Answers
Find the answer before you ask for it.
Every question we are asked about Caesar®, from where it runs to who signs the contract. Search it, or read a category.
Caesar® and Sovereign AI
What is Sovereign AI?
Sovereign AI means deciding where information is stored and processed, who can use it and whether a request may reach an outside model. You choose the environment for Caesar® and set those rules before the work begins.
What is Caesar®?
Caesar® is HAQQ’s AI engine for governments and enterprises. It works with your organization’s records, instructions and software: it retrieves the right material, coordinates the tools and uses approved models to research, compare, draft and carry out permitted actions. Your teams get work they can check, in a deployment you control.
Who is Caesar® for?
Governments and enterprises in sovereign industries: ministries, regulators, banks, public authorities and enterprises, across 25 industries.
Which industries does Caesar® serve?
Twenty-five: Law, Justice & Compliance; Defense & Military; Intelligence & National Security; Public Safety & Emergency Services; Government & Public Administration; Banking & Financial Services; Insurance & Pensions; Healthcare & Public Health; Pharmaceuticals, Biotechnology & Life Sciences; Energy; Water & Sanitation; Food & Agriculture; Mining & Critical Minerals; Environment & Natural Resources; Industry & Manufacturing; Real Estate & Construction; Transport & Logistics; Trade & Commerce; Telecommunications & Connectivity; Digital Infrastructure & Cybersecurity; AI, Software & Semiconductors; Aerospace & Space; Education, Science & Research; Labor & Social Protection; Media, Culture & Heritage.
How is Caesar® related to Justinian®?
Justinian® is Generation 1, the legal generation: it powers HAQQ’s legal products. Caesar® is Generation 2: the same architectural foundation, extended beyond law to the records, procedures and work of other sovereign industries.
Why not use a general AI model directly?
An AI model supplies reasoning and generation. Putting it to work in your institution also requires the right records, permissions, tools, workflow history and review steps. Caesar® supplies that system around the model. You can build those layers yourself; HAQQ offers an engine you can deploy and configure around your work.
We already have cloud infrastructure and access to AI models. What does Caesar® add?
Neither one knows your records, your procedures or the way your teams work. Caesar® adds that layer: it connects your institution’s knowledge, brings the right context to every task and carries the work through your workflows, under your rules. The selected model still reasons over the material and writes the result; Caesar® supplies everything around it.
What does “AI firewall” mean?
Caesar® sits between your institution and the AI models it uses. Every request passes through it first: it applies your access rules, anonymizes what must not leave, routes each part to a model you approve, checks the answers and records every call for audit.
How does Caesar® handle a request, step by step?
Seven steps. 1. The request arrives on your infrastructure. 2. Caesar® reads the material locally. 3. It extracts what matters. 4. It anonymizes on three levels: personal data, context and intent. 5. It asks approved frontier models, in parallel, only what they need. 6. It compares and verifies the answers and records every call for audit. 7. It puts the answer back in your context, and your team makes the call.
What are the ten decisions behind a request?
Clarify the request; identify the intent; apply the language and governing rules; resolve the entities; recover the working context; select the tools; route the model; set the reasoning depth; enforce the scope; maintain continuity.
What is the ontology?
The structured map of the people, organizations, assets and documents in your records, and how they relate. A supplier may appear in a contract, an invoice and an inspection report; linking those records lets Caesar® retrieve the history behind a question about that supplier. Your deployment defines what belongs in it.
What are the five layers of Caesar®?
Applications (what people use), Data (laws, regulations and your own documents, structured for the engine), Intelligence (checks, reasons and governs every request), Hardware (where data and compute sit) and Energy (the power behind every deployment). The API and MCP connect your systems to the engine.
What does the Energy layer mean?
No dependency on someone else’s data centre: Caesar® runs on the power of the site you choose.
Is Caesar® an AI model? Does HAQQ train a model on our data?
Caesar® is the engine around the models. The selected model reasons over the material and writes the result; Caesar® supplies the retrieval, integrations, permissions and review logic around it. HAQQ’s architecture excludes customer data from training HAQQ models.
How is Caesar® different from the AI assistant that comes with our office software?
An assistant inside one vendor’s suite works with that vendor’s models and cloud. Caesar® is model-independent and sits in front of every model you approve: you decide where each request is processed, Caesar® anonymizes what must not leave, routes each task to the best approved model, cites its sources and records every call for audit. It works alongside the tools you already use.
How is Caesar® different from running an open-weight model on our own servers?
Open-weight models run on your own infrastructure but fall behind the frontier on complex work. With Caesar®, local models handle the work that must stay inside, frontier reasoning is used only where you permit it, and anonymization, routing, verification and audit surround every call. Your own local models can stay part of the set-up.
Do we need a defined AI use case to start?
No. We work through it with your team, then the architecture, infrastructure, security review and deployment path your environment requires.
What do we need to get started?
A clear first task, a person responsible for reviewing it and access to the relevant source material. HAQQ works with your team to define the connections, hosting, permissions and acceptance criteria before the deployment expands.
