and a partner.
Approved tools, data rules in writing, a shared library, and someone whose job is to keep up as the models change. Their people move fast and nothing blows up.
Which AI tools are approved for client and employee data at your agency, and where is that written down?
If a carrier or your largest client asked this afternoon where their census data has been in the last 90 days, who answers, and what do they say?
Who at your agency is responsible for knowing what changed in AI last month, and what did the agency change because of it?
If all three have clear answers, you are in the third group and you don't need us. If any of them don't, keep reading.
Why is this not a tooling decision?
AI is the cost structure every business will be measured against. Agencies that build on it and agencies that don't are on two different curves, and the gap compounds with every quote, every service call and every hire. Advising employers on benefits is knowledge work, and the hours behind that advice sit squarely in what the model builders' own scenarios describe.
Impact on the scale of the internet. Gains arrive gradually.
AI can do half of all knowledge work by 2030, most of it autonomously. Most of it is still done without AI. Knowledge-worker wages flat.
AI does nearly all knowledge work autonomously. Knowledge-worker wages fall more than 10 percent by 2030.
Source: Anthropic Economics, Economic Scenarios for Transformative AI (Korinek et al.), September 2026. GDP figures are relative to a no-AI path.
Every scenario has AI reshaping knowledge work. The variables are how much and how fast. In the middle scenario, AI can do half of that work by 2030 while most firms still haven't adopted it. That gap between what AI can do and who has built on it is the three groups.
Where we think this goes for a benefits agency. This is our projection, drawn from the work we do inside agencies, where the models are heading, and the market signals below. In the next twelve months, clients and carriers form an opinion. Employers already list AI capability as a reason to switch brokers. The agency that answers a benefits question in minutes, turns an RFP in a day and can show a carrier its data policy wins the room. The one that can't is explaining itself.
In two to three years, cost structures separate. The third group serves more clients per account manager, wins more RFPs per producer, and prices accordingly. PE-backed platforms, already 72 percent of agency deals, buy the agencies that can't, and install the stack the day after closing. In five years, the split is permanent. Every AMS, carrier portal, HR platform and employer expectation assumes AI in the workflow. The third group is choosing its clients and winning on service. The first two are competing on price against agencies that aren't, whether they are still on their own or inside a platform.
What is pushing on this right now?
Four things, all live this quarter. Employers now name a broker's failure to use AI as a reason to switch. PE-backed platforms did 72 percent of agency deals last year and standardize every office they buy. A personal AI account is not a HIPAA tool. And the AMS and benefits platform vendors are shipping AI features into products your team has not been trained on.
Employers now switch brokers over it.
In Zywave's 2026 Broker Services Survey of more than 1,400 employers, a broker's failure to use modern technology and AI tools appeared among the reasons employers would leave a broker for the first time. The share of employers who expect their broker to act as a strategic advisor rose from 56 percent in 2023 to 71 percent in 2026.
Source: Zywave, 2026 Broker Services Survey, July 2026.
The competition is being standardized.
Private-equity-backed and hybrid buyers did 72 percent of insurance agency acquisitions in 2025. Every office a platform buys gets a stack, a policy, and training within months. That is the other bidder at your next RFP.
Source: OPTIS Partners, 2025 agency M&A reporting.
A personal AI account is not a HIPAA tool.
The consumer plans of ChatGPT, Claude and Gemini are not covered by a business associate agreement, and conversations can be used for model training unless each person turns it off. Census files with protected health information in a personal account are outside your control. Approved tools on a business plan with a signed BAA are a different matter, and that is what the Foundation sets up.
Source: OpenAI, Anthropic and Google business associate agreement policies.
Your AMS is becoming an AI tool.
The major AMS and benefits platform vendors are shipping AI agents inside their products this year: prospecting email agents, renewal risk flags, census collection. The features arrive whether your team has been trained on them or not.
The gap between the second group and the third is widening this quarter, not next year. That is why the plan below is four weeks long rather than a roadmap.
Why this quarter, not next?
Q4 is when the most client data moves through a benefits agency. Renewal proposals in October, renewal decisions and open enrollment in November, year-end filings in December, January 1 effective dates. Whatever tools and habits your team has in October are the ones that handle all of it. A four-week Foundation started now has the tools and rules in place before that volume peaks.
Foundation window.
Four-week Foundation starts. Audit, tools and rules, on site, live.
Renewal proposals arrive.
Carrier comparisons, RFP responses and census files start moving.
Renewal decisions and open enrollment.
Enrollment forms, OE communications and employee questions at full volume.
OE closes. Year-end filings.
Compliance filings and clean-up. Pricing on this page valid through December 31.
January 1 effective dates.
Whatever habits handled Q4 are now the agency's habits.
Volume. Renewals, open enrollment, new-business RFPs and year-end compliance filings all land between October and January. Whatever tools your team has in October are the ones they'll use on all of it. Data. Census files, claims experience, enrollment forms and employee questions peak in Q4. If there are no rules yet for where that data can go, this is the quarter that matters. Change. The models and the AMS features change every few months. What a team learns on its own in the fall is partly out of date by spring, and the person who learned it is the one you can least spare to relearn it.
What do the four weeks look like?
Getting into the third group doesn't take a big decision. It takes four weeks: audit, tools and rules, two days on site, live.
