An agency quotes you eight weeks to rebuild your marketing site. The number sounds reasonable because it always has. That's the problem.
The site you're reading was rebuilt by the founder of this firm, who has never written a line of production web code in his life. The work took roughly twenty hours of focused attention, spread across five sessions, in eleven calendar days. The git history is the receipt: seventy-plus commits during that window, all visible, all timestamped.
We're sharing this for a reason that has very little to do with websites.
If you ask an agency to deliver a similar-scope marketing site for an SMB or mid-market company, the quote you receive in May 2026 is still routinely six to ten weeks. Two months. Sometimes three, once revision rounds get added in. And the company paying that invoice signs it off, because that's what this kind of work has always cost in time.
The number you see, and the number you don't
The invoice is the visible cost. The standard the invoice teaches your company is the invisible one, and it's vastly more expensive.
When your marketing team accepts eight weeks for a website, your operations team accepts six weeks for a process redesign. Your finance team accepts a month for a board deck. Your sales team accepts two weeks for a competitive teardown that an analyst with the right tooling produces in an afternoon. Every timeline in your operation is benchmarked against pre-AI labor curves, and the benchmark gets re-affirmed every time a vendor sends a quote and nobody pushes back.
This isn't a complaint about agencies. Many of them are excellent at what they do. The point is that the ceiling for what an informed operator can ship in a working week has moved, and most companies haven't repriced anything against the new ceiling. They are still buying from a menu whose prices were set in 2022.
What the twenty-hour build actually proves
It would be easy to read the twenty-hour figure as a story about the website. It isn't. The website is a stand-in for every recurring deliverable a company outsources or staffs against pre-AI assumptions.
The compression is sharpest in work that has three traits: clear goal, lots of small decisions, easy-to-verify output. Most knowledge-work deliverables sit squarely inside those three traits. Here are the categories where companies are most overpaying for vendor time right now:
- Marketing collateral. Landing pages, sales decks, one-pagers, case studies, email sequences. Eight-week vendor cycles compress to under a week when someone who knows the business runs the build with AI tooling.
- Internal reporting and dashboards. Monthly board packs, ops dashboards, sales pipeline reports. Work that used to require a junior analyst for three days now takes ninety minutes, with cleaner reasoning attached.
- Sales enablement. Competitor teardowns, account research briefs, discovery questionnaires, proposal drafts. Tasks priced at five-figure annual retainers for a junior researcher are now sub-hour outputs.
- Light internal tooling. Forms, intake flows, scheduling glue, small data pipelines. Six-week IT tickets that an operator with the right tools ships in an afternoon, and ships better, because the operator understands the actual workflow.
None of this means the work becomes free. It means the price and the timeline are no longer set by labor hours. They're set by how clearly the company can think about what it wants. That is a very different skill from procurement, and most companies have not yet realized they need it.
Why the standard holds anyway
The honest reason companies still accept long timelines is that the alternative requires changing how they buy work, and that change makes people uncomfortable.
It means fewer vendors and more internal capability. It means smaller projects with shorter cycles. It means someone inside the company has to actually know the business well enough to direct the tooling, instead of writing a brief and waiting. The org chart, the budget categories, and the approval ladders are all calibrated against the old way. Repricing one project against what's now possible is easy. Repricing how a hundred-person company runs is the work.
That's the gap we sit in with the leaders we work with. Helping a company recalibrate its baseline for what a week of work looks like, so the operating model stops paying 2022 prices for 2026 outputs.
The uncomfortable question
Pick one project your company is about to sign off on this quarter. A campaign, a redesign, a piece of internal tooling, a recurring report. Look at the quote, look at the timeline, and ask yourself honestly what it would take, and how long it would take, if someone who actually knows your business sat down with AI tooling and built it themselves.
If your answer is "weeks faster and a fraction of the cost," that's one data point. The harder follow-up is: how many of the other line items in this year's budget were priced on the same outdated curve, and what is your company quietly paying, in time and in standards, to keep that curve in place?
The two-month website isn't the cost. The two-month baseline is.