Do we need to replace our existing software?
Caesar® is designed to connect to existing systems. We agree which records it can read, which actions it can take and which integrations need work. You can use an appropriate HAQQ application, build your own interface or integrate Caesar® into an existing application.
Can we see Caesar® working before we commit?
Yes. Book a 30-minute call to see how a deployment would run for you. Then judge Caesar® on your own work: agree a task, a representative set of records and a reviewer who knows the work, and measure time and cost alongside quality.
Where Caesar® runs
Where can Caesar® run?
In one of three set-ups: HAQQ Sovereign Cloud, fully managed by HAQQ; Hybrid, with your data, Caesar® and the models split across your premises and the cloud; or On-premise, on your own servers and GPUs. You choose where it runs, in the jurisdiction you require.
What is HAQQ Sovereign Cloud?
HAQQ manages the hosting environment. You agree on the region, data handling and permitted models as part of the deployment. There is no hardware or setup to fund, so it is the fastest start.
What is a Hybrid deployment?
Your data, Caesar® and the models split across your premises and the cloud, with external models only where you permit them. You fund hardware only for what stays on site. It suits institutions that need residency together with frontier reasoning.
What is an On-premise deployment?
Caesar® runs on your own infrastructure. Locally hosted models handle work that must stay inside the boundary, with external calls enabled only where permitted. You buy and own the hardware, and keep full control of data and compute.
How do we choose between Sovereign Cloud, Hybrid and On-premise?
Start from your rules. If nothing may leave your building, On-premise with local models keeps every step inside. If your rule is about where data is stored, Hybrid keeps your data on site and sends approved models only what a task needs. If you want the fastest start with no hardware, Sovereign Cloud runs in a region you agree. Tick every set-up you would consider on the discovery form and we will recommend one.
Can we consider more than one set-up?
Yes. The discovery form lets you select every set-up you would consider; it doesn’t lock you into one.
We’re not sure which set-up we need. Can you recommend one?
Yes. Choose “Not sure yet” and we will recommend one once we have read your requirements.
What is the difference between where our data is stored and where it is processed?
Storage is where your documents, conversations and records are kept. Processing is where the AI work happens while a request runs: reading documents, anonymizing and calling a model. Some rules cover only storage; others cover processing too. Tell us which yours covers, and the design follows the text of your rule.
Our rules say data must stay in our country. Which set-up fits?
On premise, the country is your own building. In Hybrid, your data stays on your premises and only what a task needs reaches the models you approve. In Sovereign Cloud, residency is defined in your deployment architecture and agreement, so your security team knows exactly where processing takes place.
Can all processing stay in our environment?
The architecture supports local processing with locally hosted models. The deployment must keep the required tools and data flows inside that boundary too. Your data never reaches the frontier models.
Can Caesar® run with no internet connection at all?
Yes, with locally hosted models and no external calls. Software and model updates then enter through the route your security team approves.
Where is HAQQ Sovereign Cloud hosted? Can it be in our country?
Sovereign Cloud runs in EU, US and Middle East environments. Where residency is a requirement, it is defined in your deployment architecture and agreement, so your security team knows exactly where processing takes place.
Can HAQQ staff outside our country reach our data?
Access is scoped by role, and your institution authorizes the people who may access the material. Who at HAQQ may reach your environment, and from where, is agreed for your deployment.
Can we keep a separate environment or workspace for each country we operate in?
Workspaces keep each request within its matter, case or workspace, and residency is defined for each deployment. Whether each country gets its own hosted environment is set in your deployment architecture.
Can Caesar® work with the data room or document store we already run?
Yes. Caesar® connects to document stores, databases and enterprise software through their available interfaces, and you define what each connection can read or change.
Can we use Caesar® only through its API, keeping our own portal, users and storage?
Yes. Your application stays the front door: it signs users in, holds your documents and conversations and sends each task to Caesar® through the API. Caesar® anonymizes it, routes the work to the models you approve, checks the result and returns one answer, with every call recorded for audit. The API and MCP are in early access.
In an API-only set-up, does Caesar® keep any of our data?
Your application keeps your documents and conversations. Caesar® processes each request and keeps only the audit record your deployment defines; it can run with nothing else kept once the answer is returned.
Can the split between our premises and the cloud differ by data class or by team?
Yes. Your rules decide which classes of data and which teams may use external models. Restricted classes can be configured for local processing, and the anonymization and routing rules are set to match your classification levels.
Can we start in Sovereign Cloud and move on premise later?
Yes. A deployment can start with one task in Sovereign Cloud and move to Hybrid or On-premise as your rules and volumes require. Your knowledge map, connections and workflow rules stay in the engine.
Do we have to buy hardware?
It depends on the set-up. Sovereign Cloud: no hardware or setup to fund. Hybrid: hardware for what stays on site. On-premise: hardware you buy and own.
Can Caesar® run on the servers we already have?
Tell us the servers and GPUs you have or plan to buy. We size to your existing hardware first, and we can visit to see your current infrastructure.
