ContentHomie
Challenge
Most B2B companies know that showing up on LinkedIn works, and most of them still post nothing. The reason is not that people cannot write — it is that having something worth saying requires knowing what is being discussed in your industry this week, what your buyers actually care about, and what has already been said better by someone else. That research is the expensive part, and it has to happen again every time.
The alternative that most tools offer is worse: generic AI output that reads like generic AI output. It costs less time and it costs more credibility.
Solution
ContentHomie is a LinkedIn content engine we built together with Outbound Catalyst, who brought the outbound and go-to-market side while we built the product.
The interesting part is not the writing step, it is everything that happens before it.
Research before generation
Rather than prompting a model for “a post about X”, the pipeline first goes and finds out what X should be. Several discovery tracks run in parallel:
- Industry research across the open web, looking for trends, news and live discussions in the brand’s domain
- Network analysis of the LinkedIn pages the brand already follows — what those accounts are discussing, which themes recur, what is getting engagement
- Direct search through the LinkedIn API for high-engagement posts and conversations on adjacent topics
Candidate topics are then filtered rather than accepted: is there enough substance available to say something real, does it align with what the brand actually does, and is there a gap where a distinct perspective would land.
Analysis, then drafting
For each selected topic the system assembles an inspiration package — statistics and facts, excerpts from posts that performed well, angles from web research, and relevant material from the brand’s own content sources. Only then does drafting happen, against that package and against the writer’s own voice, so the output has something specific in it rather than the average of everything ever written on the subject.
Deep research, curation, then creation — how the loop is described on contenthomie.com.
Results
- A working product — Django and SvelteKit — running the full loop from research through to drafted posts
- Multi-source topic discovery combining web research, LinkedIn network analysis and API search, rather than a single prompt
- Built jointly with Outbound Catalyst, combining their go-to-market expertise with our engineering
We never took ContentHomie to market — commercialising it properly needed more focus than either side could give it. We do still use it: for our own LinkedIn presence, and for clients. That is the honest status, and it is also the most useful thing we can say about it. It is a tool we rely on ourselves rather than a demo we built and walked away from.