AI & automation
Automate a defined part of your working day.
We build assistants, API integrations and processing pipelines for research, publishing and data work. We begin with the task you want to improve, the systems it touches and how you will check the result.
Choose a task whose result you can check
Extracting fields from a paper, classifying records and preparing a draft each need different checks. We look at the inputs, the expected output and what happens when the system gets something wrong. Some steps can use ordinary software rules; others may benefit from a model.
A prototype tests one workflow on representative data. Its timing is agreed after the workflow review. The result should establish the useful output, the review it still needs and an operating-cost estimate before a larger build is commissioned.
Workflows we connect
Our work includes research, editorial systems and data processing.
Research and extraction
Literature triage, structured extraction from papers and PDFs, dataset annotation and pipelines that can be rerun and evaluated.
Publishing and CMS
Draft review, translation, metadata, internal search and editorial assistants connected to an existing CMS.
Data evaluation
Normalise input, classify records and reconcile sources before a person reviews the report.
APIs and scheduled processing
Connect assistants to your tools, run scheduled jobs and assess self-hosting where data must remain in your infrastructure.
Your data, your rules
Model quality is only one risk. The design also has to account for where data is processed, who can access it, how long records are kept and which contractual terms apply.
Those decisions are written down before implementation.
Provider and tier named
The selected model, service tier and relevant data terms are recorded for the project.
Self-hosting assessed
Where data must remain inside your infrastructure, we assess an open-weights deployment alongside its hardware and operating cost.
Retention is deliberate
Prompts, inputs and outputs are retained only where needed, with access, redaction and deletion rules agreed in advance.
From workflow review to operation
- 01
Workflow review
A fixed-fee review of the work as it is done today, including the inputs, checks and steps suitable for automation.
- 02
A scoped prototype
Test one workflow with representative data. Agree its timing and success criteria after the review.
- 03
Build and handover
Deliver the documented workflow and assess hosting against the data and operating requirements.
- 04
Optional ongoing review
A cancellable scope for model changes, usage costs and quality regressions where those need monitoring.
Examples from our own tooling
Our CLI tools article follows documentation, code and data through small processing workflows. It distinguishes synthetic test fixtures from real model evaluation. The Raspberry Pi tuning report shows an assistant proposing system changes with human approval and before-and-after measurements. Both describe our own work, rather than client deployments.
The questions that matter
Which models do you use?
The model is selected against the task, data constraints, required context, latency and operating cost. The provider and service tier are named in the project documentation.
Is our data used for training?
Provider terms and account controls change, so the applicable terms are checked for the selected service and recorded before data is sent. If external processing is not acceptable, we assess a self-hosted design.
Can it run on-premises?
Yes, where an appropriate open-weights model can meet the task. Hardware, throughput, maintenance and evaluation requirements are part of that decision; self-hosting is not automatically cheaper.
How do you stop it inventing things?
No design eliminates model error. We constrain and detect it with validated structured output, retrieval against approved sources, citations and a human review step where the consequence warrants one.
What does it cost to run per month?
The prototype records usage against representative data. That gives the project an operating estimate and a ceiling that can be monitored after launch.
Can you just make our chatbot better?
Sometimes. First we check whether a conversational interface matches the task; search, a form or a background workflow may be simpler and easier to verify.
Describe the task you want to automate.
Tell us what arrives, what you need back and who checks it today. We will use that to scope the workflow review.