Can our staff use Caesar® while travelling, without leaving our network rules?
Yes. Users reach Caesar® only through the access routes your institution approves, such as your VPN.
Can Caesar® run with our hosting partner or in a national data centre?
Yes. On-premise means infrastructure you control, which can be a data centre you contract.
What does “Hosting countries” mean on the discovery form?
The countries where your data must stay. Add as many as you need. Residency is defined in your agreement.
Why does the discovery form ask for a go-live date?
So we can plan backwards from your budget cycle, any tender and hardware lead times, and decide whether a first task comes before the full rollout.
Models and keys
Which AI models can Caesar® use?
Frontier models and open-weight, self-hosted ones: for example GPT, Claude, Grok, Gemini, Llama, DeepSeek, Mistral, Qwen, Kimi, GLM, MiniMax, Cohere, Nova, Jamba, Palmyra, Solar, Reka and Perplexity, and your own local models. Enabled integrations and API credential options are defined for your deployment.
Are we tied to one model provider?
Caesar® is designed to work with interchangeable models. Your knowledge structure, connections and workflow rules stay in the engine when the underlying model changes. Each enabled model is configured and evaluated for the tasks and policies of your deployment.
How does Caesar® choose which model answers?
Caesar® routes each task to the best model you approve, your own keys and local models included, and matches the depth of reasoning to the request. It only uses the model routes your institution enables.
Can we choose the model or the level of reasoning for a task ourselves?
Yes, for work your developers run through the API. For people’s requests, Caesar® routes automatically to the best model you approve.
Why does Caesar® send parts of one request to several models?
Two reasons. Each part goes to the model best suited to it, and splitting the work means no single provider sees the whole request. Caesar® then compares the answers, verifies them and merges them back into your context.
Can we bring our own model API keys (BYOK)?
Yes. Use your own API credentials with supported providers, under your own contract; Caesar® then calls the model providers through your own accounts.
What are the key options on the discovery form?
Four. Yes, for all models: every call goes through your accounts. Partly: your keys for some models, HAQQ’s for the rest. No: HAQQ’s keys for every model. Or Not sure yet.
What changes if we use our own keys instead of HAQQ’s?
With your keys, the provider’s contract and guarantees stay yours, Caesar®’s controls sit in front, and the provider bills you directly, so the AI usage line of your HAQQ proposal shrinks. With HAQQ’s keys, you need no provider contract of your own, and usage is billed under your HAQQ plan.
Why might we use HAQQ’s keys rather than our own?
HAQQ’s keys need no provider contract on your side. They also strengthen the intent level of anonymization: the parts of your request travel among many customers’ calls, so a provider cannot tell they came from your institution.
Can Caesar® work through our existing enterprise agreement with a model provider?
Yes. If you already hold an enterprise agreement with a provider, Caesar® can call that provider through your own credentials, under your contract.
Can Caesar® use the AI model we already run in house?
Yes. Plug in the open-weight or industry models you have trained, and Caesar® runs them alongside the others. What you have built stays in use.
Which models can run locally, on our own GPUs?
Open-weight models such as Llama, DeepSeek, Mistral, Qwen, Kimi or GLM, chosen to fit the GPUs available. Which models fit your hardware is part of the sizing.
Can we rule out models by provider or by country of origin?
Yes. You enable only the model routes your institution permits, and Caesar® routes only to those. Restricted workloads can be configured for local processing.
What happens when a new model is released?
Caesar® orchestrates the models you choose, so you can bring in new models as soon as they are released, without rebuilding the engine. Each enabled model is configured and evaluated for the tasks and policies of your deployment.
What happens if we change models later?
Your knowledge map, source connections and workflow rules stay in the engine. Changing a model means configuring and testing the new route while keeping that investment.
Does using frontier models mean paying frontier prices on every call?
No. Caesar® prepares the task and retrieves the relevant material before asking a model to reason, and locally hosted services handle the agreed document and context work. Each task goes to the best model you approve, so frontier quality doesn’t mean a frontier price on every call.
Do the model providers keep our data?
External providers, if enabled, are subject to their own agreed processing terms. Data handling, retention and provider terms are reviewed before a route is enabled for your deployment.
Is our data used for model training?
HAQQ’s architecture excludes customer data from training HAQQ models. Where an external model is enabled, its processing and retention terms are reviewed separately for the deployment.
We already use AI tools and models. Do we have to stop?
No. Tell us the models you use today. They may already cover part of the need, and contracts you already hold can be reused rather than paid for twice.
Data protection and anonymization
Does our data reach the frontier models?
Your records stay inside the boundary you set. A frontier model receives only the anonymized part of a task it needs, and only if you have approved that model; in the site’s example, only “What’s 16 × 9?” leaves. In a fully local set-up, nothing reaches an external model at all.
What does “anonymize on three levels” mean?
Personal data, context and intent are each anonymized before anything leaves. Personal data: names, numbers and other identifiers are replaced. Context: the request is broken into smaller questions that carry no context on their own. Intent: those questions go to different models, so the purpose of the request stays behind.
