Our rates did not change when AI arrived. Our output did. At Wbcom Designs, we are delivering roughly five times the throughput we managed three years ago on certain categories of WordPress development work. The hourly rate on the invoice looks identical. The work we can actually produce in those hours has multiplied. This is the most uncomfortable conversation in the agency industry right now, and I want to have it honestly.

The Output Multiplier Is Real

Let me put specific numbers to this because I think vague claims about “AI making us more productive” do not help anyone make real decisions.

Before we integrated Claude Code, custom MCP servers, and AI-assisted testing into our development workflow in late 2023 and through 2024, a complex custom WordPress plugin typically took our team 120 to 160 hours to build from brief to production-ready. Today, comparable work takes 40 to 60 hours. The remaining time, which used to go into writing boilerplate, setting up file structures, writing repetitive CRUD patterns, and generating initial test scaffolding, is now handled by AI tools working under developer direction.

That is not a 20% efficiency gain. That is a 2x to 3x multiplier on certain categories of work. And on the automation workflows we have built for our own operations, the numbers are even more dramatic. I tracked this in detail in the n8n + MCP + Claude AI flow automation piece – nine hours of manual operations work per week reduced to under two hours of review. The same pattern applies to development work.

So Why Have Rates Not Fallen?

If output per hour has multiplied, basic economics suggests hourly rates should fall to distribute the gain to clients. Some agencies have taken this route. Competing on “same quality, lower price” is a real business strategy in a commoditizing market.

We have not taken that route, and here is why.

The Rate Was Never the Right Unit of Value

Clients do not actually want hours. They want outcomes. A client paying a WordPress agency wants a membership platform that works, a plugin that does not break on updates, a WooCommerce integration that does not lose orders at 2 AM on a Friday. The hours are the unit we used to measure effort when effort was the constraining factor.

When AI multiplies throughput, effort is no longer the constraining factor. Judgment is. The judgment about what to build, how to architect it for longevity, where the security risks are, what the user actually needs versus what they said they wanted, that judgment is still entirely human and still scarce. We are not being paid for hours; we are being paid for decisions. The rate reflects the value of decisions, not the cost of effort.

The Quality Bar Has Moved Up

The extra capacity created by AI has not gone into margin and Friday afternoons, though there is some of that. It has gone into doing work that we previously had to skip because we did not have time.

Before AI, we rarely wrote comprehensive PHPUnit tests for client plugins. Not because we did not know how, but because the time cost was prohibitive given the project budget. Now we write tests. The code we deliver is more reliable, better-documented, and more maintainable. The client is getting more value per dollar even at the same rate, because the thing we deliver is better.

We also do things like run WordPress coding standards checks automatically via our custom MCP server, run security audits on the generated code as part of the standard workflow, and validate REST API endpoints against expected schema before delivery. These quality steps used to be aspirational. Now they are default. The rate stayed the same; what it buys improved.

The Projects We Take On Have Changed

With greater throughput capacity, we have been selective about taking on projects that genuinely require our depth of WordPress expertise rather than projects we took previously because the volume made sense economically. A straightforward theme customization for a small business no longer makes sense as a Wbcom Designs project when a Tier 1 developer with good AI tools can handle it competently at a third of our rate.

We focus on work that requires what we actually have: deep BuddyPress and online community expertise, complex plugin architecture, multisite environments, WooCommerce customization at scale. The same hourly rate, applied to higher-complexity work, delivers more client value than it did when we were taking everything.

The Honest Tension: When Should Clients See Rate Reduction?

I want to be honest about where the argument for rate reduction is legitimate, because I think agencies that pretend there is no tension here are being evasive.

If an agency is billing 10 hours for work that AI has genuinely reduced to 3 hours, and the work is the same work at the same quality, the client is paying for 7 hours that do not exist. That is not the same as billing at a rate that reflects judgment and expertise. That is billing for phantom effort.

The honest version of the AI-era agency conversation is this:

AI has changed how long things take. It has not changed what good work requires. When you pay our rate, you are paying for architecture decisions, security judgment, WordPress platform expertise, and accountability for the outcome. You are not paying for hours of typing. If the project would have taken 100 hours before and takes 40 hours now, the honest thing is to scope and price for 40 hours of the right kind of work, not to keep billing 100.

This requires moving away from hourly billing for many categories of work, which is a significant structural shift for agencies that have built their entire financial model on time and materials contracts.

Same rates, 5x output is not a trick. It is what happens when the bottleneck shifts from effort to judgment. Clients pay for the judgment. AI absorbed the effort.
Same rates, 5x output is not a trick. It is what happens when the bottleneck shifts from effort to judgment. Clients pay for the judgment. AI absorbed the effort.

The Move Toward Value-Based and Outcome-Based Pricing

The logical endpoint of AI-multiplied agency output is value-based pricing. Not “how long will this take” but “what is this outcome worth to the client.”

