Case study: Semoria
From idea to production SaaS in 6 weeks
Why we built it
The story
Semoria started as bespoke tone-of-voice agents for our clients. Marketing teams across several sectors had the same problem. They had a sharp brand voice and plenty of ideas, but no capacity to publish in that voice every week.
We built custom engines that solved it one team at a time. The FPC content engine took output from 1 or 2 posts a month to 2 or more a week, without a hire.
By early 2026 the pattern had repeated so often that bespoke was clearly the wrong shape for most teams. A bespoke build cost more than most of those teams could justify. The right shape was a subscription on the same engine, with the same human control.
So in six weeks we designed, built and launched Semoria, with payments, sign-in, AI integrations, security hardening and a full marketing engine.
At launch
The build, in numbers
Every figure is from the production codebase at launch.
- 6 weeksIdea to production SaaS
- 424TypeScript files
- 73API routes
- 64Database migrations
- 47Blog posts at launch
- 27Transactional email templates
How we built it
The build
Six weeks was possible because of multi-agent orchestration, not because one AI worked harder. A lead agent split the build into independent slices and handed them to sub-agents working in parallel, each in its own isolated workspace. For example:
- One drafted a new API route
- Another refactored the billing flow
- A third wrote tests for the voice engine
- A fourth built a programmatic blog page
The lead reviewed and integrated their work. The human role shifted from writing code to setting scope, judging trade-offs and owning anything irreversible.
Where the speed comes from
Multi-agent orchestration, not single-agent generation
Single-agent AI speeds up individual tasks. Multi-agent orchestration speeds up the whole project, because independent pieces run in parallel instead of queueing. The discipline is in scoping tasks so they don't collide, isolating each workspace, and keeping a person in the lead seat. That's what makes six weeks repeatable.
Part 2 covers the full method, including what we deferred and what people still own.
What it runs on
The stack
A lean, modern stack runs Semoria for around £40 a month at 50 paying users. Each layer was chosen as a utility, not an endorsement. Part 3 has the full architecture and the costs at 0, 50, 200 and 500 users.
1 of 6
From day one
The marketing engine
The marketing engine was ready on day one, not bolted on after launch. Here's what we built alongside the product.
| Marketing channel | What we built |
|---|---|
| Programmatic blog | 47 posts at launch, MDX with FAQ frontmatter rendered as JSON-LD FAQPage schema. |
| Free tools | Voice Analyser, Post Generator, Voice Score Card. Each is a contained value unit with its own conversion funnel. |
| Persona landing pages | Programmatic SEO across niche segments with intent-aware copy. |
| Email automation | 27 templates covering trial onboarding, billing events, weekly digest, and feature discovery. |
| Analytics and funnels | PostHog funnels wired from first pageview through trial conversion to billing event. |
What we built
- Programmatic blog
- 47 posts at launch, MDX with FAQ frontmatter rendered as JSON-LD FAQPage schema.
- Free tools
- Voice Analyser, Post Generator, Voice Score Card. Each is a contained value unit with its own conversion funnel.
- Persona landing pages
- Programmatic SEO across niche segments with intent-aware copy.
- Email automation
- 27 templates covering trial onboarding, billing events, weekly digest, and feature discovery.
- Analytics and funnels
- PostHog funnels wired from first pageview through trial conversion to billing event.
Inside Semoria
The product in use
Live screenshots from semoria.io. A human reviews every output before it publishes.
Voice fingerprinting
A measurable voice signature per profile, trained on real published writing.
Research-grounded drafting
Every post grounded in recent industry research, with citations the reviewer can verify.
Series and scheduling
Generate as one-offs or 2 to 5 post mini series, then schedule across the week.
Dashboard and review
Approve, request a plain-English rewrite, or reject. Nothing publishes without a human.
Frequently asked questions
How long did Semoria take to build?
What is multi-agent orchestration in AI development?
How do you stop parallel AI agents from conflicting?
What is Semoria's tech stack?
What does it prove Agenticise can build for clients?
Is Semoria a separate company or part of Agenticise?
Ready to build yours?
Bring the platform you've been waiting to build, or the tone-of-voice problem you've been trying to solve. On a 30-minute call we'll sketch the build and the right shape for your team.