What is the first level (fact-based, personal data)?
Caesar® finds personal and sensitive facts, such as names, identification and account numbers, contact details and the terms you define, and replaces them with stand-ins before a request leaves. The real values are put back in the answer you receive.
What is the second level (context-based)?
A local model breaks the request into smaller questions that make sense only together. Each question goes out without the surrounding context, the way the site’s example sends only “16 × 9”.
What is the third level (intent-based)?
The smaller questions are spread across different models and providers, which makes it hard for any one provider to piece together what was asked or why. Caesar® then verifies the answers and merges them back into your context.
How do the real names and numbers come back into the answer?
Caesar® holds the link between each stand-in and its real value while the request runs, and puts the real values back before the answer reaches you. The models see only the stand-ins.
Which kinds of personal data does Caesar® detect?
Names, identification and passport numbers, account and card numbers, contact details, dates and the sensitive terms your institution adds, in Arabic, French and English.
Can we add our own list of sensitive terms?
Yes. Your institution can add its own sensitive terms, such as internal project or counterparty names, and Caesar® treats them like personal data.
Can anonymization follow our data classification levels?
Yes. Each document or request reaches Caesar® with its classification, and Caesar® applies the anonymization and routing rules set for that class: some classes may reach approved external models once anonymized, others stay on local models.
Which classes of data can Caesar® handle? Can it take our most confidential material?
You decide. Your rules set which classes may reach Caesar®, which may use external models once anonymized, and which must stay on local models inside your boundary. For the most sensitive material, the fully local set-up keeps every step inside.
How is this different from the masking our data-loss prevention tools already do?
Masking replaces names and numbers, which is the first of Caesar®’s three levels. Caesar® also splits the request so no part carries its context, spreads the parts across models so the purpose stays behind, verifies the answers, and keeps an audit record of every call.
Is anonymized data still personal data under our law?
That depends on the law that applies to you, and we review it with your legal team for each deployment. Caesar®’s anonymization keeps what any provider receives to a minimum, on top of the contractual controls in your agreement.
What happens if something sensitive is missed?
Caesar® is designed to fail closed: if a check cannot complete, the request does not leave. Every call is recorded, so your reviewers can inspect what was sent.
Can we see exactly what was sent to each model?
Yes. For each request, the audit record can keep the anonymized version that left at each level, which model received it and when, so you can show what was shared and with whom.
Does Caesar® keep our prompts, documents or answers?
Only what your deployment defines. Your institution sets retention for conversations, logs and generated documents, and Caesar® can run with nothing kept once the answer is returned, apart from the audit record you choose.
Is our data encrypted, and who holds the keys?
Yes, data is encrypted. For enterprise deployments your institution can manage its own encryption keys, and HAQQ cannot read the encrypted data once a key is revoked. Encryption-key arrangements are reviewed separately from model credentials.
Can we keep our encryption keys in our own hardware security module (HSM)?
Encryption-key arrangements are reviewed for each deployment; tell us your key-management standard and your HSM.
Can HAQQ see our data?
Your institution authorizes the material and the people who may access it, and HAQQ configures only the agreed connections and access scope. Where your institution manages its own keys, HAQQ cannot read the encrypted data once a key is revoked.
Are scanned documents read inside our boundary?
Yes. The document vision service turns scans and document images into usable text before analysis begins, within the engine’s hosting boundary.
Does Caesar® respect our existing access rights?
Yes. A user asking about one project should only reach the records and tools that user is permitted to access. Caesar® applies role, workspace and deployment rules when preparing the task and selecting tools and models.
Can some teams be kept fully separate from the rest?
Yes. Workspace isolation keeps each request within its matter, case or workspace, and access across teams and connected systems is configured deliberately; for example, only the supervision team can open its own files.
How is our data kept apart from other customers’?
Each deployment and its workspaces are isolated; on premise, the installation is yours alone.
Who owns our data and what Caesar® produces?
Your institution. Customer knowledge stays customer knowledge.
Can we delete our data, and what happens when the contract ends?
Retention and deletion of conversations, logs and generated documents are set for your deployment, and the exit plan, data export and deletion included, is agreed in your contract.
Audit, oversight and human control
Is every request logged?
Yes. Every call is logged for audit, and every action is logged and attributed.
What does the audit record contain?
Your deployment defines what is captured. It can include, for each request, the user, the question and the answer, the sources used, the anonymized version at each level, the models called and when, the tool activity and the approvals, so your reviewers can show who asked what, what left and what came back.
Can the audit record feed our SIEM?
Yes. The audit record can be sent to your security information and event management platform, for example Splunk, in the format your security team uses.
Where is the audit record kept?
On premise, it stays in your environment. With Sovereign Cloud or the API, it can come back to your systems with each answer, or be kept by HAQQ in the country you require.
Can we have a full audit trail without HAQQ keeping our data?
Yes. The audit record can be returned to your systems with each answer, so nothing stays at HAQQ after the response.
Is a complete end-to-end trace of each request available?
