Smeron
All Work
AI marketing automation

OSMO

  • AI integration
  • Microservices
  • Marketing tech
OSMO — AI marketing automation

A marketing automation suite that centralises online marketing workflows across social networks, search engines and direct outreach.

Who it was built for

The business, and what they came with.

Built for

OSMO

Sells one marketing platform to owners and agencies currently paying for five: content, scheduling, outreach, SEO and paid ads.

The brief

Consolidation only sells if each engine stands on its own. Every module had to be strong enough that cancelling the incumbent tool is an easy decision rather than a compromise.

What we delivered

What OSMO has now that they did not have before.

  • Five engines, one dashboard

    Content, calendar, outreach, SEO and ad management behind a single navigation, so a campaign does not get assembled across four tabs and a spreadsheet.

  • AI that produces publishable work

    GPT-4o through Azure for copy and Runware for imagery, wired into the editor rather than parked in a separate playground.

  • Scheduling that does not stall

    Redis-backed queues run posting and outreach asynchronously, so a thousand queued actions do not make the dashboard feel broken.

  • Numbers in the same place as the work

    Campaign performance sits beside the calendar and the editor, which is where decisions about next week actually get made.

Where the growth comes from

Why each piece of the build pays for itself.

  • One subscription replaces several

    The pitch is a smaller stack and one invoice, and that comparison only holds if nothing has to be kept alongside it.

  • Agencies arrive with accounts attached

    Multi-client management makes a single signup worth many, and makes leaving expensive.

  • The work stays inside the product

    Editor, calendar and reporting in one place is what turns a trial into the tool a team opens first thing.

The business analysis

Market, model and architecture. Every figure is either cited or labelled as a model.

Consolidation plays are the hardest kind of software to sell, because the customer has to be persuaded to cancel five things they already understand. OSMO's pitch is one platform and one invoice in place of content, scheduling, outreach, SEO and ad tools bought separately. That comparison only survives contact with a buyer if every module is good enough to replace its incumbent outright — because a consolidation that leaves two subscriptions running has not consolidated anything, it has added a sixth.

Average SaaS applications per company
130 → 112
The first recorded decline in over a decade, from the 2022 peak to 2023 — a 14% fall. Consolidation is a real buying behaviour, not a pitch.

Source · BetterCloud State of SaaSOps, via PR Newswire

SaaS applications per business, later tracking
144
A different tracker, counting differently, and rising again. Sprawl figures vary enormously by methodology.

Source · Vertice, company tech stack size

Engines behind one navigation
5
Content, calendar, outreach, SEO and ad management. Each has to stand alone for the bundle to make sense.

Model · Counted from the delivered scope in this case study.

Runtimes, deliberately
2
Node for platform APIs, Python for the AI services. A microservice split that follows the language each job is best written in.

Model · Architectural fact, from the build described in this case study.

Sprawl is real, and the numbers describing it are unreliable

Every consolidation pitch cites a SaaS sprawl statistic, and almost every one of those statistics is measuring something slightly different. BetterCloud recorded the first decline in over a decade in application counts, from a 2022 peak of 130 per company to 112 in 2023 — a 14% fall 1. Other trackers, counting licences rather than sanctioned applications, report figures well above that and rising 2.

The stack OSMO is asking a buyer to cancel

The five categories named in this case study, ranked by how hard each incumbent is to displace. The rankings are ours and reflect switching cost, not product quality.

Ad management
9 — hardest

Spend history, conversion tracking and optimisation data all live in the incumbent. Migrating loses the learning.

SEO tooling
7

Historical ranking and backlink data is the moat. A new tool starts with no history.

Outreach
6

Sequences and sender reputation are portable in principle, painful in practice.

Content creation
4

Little accumulated state. The easiest beachhead, which is why the AI editor matters.

Scheduling and calendar
3 — easiest

Almost no switching cost beyond reconnecting accounts.

Model · Our assessment of switching cost per category, based on how much accumulated customer data each incumbent holds. Judgement, not survey data.

The two revenue mechanics are not equal

One subscription replacing several is the pitch. Multi-client agency management is the business. Those are different mechanics with different economics, and confusing them is how consolidation products end up with high churn and a good demo.

