A tender question, a policy query or a product to describe.
The model draws only on your approved documents and data.
Drafts arrive in your format, with the sources they used.
People check and approve before anything is sent or published.
Generative AI
Generative AI tools built around your own content.
Generative AI writes text, summarises documents and answers questions in plain language. We build tools on large language models (LLMs) that draw on your own data and follow your rules.
What you get
01Answers from your contentReplies based on your documents and records, with links to the source.
02Faster first draftsReports, proposals and replies drafted in your house style for staff to finish.
03Control over useClear rules on what the tool may say, which data it sees and who can use it.
How we deliver
Four stages, each ending with something you can see and review before the next one starts.
01
Define
We agree the job the tool must do, who will use it and what a good answer looks like.
02
Prepare
We gather, clean and organise the documents and data the tool will draw on.
03
Build
We build the tool, choose a suitable model and test answers against real questions.
04
Refine
We review usage and feedback after launch and improve answers, sources and costs.
What we buildOur generative AI capabilities
Knowledge assistants (RAG)
Assistants that search your documents first, then answer with the sources they used.
Discuss thisDocument drafting
Drafts of reports, proposals, tenders and letters built from templates and past work.
Discuss thisSummarisation
Short, accurate summaries of long reports, meetings, case files and email threads.
Discuss thisContent generation
Product descriptions, marketing drafts and translations produced at volume for review.
Discuss thisPrivate LLM hosting
Open models run in your own cloud or data centre so sensitive data stays with you.
Discuss thisGuardrails and policies
Filters and rules that block off-topic answers, unsafe content and leaks of private data.
Discuss thisAnswer quality testing
Test sets that check accuracy and tone before launch and after every change.
Discuss thisApplication integration
Generative features added inside your existing apps, portals and internal tools.
Discuss thisAllThreads.ai
A sales campaign involves research, writing, sending and scheduling, usually spread across several tools and people.
Selected work · AllThreads.aiAnimated illustration
AI sales automation: multi-agent campaign orchestration
What we built
- A coordinating AI agent that assigns each step of a campaign
- Specialised agents for research, writing, sending and scheduling
- Every step tracked through to completion
What it changed
One campaign runs from a single instruction, with every step recorded for the sales team to review.
Delivered by Hoki and its engineering partners. Product and client names belong to their owners.
Discuss a similar projectFrom the Hoki portfolio
The technology we build with
We pick tools around your existing systems, your team and how the product will be run.
Models
OpenAIAnthropic ClaudeGoogle GeminiLlamaMistralQwen
Retrieval and apps
LangChainLlamaIndexpgvectorElasticsearchQdrantFastAPI
Hosting and testing
vLLMOllamaAWS BedrockAzure OpenAILangfuseDocker
Where this applies
The same engineering, shaped by the rules and realities of each sector.
Financial services
Summarising research, drafting client reports and answering policy questions.
Healthcare
Drafting discharge notes and answering staff questions from clinical guidelines.
Media and broadcast
Drafting scripts, summarising footage logs and producing captions and metadata.
Retail and e-commerce
Writing product descriptions at scale and answering shopper questions from catalogues.
Manufacturing
Answering technician questions from manuals and drafting maintenance reports.
Public sector and education
Drafting replies to public enquiries and preparing course material for teachers.
Frequently asked questions
Will the AI make things up?
Language models can produce wrong answers. We reduce this by making the tool answer from your documents, showing its sources, testing it against real questions and telling users to check important outputs.
Will our data be used to train public models?
Not if we set it up correctly. We use business services that do not train on your data, or run open models in your own environment. We confirm the arrangement with you in writing.
Can it connect to our document systems?
In most cases, yes. We connect to common file stores, intranets and databases, and respect existing access rights so people only see answers from content they are allowed to read.
Can we start small?
Yes. A single team and one type of document is a good first step. It shows real value quickly and helps decide what to extend next.
Another question? Email hello@hoki.co
Put your own knowledge to work
Tell us which documents your team searches most. We will suggest a first generative AI tool around them.
Start a project
Tell us what you want to build.
Share a few details about your generative AI project. We will reply with questions and a suggested next step.
- What problem you want solved
- What you already have in place
- Any timing or budget you are working to
Start a project
Tell us what you want to build
A few details help us bring the right people to the first call.
Thank you.
Your enquiry is with our team. Here is what happens next.
1We read your enquiry and bring in the people best placed to help.
2We reply by email to arrange a call at a time that suits you.
3On the call, we agree a clear next step, with scope and cost.