- Week 1 · Audit. We inventory every tool and account, trace where data has gone, and pick your workflows.
- Week 2 · Tools and rules. Approved tools configured, policy written, library built and loaded with your workflows.
- Week 3 · On site. Two days with your whole team on your systems, your accounts, your real client work.
- Week 4 · Live. First monthly briefing with ownership, on-call support open. The next 60 days of the retainer are on us.
What is the plan, and what does it cost?
Foundation first, then a partner who keeps up so you don't have to. The Foundation is four weeks, starting at $12,000 for the first 10 seats, $500 per additional seat, with the first 60 days of the Integrated AI Partner retainer included. The retainer is then $1,500, $2,000 or $2,500 a month by team size, month to month with 30 days' notice.
The Foundation
First 10 seats included. $500 per additional seat.
- Data and exposure audit. Every tool in use, every account, what data has gone where, and what it is costing you today. You'll have the answer to the carrier's question before they ask it.
- Approved tools and data rules, installed. Approved tools, PHI-safe configurations, an acceptable-use policy your team will actually follow because the safe path is also the fast one.
- Your first 5 to 10 workflows, built on your real accounts. Renewal proposal comparison, OE communications, RFP intake, carrier correspondence, census cleanup. You pick the ones that hurt most.
- Two days on site with your whole team. On your systems, on your live accounts. Not a webinar, not a lunch-and-learn.
- A shared prompt and skill library so the agency owns the know-how, not one person's chat history.
- First 60 days of the Integrated AI Partner retainer included. Two monthly briefings before the first invoice.
Integrated AI Partner
| Team size | Monthly |
|---|---|
| Up to 10 people | $1,500 |
| 11 to 25 people | $2,000 |
| 26 and up | $2,500 |
- Monthly briefing with ownership: what changed, what it means for a benefits agency, what we already updated for you.
- Quarterly team refresh. New hires onboarded, everyone else sharper.
- On-call support. Someone's stuck, they call us, not you.
- When the models change, we change the library. You find out in the briefing, not by discovering something stopped working.
You see two briefings before the first retainer invoice. After that it is month to month with 30 days' notice. No annual contract, nothing to unwind. Pricing is valid for engagements signed through December 31, 2026.
Each step has its own page: the Foundation and the Integrated AI Partner retainer. The full cost picture, including what is not in the number and what state training grants can take off the on-site days, is in what AI training for an insurance agency costs.
What does one workflow cost you today?
Groups renewing, times hours per renewal, times loaded hourly cost. That is what one workflow costs you today, before AI touches it. Run it on your own numbers.
Recover a third of those hours and that is 400 hours back in year one. Add your loaded hourly cost to put a dollar figure on it.
Foundation $16,000, then $2,000 a month from month three. Renewals today: 60 groups at 20 hours each, 1,200 hours a year. Recover a third of that and it is 400 hours back in year one, which at a loaded cost of $40 an hour is $16,000, the Foundation paid for. That is one workflow. The Foundation builds 5 to 10.
What results have benefits teams actually gotten?
A 5-person group benefits team left a two-day onsite with 11 production Claude Projects and now recovers 7 to 13 hours per week, with Phase 1 payback in under 9 weeks. A benefits brokerage put 10 renewal-season workflows on Claude in about six weeks, timed ahead of open enrollment.
AFC-AIS: an AI-powered back office in 30 days
- 11 production Claude Projects, one per use case
- 7 to 13 hours per week recovered across the team
- Roughly $32.5K in annualized capacity, conservatively
- Phase 1 payback in under 9 weeks
LaSalle Benefits: 10 workflows in six weeks
- 10 production Claude Projects in LaSalle's own templates
- Discovery to trained team in about six weeks
- Timed to land ahead of open enrollment
- Signed HIPAA-aware AI Governance Policy
The workflows behind those numbers are the ones a benefits team runs every week: carrier quote comparisons and renewal analysis (3 to 4 hours down to about 60 minutes), benefit booklets (2 to 3 hours down to minutes of drafting), census-to-bill reconciliation (2 plus hours per client per month down to about 20 minutes), and open enrollment campaigns (1 to 2 days down to about 2 hours). We walk through each one in the five use cases every agency builds in week one.
How is client data protected?
The rules come before the workflows. The Foundation starts with an audit of every tool and account in use and where data has gone, then installs approved tools, PHI-safe configurations, and an acceptable-use policy. A person reviews every output before it reaches a client, and nothing you provide is used to train external or shared models. You own everything your team builds.
This is the part benefits principals worry about most, and they are right to. The answer is not to avoid AI, which your competitors are not doing, but to find out where the data has already gone, put the rules in writing, and make the approved path the fast one. We go deeper on both halves of this question in client data, HIPAA and PII and in what belongs in an agency's AI governance policy.
How do we start?
We're taking on five more Foundation clients before December 31: five agencies that move from group 1 or 2 into group 3, the tools, the rules, and a partner who keeps up, before renewal season peaks and before their competitors make the same move.
The next step is a 30-minute call. Bring your client count and your team size. Leave with the exposure audit dates on the calendar, or a clear reason not to. 60 days of the retainer before the first invoice. 30 days to walk away after that. Book the 30-minute call.