Security26 Oct 20267 min read
Two critical PHP flaws, CVE-2026-109618 and CVE-2026-109619, were patched on 21 October 2026. What is affected, what to update, and what to check afterwards.
We build AI agents that do the work, and know when to ask a person.
Fullness designs, builds and runs agentic workflows, data platforms and custom software for teams that want outcomes, not demos.
Invoice approval
Run 4821
ReceivedAgent
Invoice from Northfield Supplies, 14 fields read
MatchedAgent
Purchase order 7720
FlaggedAgent
Total is 3.2% above the order
Needs a personNeeds approvalAnswered
Sent to Meera in accounts payable
ApprovedMeera
Meera accepted the difference
PostedAgent
Written to the ERP and logged
What we build
Five kinds of work, one way of doing it: start with a process that costs you real time, prove it on your own data, and keep a person in charge of what matters.
- AI agents and agentic systemsAgents that read, decide and act across your systems, with a person approving anything that matters and a log of every step.
- Workflow and process automationDependable automation that connects your tools, moves data between them and handles exceptions, with AI where judgement is needed and plain rules where it is not.
- Data engineering and analyticsPipelines, warehouses and dashboards that bring your data together, keep it clean and put the numbers in front of the people who decide.
- LLM apps, RAG and knowledge systemsAssistants and search that answer from your own documents and systems, cite their sources, and respect who is allowed to see what.
- AI-ready software engineeringProduct design and engineering for web and mobile apps, APIs and cloud platforms, with the DevOps and security work that keeps them running.
How a project runs
The same four steps whether the first project is an invoice agent or a data platform.
Step 1: Map
We sit with the people who do the work today. We map the process, the data it touches and the points where a person must stay in charge.
You get: A scoped plan, a measure of success and a fixed price for the pilot.
Step 2: Prototype
We build the narrowest version that works on your real data, with approval steps in place from the first day.
You get: A working agent your team can try on real work, and an honest view on whether to go further.
Step 3: Harden
We add evaluations, guardrails, access controls and monitoring so it can be trusted with real volume.
You get: Test results, a runbook, and a log of every decision the agent makes.
Step 4: Run
We run it with you, watch the numbers, fix what drifts and train your team to own it.
You get: Regular reporting, support, and a clear route to taking it fully in-house.
What we build with
We pick the tool for the job and design so you can swap it later. Nothing here locks you in.
- Models
- Claude
- GPT
- Gemini
- Open-weight models you host
- Agents and workflows
- LangGraph
- Temporal
- n8n
- Model Context Protocol
- Data
- PostgreSQL
- MongoDB
- BigQuery
- Snowflake
- dbt
- Airflow
- Platform
- AWS
- Google Cloud
- Azure
- Kubernetes
- Docker
- Languages
- Python
- TypeScript
- Go
Working with Fullness has been an absolute pleasure. The experienced team of designers and engineers revamped our website, mapped its performance and made quality changes to help us achieve greater visibility and increase qualified traffic on the site.
Pravin Mahajan
CTO, CIMET
Latest insights
All insightsTerminal26 Feb 20247 min read
Using a terminal can be a daunting task if you are just starting or don’t know the commands required to do the work at hand.
web development25 Jan 20245 min read
In this article, we will be explaining the step-by-step process of integrating Apple Sign-in using React.
Questions we get asked first
What is an AI agent, in plain terms?
Software that takes a goal, decides the steps, uses your tools and systems to carry them out, and checks its own work. A chatbot answers; an agent acts. We build agents so that a person approves anything consequential.
Where should we start?
With one process that is repetitive, rule-heavy and costs real time. We map it, build the narrowest version that works on your data, and measure it. If it does not pay off, we will say so.
How do you keep agents safe?
Approval steps for consequential actions, least-privilege access to your systems, evaluations before release, monitoring after it, and a log of every decision so any run can be replayed and audited.
Which AI models do you use?
Whichever fits the task and your constraints: Claude, GPT, Gemini, or open-weight models you host yourselves. We design so the model can be swapped as better ones arrive.
Will our data be used to train models?
We choose providers and settings that do not train on your data, and we can run models inside your own cloud account when your policy requires it.
What does it cost?
Pilots are fixed in scope and price. Ongoing work is a monthly retainer or a dedicated team. We quote after a short discovery call.
Do you work with teams outside India?
Yes. We are based in India and work with teams across time zones, with overlap hours agreed up front.
Have a process that eats hours? Tell us about it.
A short call is enough to tell you whether an agent is the right answer, and what it would take.