Yes. Beyond the standard record of every request and response, a full trace of each step can be added; it is scoped in your proposal.
Does audit logging cost extra?
Audit logging comes with every deployment. An extended end-to-end trace is scoped in your proposal.
How long are audit records kept?
Retention follows your rules, and the retention, deletion and recovery requirements of your deployment are agreed with your security team.
Can we require a person to approve a step before it goes ahead?
Yes. Place review and approval at the points where the workflow requires a person to take responsibility; human approval can be required at any step you choose.
Can our auditors inspect what ran and who ran it?
Yes. Authorized reviewers can inspect the source material behind an answer, the versions behind a draft and the recorded tool activity behind an action. The deployment defines what is captured and who can inspect it.
How can we check how Caesar® reached an answer?
Every claim is cited, with references a reviewer can open, and the audit record shows the sources, tool activity and approvals behind the result, so a reviewer can check the answer against the material.
Who stays responsible for decisions?
Your people. Caesar® prepares the work; your team reviews it and makes the call. You define who reviews, who approves and what may proceed after approval.
How does Caesar® keep wrong answers to a minimum?
It retrieves the relevant material before a model reasons, compares answers across models before merging them, and cites every claim so a reviewer can check it against its source.
Security and compliance
Which security standards is Caesar® built to?
Caesar® is built to the security and AI-governance standards regulators expect, and every call is logged for audit: SOC 2 (Trust Services Criteria), ISO 27001 (information security), ISO 42001 (AI management) and GDPR (processor obligations), each shown as Compliant.
Is HAQQ certified to SOC 2, ISO 27001 or ISO 42001?
Compliant, pending certification. The controls behind each standard are built and running inside HAQQ; the official certification is pending, because an external audit takes time. The available assurance evidence is shared as part of your security review.
What should our security team review before deployment?
Data location, processing flows and external providers; identity, permissions and administrative access; encryption, retention, deletion and recovery requirements; logging, human approval and incident responsibilities; applicable standards, contractual terms and available assurance evidence.
Will HAQQ answer our security questionnaire and join our security review?
Yes. Bring your security, architecture and procurement teams into the discussion. We work through the requirements that decide whether the deployment fits your institution, starting with your data boundary.
What assurance evidence can HAQQ provide?
Available assurance evidence is shared as part of the security discussion, together with the applicable standards and contractual terms.
Can we run our own penetration test, or will HAQQ provide one?
Your team or a tester you appoint can test the deployment before go-live, under rules agreed with HAQQ.
Who is responsible for what?
The deployment plan makes it explicit. Your institution authorizes the material and the people who may access it, defines residency and permitted processing, decides what workflows can do and who approves, sets acceptance criteria and owns internal escalation. HAQQ configures the agreed connections and access scope, designs the deployment and model routes to that scope, configures tools and review steps, supports testing, and agrees support, monitoring and maintenance responsibilities.
How do users sign in?
Through single sign-on, with MFA and role-based access, connected to your identity provider, for example Microsoft Entra ID, Active Directory, Okta or Ping Identity.
How does Caesar® guard against prompt injection and misuse?
Pre-generation checks run before any model is called, and Caesar® enforces the scope of every request: access boundaries, tool permissions and safety rules. A workflow uses only the tools your institution approves, and pauses for human approval where your procedures require it.
Do you sign a data processing agreement?
Yes. Data handling, residency and retention are written into your agreement, with a data processing agreement where your rules require one.
Is HAQQ a data processor or a controller?
Your institution decides what Caesar® may process; under GDPR, HAQQ meets processor obligations.
Who are your sub-processors?
The sub-processors for your deployment, hosting and any external model providers you enable, are listed in your agreement.
Can the HAQQ staff on our project be vetted, or be nationals or residents of our country?
Tell us your rules on who may work on the system: security vetting, nationality, residency or on site only. We staff the deployment to them.
Do you hand over source code or model weights, or offer escrow?
Source code and model weights are not transferred. Escrow and a license can be agreed instead.
What happens if there is a security incident?
Incident responsibilities, notification deadlines and contacts are agreed in your deployment plan.
Could a foreign government compel access to our data?
On premise, your data stays in your own environment. In Sovereign Cloud, the hosting environment and residency are defined in your agreement.
Integration, applications and languages
How do our systems connect to Caesar®?
Through the API and MCP. An API lets your software send a task to Caesar® and receive the result; MCP is a standard through which a compatible assistant can use external tools and sources. Both are in early access.
Can our own AI assistant use Caesar®?
Yes, through MCP, a standard through which a compatible assistant can use external tools and sources. The supported connection and permissions are agreed for your deployment.
Can we build our own applications on Caesar®?
Yes. Your developers can put research, document comparison or drafting inside the application your team uses, or build new applications on top of Caesar® through the API.
What kinds of applications can we build on Caesar®?
For example an internal assistant for staff, a contract review portal, or a review workspace for a supervision or compliance team.
What are the three ways to work with Caesar®?