We have been moving in this direction for two years. For some project categories, we now quote fixed project prices based on scope rather than estimated hours. The client knows exactly what they will pay. We manage the risk of the estimate. If AI tools let us deliver in 30 hours what we priced for 60, the extra capacity is ours to use on quality or capacity for the next project. If something takes longer, we absorb that too.

This is a better model for clients. Predictable cost. Aligned incentives. Agencies that invest in efficiency keep the benefit of that efficiency rather than passing it through as margin compression.

It is also a better model for agencies that have genuinely invested in AI-assisted workflows. An agency billing by the hour for AI-assisted work is either billing phantom hours or, if they pass the efficiency to the client, racing to the bottom on rates. Neither is a sustainable business. Outcome pricing decouples revenue from time and lets the investment in AI tooling show up as profit rather than rate pressure.

What the 5x Output Number Actually Means in Practice

I want to be specific about what “5x output” means in the kinds of work where we see it most, and where we do not see it.

Work CategoryAI Throughput MultiplierWhere Humans Still Dominate
Custom plugin boilerplate and CRUD4x to 5xArchitecture decisions, security review
REST API endpoint development3x to 4xPermission logic, schema design
PHPUnit test writing3x to 4xTest specification, edge case identification
Settings page UI (Settings API)4x to 5xUX decisions, validation logic
Documentation and changelogs4x to 5xAccuracy verification, tone editing
Complex debugging1.2x to 1.5xRoot cause identification, pattern recognition
Architecture planning1x (no multiplier)Entirely human judgment territory
Client requirement discovery1x (no multiplier)Human communication, domain knowledge

The 5x number is real for well-defined, pattern-heavy tasks. It is not a universal multiplier. Complex debugging is only modestly faster. Architecture work has not changed. The human-judgment-intensive work remains human-judgment-intensive. The AI multiplier applies to the execution layer, not the thinking layer.

This actually makes sense as a business model. The thinking layer is where agency expertise lives. The execution layer was always about converting thinking into code. If AI can do the conversion faster without compromising the quality, the agency that owns the thinking layer captures the benefit.

The Competitive Landscape Is Bifurcating

In 2026, the WordPress agency market is bifurcating in a specific way that I did not anticipate clearly two years ago.

At the commodity end, AI-native freelancers and micro-agencies with excellent AI workflows are genuinely competitive with agencies that cost five to ten times more for straightforward work. A solo developer with good Claude Code skills and a structured MCP workflow can build competent WordPress plugins, complete sites, and reasonable custom functionality at a price point that full-service agencies cannot touch. And they are delivering adequate quality for a large portion of the market.

At the expertise end, the premium for genuine depth has increased. Complex problems, long-term maintenance responsibility, enterprise requirements, integrations across multiple systems, these all require the kind of accumulated judgment that cannot be shortcut by AI tools. Clients with complex needs are willing to pay more for confidence that the team understands what they are doing, not just that they have good AI tooling.

The agencies that are struggling are the ones in the middle: charging agency prices for work that does not require agency expertise, and not differentiated enough to compete on depth. AI has compressed the middle of the market. You either genuinely compete on expertise and judgment (in which case, same rates with better output is a winning proposition), or you compete on price and efficiency (in which case, you need AI tools and low overhead, not a full agency structure).

The Transparency Question: What to Tell Clients

One practical question that comes up in agency circles is whether to disclose to clients that AI tools are involved in project delivery. My position is that disclosure is not just ethical, it is strategically smart.

Clients who understand how AI is being used appropriately, including what human judgment is applied at each stage, are better clients. They understand why some work is faster than it used to be and why that does not mean quality has dropped. They understand that when we say “this will take two weeks,” that reflects real complexity, not time padding. They are less susceptible to being poached by agencies promising dramatically lower costs because they understand what those lower costs are actually trading off.

The agency that hides its AI usage and bills phantom hours is taking a short-term tactical position that creates long-term credibility risk. When clients inevitably start to understand the AI landscape, they will revisit those billing patterns. Being ahead of that conversation is better than being caught behind it.

For the clients we work with on ongoing retainers, we have had explicit conversations about how our workflow has changed and what that means for their projects. The response has been positive. Clients who trust their agency’s judgment about technology are an asset. Treating them as intelligent adults about the AI question is part of building that trust.

What Clients Should Be Asking About AI in 2026

If you are a WordPress business evaluating development partners in 2026, here are the questions I would ask any agency about their AI practices:

  • Do you have code review gates on AI-generated code? Agencies that accept AI output uncritically are shipping security risks. The ones who have built review workflows get the throughput benefit without the quality tradeoff.
  • How do you handle AI hallucination of WordPress functions? This is a real problem. AI sometimes generates calls to WordPress functions that do not exist or that were deprecated. Agencies without verification steps ship these errors.
  • Has your pricing model adapted to AI-era productivity? An agency still charging for hours of work that AI has dramatically accelerated either has not adopted AI seriously or is billing phantom hours. Neither is good for the client relationship.
  • What categories of work do you still do manually? An honest answer here reveals what the agency actually understands about their own workflow. Security review, architecture, and complex debugging should still be deeply human activities. If everything is “AI-assisted,” that is a flag.

