Development agency / Assistant, Coder & model management
Ten models. One team. Control and independence.
AITJE set up a shared AI environment for a development agency, featuring Assistent, Coder and ten models to choose from. A shared token pool of €350 per month replaces nine subscriptions to €200. Model usage and token costs per task can be tracked via Asana.
The request from the development agency
AI for the team.
Control over usage.
The developers used AI throughout the working day, but separate subscriptions made its use expensive and fragmented. Nine subscriptions from €200 amounted to €1,800 per month. Not every task requires the same model, whilst a technical fault or usage limit can interrupt work with a single provider.
AITJE Installed Assistent and Coder on an BOSGAME M5 with an additional server. A linked LLM management system brings together local models and external model APIs. Developers can choose from ten models and work within a single shared API budget. This brings project knowledge, code and model usage together into a single workflow.
From separate accounts to a single pot
The team shares the capacity. With current usage, there is often budget left over.
The task determines the model
A team of models.
One way of working.
DeepSeek, Claude, Codex and smaller open-source coding models are available within the same working environment. Developers can choose the option that best suits the assignment. The LLM management system links these choices to the shared budget and the local computing environment.
AITJE To this end, they built an agent environment using OpenCode and OpenRouter. The concept is similar to model orchestration, as seen in Sakana Fugu: deploying different models as a collaborative pool. For this consultancy, this approach has been tailored to its own projects, documentation and working methods.
10 model choices within a single managed environment.
From consultation to implementation
The assistant thinks along with you.
Coder works with it.
The chat also serves as preparation for the coding work. The Assistant helps to formulate a question and gathers the project context. The coder is provided with the relevant skills and documentation, so that the agent knows how this agency builds its solutions.
How does this new feature fit into the architecture of our project?
The chatbot uses the project documentation and agreements to analyse the query and prepare context for the coding agent.
A detailed brief for Coder, with relevant background information from the same project environment.
The connection between the components
Same context.
From enquiry to task.
- 01
Discuss
The Assistant helps the developer to think through the question and prepares the context for each project.
- 02
Provide context
Skills, documentation and codebase conventions provide Coder with an understanding of the agency’s working methods.
- 03
Deploying the model
Model management connects OpenCode to local models and external APIs via OpenRouter. The task determines which route is appropriate.
- 04
Review costs
The integration with Asana shows which model was used and how many tokens the execution cost.
The integration with Asana
A task completed.
How much did AI cost?
With individual chat accounts, the costs are hidden behind a monthly subscription. In this environment, model usage can be traced back to the task in Asana: which model was used, how many tokens were processed and what did it cost to complete the task?
This enables the agency to adjust its choice of models based on its own work. Specialised local models take over suitable coding tasks; external models remain available for tasks where they offer added value. Available discounts, bonus credits and usage limits are taken into account when managing model access.
- Project & task
- The assignment in the repository
- Model used
- The chosen local or external approach
- Token usage
- The processed input and output
- Model costs
- Back to this task
Which data is collated for the task; no shared client file.
A local base for both
‘Internet’ has been omitted.
The coding doesn’t have to stop.
The M5 ‘BOSGAME’ and server make the local models available to both the Assistant and the Coder. Codebase documentation and skills remain usable within the user’s own environment. This means that even if the internet connection fails, a developer can continue working using a local coding model or ask the Assistant for help.
The full model pool
Both local models and external APIs are available. Asana and other online integrations are accessible. In the event of an outage or a limit at one provider, another configured model route remains available, provided it is accessible.
The difference in model budget
€1,800 subscriptions.
€350 for the whole team.
The agency has replaced the nine separate subscriptions with a shared token pool of €350 per month. The developers use various models throughout the working day. Given current usage, that budget is often not fully utilised.
The benefit arises from the combination of specialised local models, a targeted project context and managed API usage. This means that even the most complex external model does not need to handle every step.
€1,450 less per monthApproximately 81% lower model budget
Comparison of the previous subscription fee with the API budget set for this agency. Hardware, servers, electricity, setup and management are excluded from this. Local processing does not involve an external token account; paid model calls are funded from the shared pot.
It’s the collaboration that makes the difference
Chat, build and manage.
It all goes hand in hand.
This case brings together the ‘AITJE’ products: ‘Assistent’ for discussions and preparation, ‘Coder’ for work on the codebase, and ‘modelbeheer’ for model selection, availability and costs. The development agency retains the flexibility to use new models, whilst having its own local infrastructure to fall back on.
Other case studies
Marketing / customer management, tickets & business knowledge
350 clients. One central workspace.
Following a takeover, a marketing company’s client base grew from 150 to 350. AITJE brought together client data, tickets, projects, invoicing information and call transcripts in a central hub. Staff can search for and chat with their business experts, with access based on their role and AI integrated into an existing Mac mini.
View the approach
Marketing agency / level design & in-house coding agency
25 New game levels every week. With no external token costs.
A marketing agency took over an existing mobile puzzle game but ran into difficulties when creating new levels. AITJE analysed the game, built a reusable level design system and linked AITJE Coder to Qwen on its own server. Now, the agency 25 creates new levels every week from a single assignment in OpenCode.
View the approach
Power leads / 3D catalogue & workflow optimisation
From technical drawing to 3D product model.
AITJE built a workflow for a partner specialising in contact cables that converts product data and technical drawings from WordPress into 3D models. AITJE Coder controls Blender; Three.js makes the GLB files usable on the website. Following optimisation, API costs fell from approximately €9 to €0.11–€0.18 per product.
View the approachMOVING FORWARD TOGETHER WITH AI
Your next step begins with a question.
Got an idea, facing a challenge, or simply curious? Tell us what you want to achieve. AITJE will work with you to find a solution.
