Comparison
Serviceplan Agents compared to ChatGPT
This is not an argument that general assistants are bad. It is a description of the four things that change when the agent is built for one job, by people who have done that job commercially since 1970.
The honest version
ChatGPT and Claude are excellent general assistants, and the Serviceplan Agents run on the latest generation of Claude models themselves. If your question is well-posed and the answer lives in public text, a general assistant will handle it, and you should use one.
The difference shows up on the work that has a commercial consequence: research you will present, a plan someone will be held to, a number that will be challenged. There, four things matter.
The four differences
- Licensed data
- Statista, GWI, DataForSEO, press agencies and social APIs are included in the plan. A general assistant is limited to what it can reach publicly, which is exactly where market and audience data is thinnest.
- Persistent memory
- The agents keep an active memory of your business across requests and improve on your feedback. A general assistant starts most sessions not knowing who you are.
- Finished documents
- PDF, PowerPoint, Excel or an interactive dashboard, delivered to your inbox. Not chat text you reformat yourself.
- A point of view
- Hannah will not make weak findings look convincing, and Elena will not let an unrealistic timeline pass. A general assistant is trained to be agreeable, which is the wrong instinct for research.
Also different
Access without a login
Requests go by email, WhatsApp or Microsoft Teams. Nothing to roll out, nobody to onboard.
A shared task board
Work and files sit on one board in Sokosumi, so a request is not trapped in one person's chat history.
Active project management
Programmes, projects, work packages and tasks, rather than a single thread that grows until it is unusable.
Human escalation
Requests route to real people when they genuinely need them.
Where a general assistant is the better tool
Drafting, rewriting, summarising something you already have, thinking out loud, code. General assistants are faster and cheaper for all of it, and there is no reason to route that work through a specialist agent.
The line is roughly this: if the output is for you, use a general assistant. If the output has to convince somebody else, the sourcing and the point of view start to matter.
Questions
Which models do the Serviceplan Agents use?
The latest generation of Claude models, including on the free plan. The model is not the differentiator — the data access, the memory, the output format and the domain judgement are.
Can I use both?
Most people do, and that is the sensible arrangement. Use a general assistant for drafting and thinking; send the research that has to hold up to Hannah.
What does it cost compared to a ChatGPT subscription?
The free plan is 0 EUR with 200 credits a month. Starter is 25 EUR for 1,500 credits, Standard 75 EUR for 5,000, Pro 200 EUR for 15,000. A single competitive analysis costs under 20 EUR in credits.
Look closer
Compare the output yourself
Run the same competitive analysis question through both. Enter your URL and Hannah's version comes back free.