Your team builds: connect Caesar® to your existing software through the API. Your team uses: work in an application HAQQ already provides, such as a web chat, a Word plugin or a mobile app, with nothing to build on your side. We build together: HAQQ scopes, builds and supports an application on Caesar® for a workflow no standard interface covers.
Which HAQQ applications are available?
Ready-made: the web workspace and web chat, the Word plug-in, the mobile app, eFirm and HAQQ Legal AI. Each is connected to Caesar® and set up to your access rules.
What is the difference between ready-made and custom applications?
Ready-made applications (web, Word add-in, mobile and eFirm) work as they are and are the fastest start. A custom build is an application designed on Caesar® around how your teams submit work, check sources and approve the next step. You can choose both.
On the discovery form, what do “Caesar® licence + API”, “Caesar® API” and “HAQQ applications” mean?
Caesar® licence + API runs Caesar® for your organization, with your own API keys. Caesar® API lets you call Caesar® from your tools and applications. HAQQ applications are ready-made or built for you. Tick all that apply, or choose Not sure yet and we will help you choose.
Which tools and systems can Caesar® connect to?
We connect the systems your workflow needs, through their available interfaces, and scope each connection in implementation. The discovery form suggests common ones by group: productivity (Microsoft 365, SharePoint, Teams, Google Workspace, Slack); document management (iManage, NetDocuments, OpenText, M-Files, Box, Dropbox); CRM (Salesforce, Dynamics 365, HubSpot, Zoho CRM); ERP (SAP, Oracle, Dynamics 365 Finance, Odoo, Workday); identity (Entra ID, Okta, Active Directory, Ping Identity); practice and case management (Clio, Elite 3E, Aderant, Intapp, Relativity); ticketing and IT (ServiceNow, Jira, Confluence, Zendesk); data (Snowflake, Databricks, PostgreSQL, SQL Server, Elasticsearch); and cloud (Azure, AWS, Google Cloud, VMware, Kubernetes).
Our system isn’t on the list. Can Caesar® still connect to it?
Yes, through its available interfaces. Add it on the discovery form, and tell us how you connect to it today, for example through an API, a database or a file export. In-house systems are welcome.
Can Caesar® act in our systems, or only read them?
You decide. Define what each connected tool can read or change; read-only is a fine start. Actions can wait for human approval where your procedures require it.
Which file types can Caesar® read?
Word, Excel and PDF files, scanned PDFs included, and document images; the document vision service turns scans into text inside your hosting boundary.
Which languages does Caesar® work in?
Caesar® is built on Justinian®’s foundation, which was built for Arabic, French and English.
Do HAQQ’s applications work in Arabic, right to left?
Yes.
Does every answer cite its sources?
Yes. Every claim is cited, with references a reviewer can open to check the answer against the material.
Can Caesar® produce Word, Excel or PDF documents?
Caesar® prepares and revises documents using the task’s sources, instructions and templates, and keeps version history so your team can see how the document developed.
Can Caesar® run work with no person involved, such as batch jobs or scheduled reviews?
Yes. Agents plan and carry out multi-step work, such as reviewing every new submission or checking records on a schedule, and pause for human approval where required. Work that runs without a person is sized separately from people’s use.
Can we use Caesar® inside Microsoft Word?
Yes, through HAQQ’s Word plug-in.
Can Caesar® keep earlier records and decisions in context?
Yes. Caesar® brings earlier records and decisions into the current task, with their dates and workspace boundaries. A supplier’s previous review can inform the next one without being treated as a current fact.
Infrastructure and on-premise
What hardware does an on-premise deployment need?
Servers and GPUs sized to your users, your busiest hours and the work that must run on local models. Tell us what you have or plan to buy (model, number and memory per GPU); if you have none yet, we will size it for you.
How do you size the hardware?
From how many people will use Caesar® and how many at the busiest hour, the work that runs without a person, and how much must stay on local models rather than approved external ones. Your current AI usage, even a rough range, gives us a baseline.
Can you size to the GPUs we already own?
Yes. We size to your existing hardware first, and tell you if anything must be added.
Can HAQQ help us buy servers and GPUs?
Yes. Procurement support covers help with tenders, hardware and vendor paperwork. Tick Procurement in the support step of the discovery form.
Which platforms does Caesar® run on?
Your existing platform: Kubernetes (including VMware Tanzu), Docker hosts or bare-metal GPU servers, on Ubuntu or your approved operating system.
Which databases and storage does an on-premise install need?
Standard components your team can run, a relational database, a cache and object storage, aligned to your approved software list.
How do software and model updates reach an on-premise installation?
Through the route your security team approves: your recorded VPN, a supervised session, or an offline package your team imports. You set the release cadence and the maintenance windows.
Can HAQQ support an on-premise system without remote access?
Yes, with engineers on site. Where your policy allows remote access, through supervised sessions or your VPN, support can also be remote.
Do we need test and disaster-recovery environments?
That is your choice, and each environment adds hardware. Tell us the environments you need (production, test, disaster recovery) and your availability and recovery targets.
What uptime can HAQQ commit to?
