Support triage
Classify tickets, pull order context, draft replies for an agent to approve.
LLM agents can read context, decide on a next step, and call tools: create a quote, look up stock, draft a reply. We build them with tight permissions, logs, and approval gates.
{
}
Classify tickets, pull order context, draft replies for an agent to approve.
Read an RFQ email, find products and prices, draft the quote in the ERP.
Gather facts across internal systems and summarize with citations.
Chase missing documents, reconcile small mismatches, update records.
| Action type | Example | Control |
|---|---|---|
| Read | Look up an order status | Allowed with scoped credentials |
| Draft | Prepare a quote or email | Saved as draft for review |
| Write | Update a delivery date | Allowed for low-risk fields, logged |
| Irreversible | Send payment, delete records | Human approval required |
An agent with access to everything is just an intern with root. Give it a badge, not the master key.
An LLM agent is a loop: the model reads a goal and some context, decides on a next step, calls a tool such as a search, an API, or an ERP action, looks at the result, and decides again until the job is done or it needs help. The model supplies judgment; the tools supply the ability to act.
That combination is powerful and risky in equal measure. A model that can call tools can also call the wrong one, misread an instruction hidden in an email, or loop on a task it cannot finish. Production agents need the same engineering discipline as any system with write access to business data: least privilege, validation, limits, and logs.
We build agents for specific, bounded jobs where the steps are clear and the value is measurable. A narrow agent that reliably prepares quotes is worth far more than a general assistant that sometimes does everything.
Step 01
Write down the task, the inputs, the expected output, and what 'done' means, as you would for a new hire.
Step 02
Expose the smallest set of tools that can do the job, each with typed arguments and validation.
Step 03
Give the agent its own scoped credentials, and decide which actions need human approval.
Step 04
Collect real examples with expected outcomes so every change is tested before release.
Step 05
Log every step, tool call, and result, with alerts on failures, loops, and cost spikes.
| Guardrail | What it prevents |
|---|---|
| Tool allow-list | Calling anything outside the job's scope |
| Argument validation | Malformed or out-of-range values reaching your systems |
| Approval gates | Irreversible actions without a person's sign-off |
| Step and cost limits | Runaway loops and surprise bills |
| Content isolation | Instructions hidden in emails or documents being treated as commands |
| Full trace | Unexplained outcomes; every decision is reviewable |
Agents are not always the answer. A fixed workflow is cheaper, faster, and easier to trust when it fits.
The most common pitfall is giving an agent too broad a goal. “Handle customer emails” fails in unpredictable ways; “classify the email, look up the order, and draft a reply for approval” succeeds because every step can be checked.
The second is skipping the evaluation set. Without real examples and expected outcomes, nobody can tell whether a prompt change made the agent better or worse, and regressions reach users first.
The third is treating cost and latency as afterthoughts. Agents that call large models many times per task can become slow and expensive at volume, so we measure both during the pilot and design around them.
| Measure | Why it matters |
|---|---|
| Task success rate | The share of tasks completed correctly without help |
| Escalation rate | How often the agent hands over to a person, and why |
| Cost per task | Model and tool costs, compared with manual effort |
| Time per task | End-to-end latency users actually experience |
Imagine a distributor whose sales team receives dozens of requests for quotation by email each day. A quoting agent reads each email, identifies the customer in the CRM, matches requested items to products, checks stock and price lists, and saves a draft quote in the ERP with any unmatched lines highlighted.
A salesperson opens the draft, fixes the highlighted lines, and sends it. The agent never sends anything itself, and every step it took is visible in the trace. Over time, the evaluation set grows from real corrections, and the share of drafts sent without changes becomes the measure of progress.
Anthropic distinguishes workflows, where models and tools follow predefined code paths, from agents that choose their own steps, and recommends finding the simplest solution possible, adding complexity only when needed, with stopping conditions such as a maximum number of iterationsSource [1].
On security, OWASP ranks prompt injection first in its 2025 Top 10 for LLM applications, including indirect injection through websites and files an agent readsSource [2]. Its guidance on excessive agency names excessive functionality, permissions, and autonomy as root causes, and recommends minimal permissions and human approval for high-impact actionsSource [3]. Our agents get narrow tools, scoped credentials, and approval steps for anything irreversible.
Production AI agents
LLM demos are easy; production AI is hard. Structured outputs, observability, human review, and adversarial testing for real agents.
AIAI security & governance
Deploy AI safely: private model hosting, data minimization, access controls, prompt-injection defenses, audit logs, and usage policies.
AIAI in your ERP
Embed AI in Odoo workflows: smart bank reconciliation, anomaly detection, document capture, and suggestions inside the screens users know.
Next step
Thirty minutes with an engineer, no slide deck required. You'll leave with a clearer picture, whether or not we work together.