These questions will produce very different answers from agencies at different points in the AI adoption curve, and the differences are instructive. Agencies that have thought carefully about this will give specific, honest answers. Agencies that have not will give vague ones about being “AI-forward” without specifics.

The Retainer Model Rethink

Traditional monthly retainers were priced around a predictable number of development hours. When AI tools change how many hours a task actually takes, retainer models need to be rethought. A retainer priced for 40 development hours a month that now produces the equivalent of 80 to 120 hours of pre-AI work creates a genuine question about where the efficiency benefit should go. The answer is not simply to charge more or pocket the difference. It is to redesign the contract model so both sides are clear on what they are getting and why.

At Wbcom, we have restructured retainers over the past 18 months to be scope-based rather than hour-based where the client relationship allows it. Instead of selling a block of hours, we define what will be delivered each month: plugin updates maintained to a specific quality standard, a defined set of feature additions, a quality bar for the code that gets delivered. The client knows what they are getting. We manage the efficiency gains internally and use them to invest in quality rather than inflate hour counts that no longer exist at the pre-AI level.

This model removes the implicit incentive to work slowly to fill hours and creates a natural conversation about value rather than time. When we propose renewing a retainer, the conversation is about what got delivered and what the client needs next, not about whether they used all their hours. The transition takes time and trust, but it consistently produces deeper client relationships and more predictable revenue than hourly billing ever did. The agencies that navigate this transition successfully are also better positioned when clients start asking pointed questions about AI and pricing, because they have already structured the answer into how they work.

Building the Workflow Infrastructure

The productivity gains from AI tools require investment in internal tooling and process. An agency that hands Claude Code to a developer without any supporting infrastructure will see modest gains and significant quality risk. At Wbcom, what drives the output multiplier is the combination of custom MCP servers for WordPress built for our specific workflows, automated testing pipelines on every commit, documented patterns for common project types, and code review processes that ensure AI-generated output gets proper human verification before delivery.

Building this infrastructure took deliberate effort over 18 months. The first phase was identifying where AI tools added real value and where they did not. The second phase was systematizing the workflows that proved out. The third phase, where we are now, is incremental refinement based on live client work. Each phase built on the last. The accumulated result is a development environment that is genuinely faster than what we ran two years ago without trading quality for speed.

The agencies that will sustain the output advantage over time are the ones who invest in this infrastructure seriously rather than treating AI tools as a drop-in booster. The investment competes with billable hours in the short term, but it compounds in ways that are hard to replicate quickly. First movers who have built this infrastructure have a structural advantage that is not purely about tool access. That advantage shows up in delivery speed, code quality, and the ability to take on more complex projects without adding headcount.

The Long-Term Trajectory

Same rates, 5x output is a temporary equilibrium. The market will reprice over time. Some of the output gains will flow to clients through lower fixed project prices for well-defined work. Some will flow to agencies as margin for reinvestment in tools and people. Some will compress commodity rates as AI-native operators enter the market.

The direction is clear: the economic unit of agency work is shifting from hours to outcomes. The agencies that figure this out early, and build the pricing models and project structures to match, will be in a better position than those that keep billing time and materials on work that AI has made dramatically faster.

At Wbcom, we are not all the way through this transition. We have parts of our client work on fixed-scope project pricing and parts still on time and materials. The transition is ongoing. What I am confident about is the direction: the agencies that survive and thrive through this transition are the ones who own genuine expertise, deliver measurable outcomes, and price accordingly. The ones who try to preserve the old hourly model while using AI to inflate their margins are building toward a credibility problem with clients who will eventually notice the discrepancy.

Same rates, 5x output is an honest description of a transitional moment. The endpoint is outcome-based pricing that reflects what AI-era development actually delivers. Getting there requires honesty about what the agency is actually providing and what part of that value AI has affected. That is a conversation worth having openly.

For more on how the specific tools driving this productivity shift fit together in practice, the agentic coding discussion covers where the limits are and why human oversight of AI workflows is not a bottleneck but a feature. The agencies building durable practices are the ones treating AI as an accelerant for human judgment, not a replacement for it.

Thinking About Agency Partnerships?

If you are evaluating WordPress development partners and want to understand exactly what AI-augmented development looks like in practice, including what it means for your project scope, timeline, and cost, Wbcom Designs is happy to walk through the specifics with you. We prefer honest conversations to polished proposals.