Resources
Glossary
39 terms we use with clients, each explained in plain language.
AI
Models, agents and the methods that keep their answers accurate.
16 terms
- AI agent
- Software that uses a language model to plan and take actions with tools, such as reading email, updating a record or calling an API, usually with limits and human approval for risky steps.
- Agentic workflow
- A process where one or more AI agents carry out several connected steps toward a goal, such as reading an order, checking stock and drafting a reply, in one run.
- Large language model (LLM)
- A model trained on large amounts of text that can read, write, summarise and reason in natural language. Examples include Claude, GPT and Gemini.
- Open weight model
- A model whose weights are published so it can run on your own infrastructure, such as Llama, Qwen or Mistral.
- RAG (retrieval augmented generation)
- A method where the system first searches your documents or data, then asks the model to answer using only what it found, and to cite where each part came from.
- Embedding
- A list of numbers that represents the meaning of text or an image, used to find similar content in semantic search.
- Vector database
- A database that stores embeddings and finds the closest matches quickly. Examples include pgvector and Qdrant.
- MCP (Model Context Protocol)
- An open standard that lets AI assistants connect to tools and data sources in a consistent, permissioned way.
- Evaluation (evals)
- A set of test cases with expected results, used to measure the quality of an AI system before and after changes.
- Prompt injection
- An attack where text hidden in a document, email or web page tries to override the instructions given to an AI system, for example telling it to approve a payment.
- Hallucination
- When a model states something confidently that is not true. Retrieval with citations and regular evaluation reduce it, but do not remove it entirely.
- Fine tuning
- Further training of a model on your examples so it follows a style or task more closely. Needed less often than people expect: good retrieval and prompts solve most problems first.
- MLOps and LLMOps
- The practices for deploying, monitoring, updating and controlling the cost of machine learning and language model systems in production.
- Human in the loop
- A design where a person reviews or approves certain AI outputs or actions before they take effect.
- Document intelligence
- Reading documents such as invoices, purchase orders, KYC forms and contracts, and turning them into structured data that can be posted to Tally, an ERP or a CRM.
- Copilot
- An AI assistant built into a product or role that helps a person do their work, such as drafting, searching or summarising.
Delivery
How engagements are scoped, staffed, priced and accepted.
7 terms
- Forward Deployed Engineer (FDE)
- An engineer who works inside the client's team and systems, takes a use case into production and stays accountable for it running.
- Forward Deployed Strategist (FDS)
- A business focused role that maps the process, agrees success measures and checks after launch that the value is real, working alongside FDEs.
- Delivery pod
- Our standard forward deployed team: one strategist and two engineers, with QA shared across pods.
- O2E (Output Oriented Engagement)
- Our engagement model where work is arranged in fixed capacity blocks and paid for after the client reviews and accepts the output.
- TSE (The Startup Engine)
- Our execution partnership for startups, where we share outcomes and align through equity.
- Global Capability Centre (GCC)
- A centre that provides specialised capabilities, such as technology, AI and consulting, to businesses as if it were their own in house unit.
- Technical specification
- A signed document that defines what will be built and how acceptance will be judged.
Engineering
The build and run vocabulary behind the software we ship.
6 terms
- CI/CD
- Continuous integration and continuous delivery: automatically building, testing and deploying software on every change.
- Headless commerce
- An e-commerce setup where the storefront is separate from the commerce backend and connected by APIs.
- Microservices
- An architecture where an application is built as a set of small, independently deployable services.
- Infrastructure as code
- Defining servers, networks and cloud resources in version controlled files, for example with Terraform.
- DevSecOps
- Building security checks into the development and delivery pipeline, so they run on every change, all the way to launch and beyond.
- Penetration test
- An authorised, simulated attack on a system to find security weaknesses before real attackers do.
Compliance
The laws and standards that shape how personal data is handled.
10 terms
- DPDP Act 2023
- India's Digital Personal Data Protection Act, which governs how digital personal data of people in India is collected and used. Its Rules were notified in November 2025, and most duties apply from May 2027 as currently notified.
- Data fiduciary
- Under the DPDP Act, the organisation that decides why and how personal data is processed.
- Data principal
- Under the DPDP Act, the person the personal data is about.
- Consent manager
- A registered entity under the DPDP Act that helps people give, manage and withdraw consent through one platform.
- Significant Data Fiduciary
- A data fiduciary notified by the government as higher risk, with extra duties such as a DPO, audits and impact assessments.
- GDPR
- The EU General Data Protection Regulation, which protects personal data of people in the European Union.
- EU AI Act
- The European Union law that regulates AI systems according to their level of risk.
- ISO/IEC 42001
- An international standard for an AI management system, covering governance, risk and controls for AI.
- HIPAA
- The US law that protects health information handled by healthcare organisations and their business associates.
- PCI DSS
- The security standard for organisations that handle payment card data.
Start a conversation
Tell us what you are trying to build.
Share the problem in a short call. You leave with a clear view of the approach, the team and the cost.
Or get a ballpark proposal in two minutes
A specialist replies within one working day. The first conversation is free.