Introducing Sukarma Read the announcement

Every good deed
travels further.

Sukarma builds software, including AI, for councils, charities and regulated organisations. A person stays in charge of every decision, data stays protected, and the service works for everyone who uses it.

  • People decide
  • Data protected
  • Accessible to WCAG 2.2 AA
  • UK registered company

We build with leading AI models and cloud platforms

Anthropic ClaudeOpenAI GPTGoogle GeminiMeta LlamaMistralModel Context ProtocolGoogle Cloud Vertex AIAmazon BedrockAzure AI FoundryLangGraphPostgreSQLpgvectorTypeScriptPythonNode.jsKubernetesTerraformOpenTelemetryAnthropic ClaudeOpenAI GPTGoogle GeminiMeta LlamaMistralModel Context ProtocolGoogle Cloud Vertex AIAmazon BedrockAzure AI FoundryLangGraphPostgreSQLpgvectorTypeScriptPythonNode.jsKubernetesTerraformOpenTelemetry

People are right to be wary of powerful technology. We answer that with evidence: software you can test, and a person accountable for what it does.

The Sukarma Promise

AI, without fear.

Five promises we make to every client. Why we are called Sukarma.

  1. 01

    Fear not · for people

    Services that work for everyone

    We build to WCAG 2.2 AA, write in plain English and always give users a way to reach a person.

  2. 02

    Fear not · for your people

    AI assists. People decide.

    AI handles the repetitive work; your staff make the decisions. When we leave, your team has the documentation and skills to carry on.

  3. 03

    Fear not · for your data

    Protected by design

    We host in the UK or EU, in your own cloud account where possible, and start a data protection impact assessment in week one. Your data never trains anyone else's AI.

  4. 04

    Fear not · for accountability

    A named person, always

    Every important output has a named owner and an audit trail, and we label anything AI wrote. We draft Algorithmic Transparency Recording Standard records as we build.

  5. 05

    Fear not · for your budget

    No lock-in, no surprises

    Prices are fixed or capped. You can switch AI providers at any time, and you own everything we build.

Forge

AI does the groundwork. A person approves every step.

Forge is the system we use on every project. AI agents analyse, build and test; a senior member of our team reviews and signs off each stage.

forge · example rundemo
Agents5
Tests passing0
WCAG 2.2 AA—
Human approvals0/3

Analyst agents + senior lead

Understand

Agents read your policies, specifications and existing systems, then draft user needs and requirements in hours. Our lead checks every one.

Requirements · Risks · Assumptions

How Forge works

176tender answers drafted by Forge Tender, each labelled with its source
13requirement areas covered by a working platform we built during a bid
5AI agent roles on every project, each checked by a person
2.2 AAthe WCAG accessibility level we build to

Services

From first idea to live service

Six ways we can help, from a two-week review to running a live service.

How we work

What to expect from us

Typical approachSukarma
First working softwareSome months after contract award Often during discovery
TeamLarger teams A small senior team, supported by AI
DocumentationOften written at the end Written and checked as we go
AccessibilityOften audited late Tested on every change
AI modelsOften tied to one vendor Any provider, chosen per task
PricingDay rates Fixed or capped prices

Products

Built on real projects

Each product started as a tool we needed on a bid or a client project.

In useDelivery platform

Forge

The system behind all our work. AI agents analyse, build and test; senior people review and approve every step.

  • Several AI models working together
  • Requirements linked to tests
  • Automated accessibility and security checks
  • A person approves every release
How Forge works
In developmentAI agent assurance

Forge Guard

The controls from Forge, packaged for any AI agent. It helps security teams decide whether an agent is safe to use, and keeps it within agreed limits once it runs.

  • A risk assessment for a specific agent in a specific context
  • Role limits, approval gates and rollback
  • An audit trail of every action and approval
  • Core framework to be released as open source
How Forge Guard works
In useProcurement

Forge Tender

Reads a whole tender pack, tells you whether to bid and why, drafts the answers with their sources, and submits only after you confirm.

  • Reads every document in the pack
  • Each answer labelled with its source and confidence
  • Three confirmations before anything is sent
  • A fingerprint (SHA-256) of every file submitted
Ask about Forge Tender
Demo availableResident engagement

Forge Engage

Surveys, consultations, idea boards and events for residents and customers, with consent records, moderation and live dashboards.

  • MFA, consent records and data-rights requests
  • Reports on who is taking part
  • Core journeys work without JavaScript
  • REST API, SCIM 2.0 and an MCP server
Book a walkthrough
In developmentCasework

Forge Assist

Helps casework teams summarise case histories and draft letters in your house style. It checks drafts against policy and passes anything sensitive to a person.

  • Drafts that cite the policy they follow
  • Personal data removed where needed
  • Plain-English drafting
  • Test reports for assurance boards
Register interest
Demo availableWorkforce transformation

Forge Future

Grow instead of cut. Forge Future shows how an organisation can reshape roles, retrain its people and fund new businesses with the workforce it already has.

  • Starts from the real workforce: functions, grades, costs and ratings
  • One plan seen by leadership, technology and team managers
  • Every count and cost traced back to the people data
  • Live demo on a fictional bank with synthetic data
See Forge Future

Case studies

Two things we built this year

Standards

Built to the standards your assurance board will ask about.

We work to the GDS Service Standard, the Technology Code of Practice and the UK Government's AI Playbook. Security and data protection are part of the design from the first day.

  • GDS Service Standard
  • Technology Code of Practice
  • AI Playbook for the UK Government
  • Algorithmic Transparency Recording Standard
  • NCSC Secure AI Guidelines
  • NCSC Cloud Security Principles
  • WCAG 2.2 AA
  • UK GDPR & DPA 2018
  • OWASP Top 10 for LLM Apps
  • Procurement Act 2023
  • Welsh Language Standards

Insights

Latest articles

All articles

Programmes and registrations