With a Guaranteed SLA, uptime and response times are written into the contract.
Can several institutions share one sovereign AI infrastructure?
Caesar® can run on infrastructure shared across a sector, such as one data centre serving several institutions, with each institution’s data and workspaces kept apart. It is scoped as its own project.
Support and services
Which support levels do you offer?
Four: Self-serve, On-demand, Enterprise and Government grade. Each level comes with its usual set of services, and you can change any of them.
What does Self-serve include?
Documentation and standard support.
What does On-demand include?
HAQQ when you call: onboarding and help as you need it. It usually includes implementation and training.
What does Enterprise include?
A dedicated HAQQ team, with an SLA and regular reviews. It usually includes a guaranteed SLA, customization, implementation and training.
What does Government grade include?
Dedicated engineers on site and every guarantee in the contract. It usually includes a guaranteed SLA, HAQQ engineers on site, customization, implementation, training and procurement help.
Can we pick services from different levels?
Yes. Each level ticks its usual set of services; change any of them.
What is a Guaranteed SLA?
Uptime and response times written into the contract.
Can HAQQ engineers work at our premises?
Yes. HAQQ staff can be deployed at your premises. It comes with Government grade support and can be added to any level.
What does Customization cover?
Caesar® adapted to your workflows and data.
What does Implementation cover?
We set Caesar® up and connect it with your team: data connections, workflow configuration and integration with your systems.
What training do you provide?
Sessions for your users and your administrators.
What does Procurement help mean?
Help with tenders, hardware and vendor paperwork: the technical specification, the hardware list and the vendor documents a public procurement needs.
Who runs Caesar® day to day: our team or HAQQ?
Your team, HAQQ, or both. Your institution agrees internal ownership and escalation; HAQQ agrees support, monitoring and maintenance responsibilities, written for each deployment.
What can HAQQ see when it monitors our deployment?
Only what you allow: nothing, read-only dashboards, or supervised sessions.
How often are updates released, and can we control when they are applied?
The release cadence, maintenance windows and testing before production are agreed with HAQQ and run through your change process.
Do you work through local partners for support?
In some countries a local partner gives first-line support, with HAQQ engineering behind it.
How long does a deployment take?
It depends on the set-up and on your approvals. Sovereign Cloud is the fastest start, with no hardware to fund; an on-premise deployment follows your hardware lead times. We start with one task and a reviewer who knows the work, then expand.
Pricing and buying
How is Caesar® priced?
Your proposal identifies the engine license, usage, applications and implementation services you need, as four lines: engine license, AI usage, applications, and implementation and support.
What is the engine license?
The license to run Caesar® in the agreed environment. It supplies the shared knowledge, retrieval, model routing and controls behind your AI workflows.
How is AI usage measured and billed?
You pay for consumption under the agreed plan. The proposal defines whether usage is measured by API call, credit or token, and how your own model credentials affect billing.
What do the Applications and the Implementation and support lines cover?
Applications: subscriptions to applicable HAQQ applications, or software commissioned for your teams on top of Caesar®. Implementation and support: data connections, workflow configuration, custom development, training and support.
How do our own model keys change the price?
With your own keys, the model provider bills you directly for those calls, so the AI usage line of your proposal shrinks.
Do you publish prices?
No. We don’t quote before we understand your requirements; your proposal is built from your set-up, models, applications and services.
Can you give us a budgetary price for our next budget cycle?
Yes. Once we understand your requirements, we prepare a budgetary proposal in separate lines (engine license, AI usage, applications, implementation and support) that maps onto your budget and any tender.
Can usage be billed monthly against an annual commitment?
Yes: an annual commitment, billed monthly on actual usage, at an agreed unit price.
Is the license a one-time fee or a subscription, and can we budget it as capital or operating spend?
Your proposal sets the license model for your environment. Tell us whether you budget it as capital or operating spend.
Why does the discovery form ask about our current AI spend, token use and budget?
They size your deployment: where you are today and where you want to go. Your current spend is the baseline for your cost-benefit evaluation, token use sizes GPUs on premise and usage in the cloud, and your budget tells us whether to propose one phase or several.
We don’t know our token use or our budget yet. Can we still fill in the form?
Yes. A rough range is fine, zero is fine, and “I don’t know” is a valid answer.
Can we run a pilot or proof of concept before we buy?
Yes. A first deployment starts with one task, a representative set of records and a reviewer who knows the work, judged on agreed acceptance criteria: quality, time and cost.
Can you respond to our tender or RFP?
Yes. Share the tender specification if one exists and we will answer to its structure. Procurement support also helps with the technical specification and the vendor paperwork.
Is Hybrid priced differently?
Hybrid splits differ a lot from one institution to the next, so we map yours on a short call before we quote.
Do you work with resellers or partners in our country?
In some countries HAQQ works with local partners. Contact us through the partnership inquiry.
The discovery form and next steps
What is the Caesar® discovery form?
A short intake, about 5 minutes over 7 steps (about you, what you need, your tools, deployment, models and keys, support, and usage and budget), that tells us enough to shape your offer and prepare the next conversation.