MechanicWhat it changesEconomic character
One platform instead of fiveA smaller stack and one invoiceWins the trial, loses to any gap in module quality
Agency multi-client managementOne signup carries many client accountsThe compounding one — expands with the agency and is expensive to unwind
Work happening inside the productEditor, calendar and reporting in one placeRetention, because leaving means rebuilding a workflow

Where the durable revenue sits

Our apportionment of expected account value across the three mechanics, weighted for expansion and switching cost rather than for headline appeal.

55%of durable value comes from agency accounts
Agency multi-client accounts55%
Each new client the agency wins is expansion revenue with no new sale attached.
Workflow lock-in25%
Calendars, drafts and reporting history accumulated in one place.
Single-business subscriptions20%
The volume segment. Lowest expansion, highest churn.

Model · Our weighting of the three revenue mechanics documented in this case study by expansion potential and switching cost. Illustrative — no customer data was shared with us.

Where the architecture earns its complexity

Five engines behind one navigation is a lot of surface area, and the temptation is a single application that does all of it. The build resists that in two specific places, and both are commercial rather than aesthetic decisions.

The stack, by responsibility

The two deliberate splits: AI services in their own runtime, and two data stores chosen for two different shapes of data.

  1. Application shell

    One navigation over five engines, with the editor and calendar treated as first-class surfaces rather than as modals.

    • Next.js 16
    • React 19
    • TypeScript
    • Tailwind CSS v4
  2. Working surfaces

    Where the work actually happens, which is why these libraries are named: the editor and the calendar are the product for most users.

    • Tiptap
    • FullCalendar
    • ApexCharts
    • React Query
    • Redux
  3. Platform services

    Accounts, campaigns, channels, billing. Conventional API work, in the runtime best suited to it.

    • Node.js
    • Express
    • Socket.io
  4. AI services

    Separated into Python because model and data tooling lives there. Also isolates the slowest, most failure-prone calls from the platform API.

    • Python
    • FastAPI
    • Azure AI GPT-4o
    • Runware
  5. Data and queues

    Two stores on purpose: relational for accounts and billing, document for generated content and campaign artefacts. Redis carries the queues.

    • PostgreSQL
    • MongoDB
    • Redis
  6. Channels and payments

    Every outbound surface, and the two payment processors that make the product sellable in more than one market.

    • LinkedIn
    • Meta
    • Google Ads
    • X
    • Stripe
    • Razorpay

One campaign, across five engines

The point of the product is that this is one path rather than four tabs and a spreadsheet. The offshoots are where multi-channel automation actually breaks.

  1. Draft created in the editor

    Copy generated through Azure AI and imagery through Runware, inside the editor rather than in a separate playground.

  2. Scheduled on the calendar

    The same artefact placed across channels. One object, several destinations, which is the part a five-tool stack cannot do.

  3. Queued for dispatchdecision

    Posting and outreach run asynchronously through Redis-backed workers, so volume does not make the dashboard feel broken.

    • ↳ Channel rate limit → retried on backoff, not failed
    • ↳ Token expired → account flagged, remaining channels still go out
  4. Outreach sequences run

    Direct contact handled by the same platform that made the asset, so replies land next to the campaign rather than in a separate inbox.

  5. Performance beside the work

    Results sit next to the calendar and the editor, which is where decisions about next week are actually made.

What we would watch

RiskWhy it bitesEarly indicator
The weakest module sets the priceA consolidation bundle is judged on its worst engine. One thin module and the buyer keeps an incumbent, which invalidates the whole pitchCustomers using four of five engines and still paying a competitor
Channel API dependencyFive outbound platforms, each of which changes its API and its policy on its own scheduleEngineering time spent on channel maintenance outgrowing time spent on features
AI cost per active accountGeneration is a variable cost that scales with usage, and heavy agency accounts are the most valuable and the most expensiveGross margin falling as agency seat count rises

References

  1. 1.The number of SaaS applications at companies declined for the first time in over a decade · BetterCloud State of SaaSOps, via PR Newswire
  2. 2.Company tech stack size — SaaS applications per business · Vertice

Who it is for

The people whose problem this solves, and what they came for.

  • Owners and founders

    Want reach and leads without hiring a marketing team to run five tools.

  • Agencies

    Manage many client accounts and automate the repetitive half of the job.

  • Creators and SEO specialists

    Need blogging, backlink tracking and scheduling in one workflow.

Reference

For the technical reader. Everyone else has what they need above.

Recommended case studies

The closest work to OSMO, scored for relevance rather than picked by position.

Want the version of this built for your business?

Fifteen minutes with the people who shipped it. No deck, no account manager.