Which questions are required?
About you, what you need, the deployment type, models and keys, and usage and budget. Your tools and support are optional.
Does filling in the form commit us to anything?
No. It shapes your offer; it doesn’t lock you into a set-up, a price or a contract.
Who reads our answers?
HAQQ’s sales lead, with the technical and leadership team working on your file in copy.
What happens after we submit?
We read your answers and documents and come back with the next step. If you chose Hybrid, we set up a short call to map the split before we quote; otherwise we prepare your proposal.
Why does Hybrid need a follow-up call?
A hybrid deployment can split your data, Caesar® and the models across your premises and the cloud in many ways. A short call lets us map yours before we quote.
Which documents should we share?
Any of these you have; none is required: your requirements or RFP, or a short note on what goes live first; a high-level architecture diagram of where Caesar® would sit; your hardware and platform, if any part runs on your premises; and a recent export of your current AI usage.
What should we leave out of what we share?
Send only what you’re allowed to share outside your organization, and remove hostnames, IP addresses, credentials and personal names. A short note is as useful as the full document.
How many files can we upload, and in which formats?
Up to 100 files: PDF, Office, Visio, CSV, text or images.
Can we talk to someone before filling in the form?
Yes. Book a 30-minute call: set out your rules (where Caesar® may run, which records it may reach and which models it may call) and leave with the next step agreed.
How do we contact HAQQ about Sovereign AI?
Book a call on this site, or write to sovereign@haqq.ai.
We are a partner, integrator or reseller. Should we use the discovery form?
The discovery form is for organizations deploying Caesar® for themselves. Partners: use the partnership inquiry on the contact form.
Central banks, regulators and other industries
How can a central bank or financial regulator use Caesar®?
Three examples. Supervision: review every submission against the rules that apply. Circulars and policy: trace a new circular to everything it changes. Legal and contracts: prepare legal work from the sources your counsel trusts. Your teams review every decision.
Can Caesar® check a bank’s submission against the circulars that apply?
Yes. Ask, for example: “Check this bank’s quarterly return against the circulars that apply. Separate breaches, gaps and questions for management.” Caesar® draws on bank submissions, circulars in force, previous findings and examination manuals, and delivers a sourced compliance review, breaches and missing data by rule, and follow-up questions for the bank. The supervision team keeps every finding and enforcement decision.
Can Caesar® trace what a new circular changes?
Yes. Ask: “Show which circulars and procedures this draft amends, and what each bank would have to change.” Caesar® draws on the draft, circulars in force, laws and decrees and consultation responses, and delivers an amendment map with sources, the affected institutions and obligations, and a consultation and impact note. The regulation department decides the final text, its timing and its scope.
Can Caesar® review our legal agreements?
Yes. Ask: “Review this bond purchase agreement against our standard terms and the governing law. Flag deviations and open points.” It delivers a clause-by-clause review, deviations each with its source, and a draft opinion for counsel. It is built on Justinian®, HAQQ’s sovereign legal AI, and the legal department keeps every legal position.
How do other industries use Caesar®?
Each industry page shows work Caesar® can be configured to support, the sources it needs and the people who review the result: a ministry reviewing a policy change, a manufacturer investigating a recurring fault, an insurer preparing a claim file.
Does Caesar® know our country’s laws and regulations?
Caesar®’s data layer structures laws, regulations and your own documents for the engine, and it applies the right language, jurisdiction or domain framework to each request. You can also connect the reference content you maintain, such as your own circulars.
Can Caesar® review transactions continuously rather than in periodic audits?
Agents can run reviews on a schedule, for example daily or weekly, and set out findings for your auditors to review.
Can our legal team use HAQQ Legal AI on its own?
Yes. Justinian® powers HAQQ’s legal products and is the foundation Caesar® grew from. Your legal team can use HAQQ Legal AI on its own, or as part of a Caesar® deployment.
About HAQQ
Who is HAQQ?
HAQQ builds Sovereign AI for sovereign industries. Its engines let governments and enterprises use frontier AI models at a lower cost, without compromising the sovereignty, security or safety of their data. Justinian® powers HAQQ’s legal products; Caesar® takes the engine to every other sovereign industry.
Where is HAQQ based?
HAQQ Inc. is incorporated in Delaware, with teams in Beirut and Dubai.
How does HAQQ evaluate Caesar®?
Against every way to put AI to work: industry harnesses, industry-tuned general AI, generic LLMs and industry AI tools. The same protocol runs in every sector: real tasks written with practitioners, the same inputs for every route, blind grading, and every result published, including the ones Caesar® does not lead.
Nothing here answers that yet. Ask us on a call and we will answer it in person.
Bring the question we have not answered.
Tell us what your institution needs to know about running Caesar®, and we will work through it with your team.
Bring the problem. We’ll shape the use case with you.
You don’t need a defined AI use case to start. We work through it with your team, then the architecture, infrastructure, security review and deployment path your environment requires.
Prefer email? Write to our team